TVS/Zener Theory and Design Considerations -...

http://onsemi.com1

TVS/Zener Theory and Design Considerations

Handbook

HBD854/DRev. 0, Jun−2005

© SCILLC, 2005Previous Edition © 2001 as Excerpted from DL150/D“All Rights Reserved’’

http://onsemi.com2

Technical Information, Application Notes and ArticlesZener Diode Theory 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zener Diode Fabrication Techniques 8. . . . . . . . . . . . . . . Reliability 12. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zener Diode Characteristics 18. . . . . . . . . . . . . . . . . . . . . Temperature Compensated Zeners 30. . . . . . . . . . . . . . . Basic Voltage Regulation Using Zener Diodes 34. . . . . Zener Protective Circuits and Techniques:

Basic Design Considerations 44. . . . . . . . . . . . . . . . . Zener Voltage Sensing Circuits and Applications 54. . . Miscellaneous Applications of

Zener Type Devices 61. . . . . . . . . . . . . . . . . . . . . . . . . Transient Voltage Suppression 63. . . . . . . . . . . . . . . . . . AN784 82. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . AN843 84. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Design Considerations and Performance of

Temperature Compensated Zener Diodes 97. . . . . . MOSORBs 102. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . AR450 106. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurement of Zener Voltage to Thermal

Equilibrium with Pulsed Test Current 119. . . . . . . . . .

http://onsemi.com3

ZENER DIODE THEORY

INTRODUCTION

The zener diode is a semiconductor device unique in itsmode of operation and completely unreplaceable by anyother electronic device. Because of its unusual properties itfills a long-standing need in electronic circuitry. It provides,among other useful functions, a constant voltage referenceor voltage control element available over a wide spectrum ofvoltage and power levels.

The zener diode is unique among the semiconductorfamily of devices because its electrical properties arederived from a rectifying junction which operates in thereverse breakdown region. In the sections that follow, thereverse biased rectifying junction, some of the termsassociated with it, and properties derived from it will bediscussed fully.

The zener diode is fabricated from the element silicon.Special techniques are applied in the fabrication of zenerdiodes to create the required properties.

This manual was prepared to acquaint the engineer, theequipment designer and manufacturer, and the experimenterwith the fundamental principles, design characteristics,applications and advantages of this importantsemiconductor device.

SEMICONDUCTOR THEORY

The active portion of a zener diode is a semiconductor PNjunction. PN junctions are formed in various kinds ofsemiconductor devices by several techniques. Among theseare the widely used techniques known as alloying anddiffusion which are utilized in fabricating zener PNjunctions to provide excellent control over zener breakdownvoltage.

At the present time, zener diodes use silicon as the basicmaterial in the formation of their PN junction. Silicon is inGroup IV of the periodic table (tetravalent) and is classed asa “semiconductor” due to the fact that it is a poor conductorin a pure state. When controlled amounts of certain“impurities” are added to a semiconductor it becomes abetter conductor of electricity. Depending on the type ofimpurity added to the basic semiconductor, its conductivitymay take two different forms, called P- and N-typerespectively.

N-type conductivity in a semiconductor is much like theconductivity due to the drift of free electrons in a metal. Inpure silicon at room temperature there are too few freeelectrons to conduct current. However, there are ways ofintroducing free electrons into the crystal lattice as we shall

now see. Silicon is a tetravalent element, one with fourvalence electrons in the outer shell; all are virtually lockedinto place by the covalent bonds of the crystal latticestructure, as shown schematically in Figure 1a. Whencontrolled amounts of donor impurities (Group V elements)such as phosphorus are added, the pentavalent phosphorusatoms entering the lattice structure provide extra electronsnot required by the covalent bonds. These impurities arecalled donor impurities since they “donate” a free electronto the lattice. These donated electrons are free to drift fromnegative to positive across the crystal when a field is applied,as shown in Figure 1b. The “N” nomenclature for this kindof conductivity implies “negative” charge carriers.

In P-type conductivity, the charges that carry electriccurrent across the crystal act as if they were positive charges.We know that electricity is always carried by driftingelectrons in any material, and that there are no mobilepositively charged carriers in a solid. Positive chargecarriers can exist in gases and liquids in the form of positiveions but not in solids. The positive character of the currentflow in the semiconductor crystal may be thought of as themovement of vacancies (called holes) in the covalent lattice.These holes drift from positive toward negative in an electricfield, behaving as if they were positive carriers.

P-type conductivity in semiconductors result from addingacceptor impurities (Group III elements) such as boron tosilicon to the semiconductor crystal. In this case, boronatoms, with three valence electrons, enter the tetravalentsilicon lattice. Since the covalent bonds cannot be satisfiedby only three electrons, each acceptor atom leaves a hole inthe lattice which is deficient by one electron. These holesreadily accept electrons introduced by external sources orcreated by radiation or heat, as shown in Figure 1c. Hencethe name acceptor ion or acceptor impurity. When anexternal circuit is connected, electrons from the currentsource “fill up” these holes from the negative end and jumpfrom hole to hole across the crystal or one may think of thisprocess in a slightly different but equivalent way, that is asthe displacement of positive holes toward the negativeterminal. It is this drift of the positively charged holes whichaccounts for the term P-type conductivity.

When semiconductor regions of N- and P-typeconductivities are formed in a semiconductor crystaladjacent to each other, this structure is called a PN junction.Such a junction is responsible for the action of both zenerdiodes and rectifier devices, and will be discussed in the nextsection.

http://onsemi.com4

Figure 1. Semiconductor Structure

−

Si

Si Si Si

Si

Si Si

Si

Si P Si

Si

Si Si

Si

Si B Si

Si

Si Si

APPLIED FIELD

APPLIED FIELD

+

ELECTRONS ARE LOCKED

IN COVALENT BONDS

LOCKED COVALENT

BOND ELECTRONS

FREE ELECTRON

FROM PHOSPHOROUS

ATOM DRIFTS TOWARD

APPLIED POSITIVE POLE.

INCOMPLETED

COVALENT BOND

THIS ELECTRON JUMPS INTO

HOLE LEFT BY BORON ATOM.

HOLE POSITION IS DISPLACED

TO RIGHT. THIS RESULTS IN

A DRIFT OF HOLES TOWARD

THE NEGATIVE POLE, GIVING

THEM THE CHARACTER OF

MOBILE POSITIVE CHARGES.

(a) Lattice Structures ofPure Silicon

(b) N-Type Silicon

(c) P-Type Silicon

−

THE SEMICONDUCTOR DIODE

In the forward-biased PN junction, Figure 2a, the P regionis made more positive than the N region by an externalcircuit. Under these conditions there is a very low resistanceto current flow in the circuit. This is because the holes in thepositive P-type material are very readily attracted across the

junction interface toward the negative N-type side.Conversely, electrons in the N-type are readily attracted bythe positive polarity in the other direction.

When a PN junction is reverse biased, the P-type side ismade more negative than the N-type side. (See Figure 2b.)At voltages below the breakdown of the junction, there isvery little current flow across the junction interface. At firstthought one would expect no reverse current under reversebias conditions, but several effects are responsible for thissmall current.

Under this condition the positive holes in the P-typesemiconductor are repelled from the junction interface bythe positive polarity applied to the N side, and conversely,the electrons in the N material are repelled from the interfaceby the negative polarity of the P side. This creates a regionextending from the junction interface into both P- andN-type materials which is completely free of charge carriers,that is, the region is depleted of its charge carriers. Hence,this region is usually called the depletion region.

Although the region is free of charge carriers, the P-sideof the depletion region will have an excess negative chargedue to the presence of acceptor ions which are, of course,fixed in the lattice; while the N-side of the depletion regionhas an excess positive charge due to the presence of donorions. These opposing regions of charged ions create a strongelectric field across the PN junction responsible for thecreation of reverse current.

The semiconductor regions are never perfect; there arealways a few free electrons in P material and few holes in Nmaterial. A more significant factor, however, is the fact thatgreat magnitudes of electron-hole pairs may be thermallygenerated at room temperatures in the semiconductor. Whenthese electron-hole pairs are created within the depletionregion, then the intense electric field mentioned in the aboveparagraph will cause a small current to flow. This smallcurrent is called the reverse saturation current, and tends tomaintain a relatively constant value for a fixed temperatureat all voltages. The reverse saturation current is usuallynegligible compared with the current flow when the junctionis forward biased. Hence, we see that the PN junction, whennot reverse biased beyond breakdown voltage, will conductheavily in only one direction. When this property is utilizedin a circuit we are employing the PN junction as a rectifier.Let us see how we can employ its reverse breakdowncharacteristics to an advantage.

As the reverse voltage is increased to a point called thevoltage breakdown point and beyond, current conductionacross the junction interface increases rapidly. The breakfrom a low value of the reverse saturation current to heavyconductance is very sharp and well defined in most PNjunctions. It is called the zener knee. When reverse voltagesgreater than the voltage breakdown point are applied to thePN junction, the voltage drop across the PN junctionremains essentially constant at the value of the breakdownvoltage for a relatively wide range of currents. This region

http://onsemi.com5

N

APPLIED FIELD

P

LARGE

CURRENT

SEVERAL VOLTS

APPLIED FIELD

N P

VERY

SMALL

CURRENT

Figure 2. Effects of Junction Bias

CHARGES FROM BOTH P AND N REGIONS

DRIFT ACROSS JUNCTION AT VERY LOW

APPLIED VOLTAGES.

AT APPLIED VOLTAGES BELOW THE CRITICAL

BREAKDOWN LEVEL ONLY A FEW CHARGES

DRIFT ACROSS THE INTERFACE.

(a) Forward-Based PNJunction

(b) Reverse-Biased PNJunction

beyond the voltage breakdown point is called the zenercontrol region.

ZENER CONTROL REGION: VOLTAGEBREAKDOWN MECHANISMS

Figure 3 depicts the extension of reverse biasing to thepoint where voltage breakdown occurs. Although all PNjunctions exhibit a voltage breakdown, it is important toknow that there are two distinct voltage breakdownmechanisms. One is called zener breakdown and the other iscalled avalanche breakdown. In zener breakdown the valueof breakdown voltage decreases as the PN junctiontemperature increases; while in avalanche breakdown thevalue of the breakdown voltage increases as the PN junctiontemperature increases. Typical diode breakdowncharacteristics of each category are shown in Figure 4. Thefactor determining which of the two breakdownmechanisms occurs is the relative concentrations of theimpurities in the materials which comprise the junction. Iftwo different resistivity P-type materials are placed againsttwo separate but equally doped low-resistivity pieces ofN-type materials, the depletion region spread in the lowresistivity P-type material will be smaller than the depletionregion spread in the high resistivity P-type material.Moreover, in both situations little of the resultant depletionwidth lies in the N material if its resistivity is low comparedto the P-type material. In other words, the depletion regionalways spreads principally into the material having thehighest resistivity. Also, the electric field (voltage per unit

Figure 3. Reverse Characteristic Extendedto Show Breakdown Effect

SLOPEIREV

VBREAKDOWNVREV

length) in the less resistive material is greater than theelectric field in the material of greater resistivity due to thepresence of more ions/unit volume in the less resistivematerial. A junction that results in a narrow depletion regionwill therefore develop a high field intensity and breakdownby the zener mechanism. A junction that results in a widerdepletion region and, thus, a lower field intensity will breakdown by the avalanche mechanism before a zenerbreakdown condition can be reached.

The zener mechanism can be described qualitatively asfollows: because the depletion width is very small, theapplication of low reverse bias (5 volts or less) will cause afield across the depletion region on the order of 3 x 105V/cm.A field of such high magnitude exerts a large force on the

http://onsemi.com6

Figure 4. Typical Breakdown Diode Characteristics. Note Effects of Temperature for Each Mechanism

(A)

ZENER BREAKDOWN

OF A PN FUNCTION

VREV (VOLTS)

(B)

AVALANCHE BREAKDOWN OF A PN FUNCTION

IREV IREV

4 3 2 1

VREV (VOLTS)

30 25 20 15 10 5

25°C 65°C 25°C65°C

valence electrons of a silicon atom, tending to separate themfrom their respective nuclei. Actual rupture of the covalentbonds occurs when the field approaches 3 x 105V/cm. Thus,electron-hole pairs are generated in large numbers and asudden increase of current is observed. Although we speakof a rupture of the atomic structure, it should be understoodthat this generation of electron-hole pairs may be carried oncontinuously as long as an external source suppliesadditional electrons. If a limiting resistance in the circuitexternal to the diode junction does not prevent the currentfrom increasing to high values, the device may be destroyeddue to overheating. The actual critical value of field causingzener breakdown is believed to be approximately3 x 105V/cm. On most commercially available silicondiodes, the maximum value of voltage breakdown by thezener mechanism is 8 volts. In order to fabricate deviceswith higher voltage breakdown characteristics, materialswith higher resistivity, and consequently, wider depletionregions are required. These wide depletion regions hold thefield strength down below the zener breakdown value(3 x 105V/cm). Consequently, for devices with breakdownvoltage lower than 5 volts the zener mechanismpredominates, between 5 and 8 volts both zener and anavalanche mechanism are involved, while above 8 volts theavalanche mechanism alone takes over.

The decrease of zener breakdown voltage as junctiontemperature increases can be explained in terms of theenergies of the valence electrons. An increase of temperatureincreases the energies of the valence electrons. This weakensthe bonds holding the electrons and consequently, less appliedvoltage is necessary to pull the valence electrons from theirposition around the nuclei. Thus, the breakdown voltagedecreases as the temperature increases.

The dependence on temperature of the avalanchebreakdown mechanism is quite different. Here the depletionregion is of sufficient width that the carriers (electrons orholes) can suffer collisions before traveling the regioncompletely i.e., the depletion region is wider than onemean-free path (the average distance a carrier can travel

before combining with a carrier of opposite conductivity).Therefore, when temperature is increased, the increasedlattice vibration shortens the distance a carrier travels beforecolliding and thus requires a higher voltage to get it acrossthe depletion region.

As established earlier, the applied reverse bias causes asmall movement of intrinsic electrons from the P material tothe potentially positive N material and intrinsic holes fromthe N material to the potentially negative P material (leakagecurrent). As the applied voltage becomes larger, theseelectrons and holes increasingly accelerate. There are alsocollisions between these intrinsic particles and boundelectrons as the intrinsic particles move through thedepletion region. If the applied voltage is such that theintrinsic electrons do not have high velocity, then thecollisions take some energy from the intrinsic particles,altering their velocity. If the applied voltage is increased,collision with a valence electron will give considerableenergy to the electron and it will break free of its covalentbond. Thus, one electron by collision, has created anelectron-hole pair. These secondary particles will also beaccelerated and participate in collisions which generate newelectron-hole pairs. This phenomenon is called carriermultiplication. Electron-hole pairs are generated so quicklyand in such large numbers that there is an apparent avalancheor self-sustained multiplication process (depictedgraphically in Figure 5). The junction is said to be inbreakdown and the current is limited only by resistanceexternal to the junction. Zener diodes above 7 to 8 voltsexhibit avalanche breakdown.

As junction temperature increases, the voltage breakdownpoint for the avalanche mechanism increases. This effect canbe explained by considering the vibration displacement ofatoms in their lattice increases, and this increaseddisplacement corresponds to an increase in the probabilitythat intrinsic particles in the depletion region will collidewith the lattice atoms. If the probability of an intrinsicparticle-atom collision increases, then the probability that agiven intrinsic particle will obtain high momentum

http://onsemi.com7

Figure 5. PN Junction in Avalanche Breakdown

WHEN THE APPLIED VOLTAGE IS

ABOVE THE BREAKDOWN POINT, A

FEW INJECTED ELECTRONS RECEIVE

ENOUGH ACCELERATION FROM THE

FIELD TO GENERATE NEW ELECTRONS

BY COLLISION. DURING THIS PROCESS

THE VOLTAGE DROP ACROSS THE

JUNCTION REMAINS CONSTANT.

RS ABSORBS EXCESS VOLTAGE.

R

S

N P

LARGE CURRENT

CONSTANT VOLTAGE DROP

REVERSE-BIASED

PN JUNCTION IN AVALANCHE

decreases, and it follows that the low momentum intrinsicparticles are less likely to ionize the lattice atoms. Naturally,increased voltage increases the acceleration of the intrinsicparticles, providing higher mean momentum and moreelectron-hole pairs production. If the voltage is raisedsufficiently, the mean momentum becomes great enough tocreate electron-hole pairs and carrier multiplication results.Hence, for increasing temperature, the value of theavalanche breakdown voltage increases.

VOLT-AMPERE CHARACTERISTICS

The zener volt-ampere characteristics for a typical 30 voltzener diode is illustrated in Figure 6. It shows that the zenerdiode conducts current in both directions; the forwardcurrent IF being a function of forward voltage VF. Note thatIF is small until VF ≈ 0.65 V; then IF increases very rapidly.For VF > 0.65 V IF is limited primarily by the circuitresistance external to the diode.

ZZK

VZ

ZZT

IR

30 20 10 0 0.5 1 1.5

15

10

5

0

0.5

1

1.5

REVERSE

CHARACTERISTIC

VR

(VOLTS)

VF

(VOLTS)

I (

AM

PS

)F

RE

VE

RS

E C

UR

RE

NT

(AM

PS

)

Figure 6. Zener Diode Characteristics

IZT

IZM

1.40 A

IZK = 5 mA

FORWARD

CHARACTERISTIC TYPICAL

420 mA

The reverse current is a function of the reverse voltage VRbut for most prNO TAGactical purposes is zero until thereverse voltage approaches VZ, the PN junction breakdownvoltage, at which time the reverse current increases veryrapidly. Since the reverse current is small for VR < VZ, butgreat for VR > VZ each of the current regions is specified by

a different symbol. For the leakage current region, i.e.non-conducting region, between 0 volts and VZ, the reversecurrent is denoted by the symbol IR; but for the zener controlregion, VR ≥ VZ, the reverse current is denoted by thesymbol IZ. IR is usually specified at a reverse voltageVR ≈ 0.8 VZ.

The PN junction breakdown voltage, VZ, is usually calledthe zener voltage, regardless whether the diode is of thezener or avalanche breakdown type. Commercial zenerdiodes are available with zener voltages from about1.8 V − 400 V. For most applications the zener diode isoperated well into the breakdown region (IZT to IZM). Mostmanufacturers give an additional specification of IZK(= 5 mA in Figure 6) to indicate a minimum operatingcurrent to assure reasonable regulation.

This minimum current IZK varies in the various types ofzener diodes and, consequently, is given on the data sheets.The maximum zener current IZM should be considered themaximum reverse current recommended by themanufacturer. Values of IZM are usually given in the datasheets.

Between the limits of IZK and IZM, which are 5 mA and1400 mA (1.4 Amps) in the example of Figure 6, the voltageacross the diode is essentially constant, and ≈ VZ. Thisplateau region has, however, a large positive slope such thatthe precise value of reverse voltage will change slightly asa function of IZ. For any point on this plateau region one maycalculate an impedance using the incremental magnitudes ofthe voltage and current. This impedance is usually called thezener impedance ZZ, and is specified for most zener diodes.Most manufacturers measure the maximum zenerimpedance at two test points on the plateau region. The firstis usually near the knee of the zener plateau, ZZK, and thelatter point near the midrange of the usable zener currentexcursion. Two such points are illustrated in Figure 6.

This section was intended to introduce the reader to a fewof the major terms used with zener diodes. A completedescription of these terms may be found in chapter four. Inchapter four a full discussion of zener leakage, DCbreakdown, zener impedance, temperature coefficients andmany other topics may be found.

http://onsemi.com8

ZENER DIODE FABRICATION TECHNIQUES

INTRODUCTION

A brief exposure to the techniques used in the fabricationof zener diodes can provide the engineer with additionalinsight using zeners in their applications. That is, anunderstanding of zener fabrication makes the capabilitiesand limitations of the zener diode more meaningful. Thischapter discusses the basic steps in the fabrication of thezener from crystal growing through final testing.

ZENER DIODE WAFER FABRICATION

The major steps in the manufacture of zeners are providedin the process flow in Figure 1. It is important to point outthat the manufacturing steps vary somewhat frommanufacturer to manufacturer, and also vary with the type ofzener diode produced. This is driven by the type of packagerequired as well as the electrical characteristics desired. Forexample, alloy diffused devices provide excellent lowvoltage reference with low leakage characteristics but do not

have the same surge carrying capability as diffused diodes.The manufacturing process begins with the growing of highquality silicon crystals.

Crystals for ON Semiconductor zener diodes are grownusing the Czochralski technique, a widely used processwhich begins with ultra-pure polycrystalline silicon. Thepolycrystalline silicon is first melted in a nonreactivecrucible held at a temperature just above the melting point.A carefully controlled quantity of the desired dopantimpurity, such as phosphorus or boron is added. A highquality seed crystal of the desired crystalline orientation isthen lowered into the melt while rotating. A portion of thisseed crystal is allowed to melt into the molten silicon. Theseed is then slowly pulled and continues to rotate as it israised from the melt. As the seed is raised, cooling takesplace and material from the melt adheres to it, thus forminga single crystal ingot. With this technique, ingots withdiameters of several inches can be fabricated.

Figure 1. General Flow of the Zener Diode Process

SILICON CRYSTAL

GROWING

WAFER

PREPARATION

OXIDE

PASSIVATION

WAFER THINNINGANODE

METALLIZATION

JUNCTION

FORMATION

CATHODE

METALLIZATION

WAFER

TESTINGWAFER DICING

TESTLEAD

FINISHASSEMBLY

MARK TEST PACKAGE

SHIP

http://onsemi.com9

Figure 2. Basic Fabrication Steps in the Silicon Planar Process: a) oxide formation, b) selective oxide removal,c) deposition of dopant atoms, d) junction formation by diffusion of dopant atoms.

SILICON DIOXIDE GROWTH

SiO2

Si

(a)

SILICON DIOXIDE SELECTIVELY REMOVED

(b)

(c)

DOPANT ATOMS DEPOSITED ONTO

THE EXPOSED SILICON

(d)

DOPANT ATOMS DIFFUSE INTO SILICON

BUT NOT APPRECIABLY INTO THE SILICON DIOXIDE

Once the single-crystal silicon ingot is grown, it is testedfor doping concentration (resistivity), undesired impuritylevels, and minority carrier lifetime. The ingot is then slicedinto thin, circular wafers. The wafers are then chemicallyetched to remove saw damage and polished in a sequence ofsuccessively finer polishing grits until a mirror-like defectfree surface is obtained. The wafers are then cleaned andplaced in vacuum sealed wafer carriers to prevent anycontamination from getting on them. At this point, thewafers are ready to begin device fabrication.

Zener diodes can be manufactured using differentprocessing techniques such as planar processing or mesaetched processing. The majority of ON Semiconductorzener diodes are manufactured using the planar technique asshown in Figure 2.

The planar process begins by growing an ultra-cleanprotective silicon dioxide passivation layer. The oxide istypically grown in the temperature range of 900 to 1200degrees celcius. Once the protective layer of silicon dioxidehas been formed, it must be selectively removed from thoseareas into which dopant atoms will be introduced. This isdone using photolithographic techniques.

First a light sensitive solution called photo resist is spunonto the wafer. The resist is then dried and a photographicnegative or mask is placed over the wafer. The resist is thenexposed to ultraviolet light causing the molecules in it tocross link or polymerize becoming very rigid. Those areasof the wafer that are protected by opaque portions of themask are not exposed and are developed away. The oxide isthen etched forming the exposed regions in which the dopantwill be introduced. The remaining resist is then removed andthe wafers carefully cleaned for the doping steps.

Dopant is then introduced onto the wafer surface usingvarious techniques such as aluminum alloy for low voltagedevices, ion-implantation, spin-on dopants, or chemicalvapor deposition. Once the dopant is deposited, thejunctions are formed in a subsequent high temperature (1100to 1250 degrees celcius are typical) drive-in. The resultantjunction profile is determined by the backgroundconcentration of the starting substrate, the amount of dopantplaced at the surface, and amount of time and temperatureused during the dopant drive-in. This junction profiledetermines the electrical characteristics of the device.During the drive-in cycle, additional passivation oxide isgrown providing additional protection for the devices.

After junction formation, the wafers are then processedthrough what is called a getter process. The getter steputilizes high temperature and slight stress provided by ahighly doped phosphosilicate glass layer introduced into thebackside of the wafers. This causes any contaminants in thearea of the junction to diffuse away from the region. Thisserves to improve the reverse leakage characteristic and thestability of the device. Following the getter process, a secondphoto resist step opens the contact area in which the anodemetallization is deposited.

Metal systems for ON Semiconductor’s zener diodes aredetermined by the requirements of the package. The metalsystems are deposited in ultra-clean vacuum chambersutilizing electron-beam evaporation techniques. Once themetal is deposited, photo resist processing is utilized to formthe desired patterns. The wafers are then lapped to their finalthickness and the cathode metallization deposited using thesame e-beam process.

http://onsemi.com10

The quality of the wafers is closely monitored throughoutthe process by using statistical process control techniquesand careful microscopic inspections at critical steps. Specialwafer handling equipment is used throughout themanufacturing process to minimize contamination and toavoid damaging the wafers in any way. This furtherenhances the quality and stability of the devices.

Upon completion of the fabrication steps, the wafers areelectrically probed, inspected, and packaged for shipment tothe assembly operations. All ON Semiconductor zenerdiode product is sawn using 100% saw-through techniquesstringently developed to provide high quality silicon die.

ZENER DIODE ASSEMBLY

Surmetic 30, 40 and MOSORB

The plastic packages (Surmetic 30, 40 and MOSORBs)are assembled using oxygen free high conductivity copperleads for efficient heat transfer from the die and allowingmaximum power dissipation with a minimum of externalheatsinking. Figure 3 shows typical assembly. The leads areof nail head construction, soldered directly to the die, whichfurther enhances the heat dissipating capabilities of thepackage.

The Surmetic 30s, 40s and MOSORBs are basicallyassembled in the same manner; the only difference being theMOSORBs are soldered together using a solder discbetween the lead and die whereas the Surmetic 30s andSurmetic 40s utilize pre-soldered leads.

Assembly is started on the Surmetic 30 and 40 by loadingthe leads into assembly boats and pre-soldering the nailheads. After pre-soldering, one die is then placed into eachcavity of one assembly boat and another assembly boat isthen mated to it. Since the MOSORBs do not usepre-soldered leads, the leads are put into the assembly boat,

a solder disc is placed into each cavity and then a die is putin on top. A solder disc is put in on top of the die. Anotherassembly boat containing only leads is mated to the boatcontaining the leads, die, and two solder discs. The boats arepassed through the assembly furnace; this operation requiresonly one pass through the furnace.

After assembly, the leads on the Surmetic 30s, 40s andMOSORBs are plated with a tin-lead alloy making themreadily solderable and corrosion resistant.

Double Slug (DO-35 and DO-41)

Double slugs receive their name from the dumet slugs, oneattached to one end of each lead. These slugs sandwich thepre-tinned die between them and are hermetically sealed tothe glass envelope or body during assembly. Figure 4 showstypical assembly.

The assembly begins with the copper clad steel leadsbeing loaded into assembly “boats.” Every other boat loadof leads has a glass body set over the slug. A pre-tinned dieis placed into each glass body and the other boat load of leadsis mated to the boat holding the leads, body and die. Thesemated boats are then placed into the assembly furnace wherethe total mass is heated. Each glass body melts; and as theboat proceeds through the cooling portion of the furnacechamber, the tin which has wetted to each slug solidifiesforming a bond between the die and both slugs. The glasshardens, attaching itself to the sides of the two slugs formingthe hermetic seal. The above illustrates how the diodes arecompletely assembled using a single furnace passminimizing assembly problems.

The encapsulated devices are then processed through leadfinish. This consists of dipping the leads in molten tin/leadsolder alloy. The solder dipped leads produce an externalfinish which is tarnish-resistant and very solderable.

Figure 3. Double-Slug PlasticZener Construction

Figure 4. Double Slug GlassZener Construction

OFHC COPPER LEAD,

SOLDER PLATED

PLASTIC

(THERMO SET)

ENCAPSULATED

NAILHEAD LEAD

ZENER DIE Sn Pb

OFHC COPPER LEAD,

SOLDER PLATED

LEAD, STEEL, CU CLAD

SOLDER DIPPED

SLUG DUMET

GLASS SLEEVE

PASSIVATED

ZENER DIENAILHEAD LEAD

http://onsemi.com11

ZENER DIODE TEST, MARK AND PACKAGING

Double Slug, Surmetic 30, 40 and MOSORB

After lead finish, all products are final tested, whetherthey are double slug or of Surmetic construction, all are 100percent final tested for zener voltage, leakage current,impedance and forward voltage drop.

Process average testing is used which is based upon theaverages of the previous lots for a given voltage line andpackage type. Histograms are generated for the various

parameters as the units are being tested to ensure that the lotis testing well to the process average and compared againstother lots of the same voltage.

After testing, the units are marked as required by thespecification. The markers are equipped to polarity orientthe devices as well as perform 100% redundant test prior topackaging.

After marking, the units are packaged either in “bulk”form or taped and reeled or taped and ammo packed toaccommodate automatic insertion.

http://onsemi.com12

RELIABILITY

INTRODUCTION

ON Semiconductor’s Quality System maintains“continuous product improvement” goals in all phases of theoperation. Statistical process control (SPC), quality controlsampling, reliability audits and accelerated stress testingtechniques monitor the quality and reliability of its products.Management and engineering skills are continuouslyupgraded through training programs. This maintains aunified focus on Six Sigma quality and reliability from theinception of the product to final customer use.

STATISTICAL PROCESS CONTROL

ON Semiconductor’s Discrete Group is continuallypursuing new ways to improve product quality. Initial designimprovement is one method that can be used to produce asuperior product. Equally important to outgoing productquality is the ability to produce product that consistentlyconforms to specification. Process variability is the basicenemy of semiconductor manufacturing since it leads toproduct variability. Used in all phases ofON Semiconductor’s product manufacturing,STATISTICAL PROCESS CONTROL (SPC) replacesvariability with predictability. The traditional philosophy inthe semiconductor industry has been adherence to the datasheet specification. Using SPC methods assures the productwill meet specific process requirements throughout themanufacturing cycle. The emphasis is on defect prevention,not detection. Predictability through SPC methods requiresthe manufacturing culture to focus on constant andpermanent improvements. Usually these improvementscannot be bought with state-of-the-art equipment orautomated factories. With quality in design, process andmaterial selection, coupled with manufacturingpredictability, ON Semiconductor can produce world classproducts.

The immediate effect of SPC manufacturing ispredictability through process controls. Product centered anddistributed well within the product specification benefits ONSemiconductor with fewer rejects, improved yields and lowercost. The direct benefit to ON Semiconductor’s customersincludes better incoming quality levels, less inspection timeand ship-to-stock capability. Circuit performance is oftendependent on the cumulative effect of component variability.Tightly controlled component distributions give the customergreater circuit predictability. Many customers are alsoconverting to just-in-time (JIT) delivery programs. Theseprograms require improvements in cycle time and yieldpredictability achievable only through SPC techniques. Thebenefit derived from SPC helps the manufacturer meet thecustomer’s expectations of higher quality and lower costproduct.

Ultimately, ON Semiconductor will have Six Sigmacapability on all products. This means parametric

distributions will be centered within the specification limitswith a product distribution of plus or minus Six Sigma aboutmean. Six Sigma capability, shown graphically in Figure 1,details the benefit in terms of yield and outgoing qualitylevels. This compares a centered distribution versus a 1.5sigma worst case distribution shift.

New product development at ON Semiconductor requiresmore robust design features that make them less sensitive tominor variations in processing. These features make theimplementation of SPC much easier.

A complete commitment to SPC is present throughoutON Semiconductor. All managers, engineers, productionoperators, supervisors and maintenance personnel havereceived multiple training courses on SPC techniques.Manufacturing has identified 22 wafer processing and 8assembly steps considered critical to the processing of zenerproducts. Processes, controlled by SPC methods, that haveshown significant improvement are in the diffusion,photolithography and metallization areas.

To better understand SPC principles, brief explanationshave been provided. These cover process capability,implementation and use.

Figure 1. AOQL and Yield from a Normal Distribution ofProduct With 6σ Capability

Standard Deviations From Mean

Distribution Centered Distribution Shifted ± 1.5At ± 3 σ 2700 ppm defective

99.73% yield

At ± 4 σ 63 ppm defective99.9937% yield

At ± 5 σ 0.57 ppm defective99.999943% yield

At ± 6 σ 0.002 ppm defective99.9999998% yield

66810 ppm defective93.32% yield



3.4 ppm defective99.99966% yield

�-6σ �-5s �-4σ �-3σ �-2σ �-1σ �0 �1σ �2σ �3σ �4σ �5σ �6σ

PROCESS CAPABILITY

One goal of SPC is to ensure a process is CAPABLE.Process capability is the measurement of a process toproduce products consistently to specificationrequirements. The purpose of a process capability study isto separate the inherent RANDOM VARIABILITY fromASSIGNABLE CAUSES. Once completed, steps are takento identify and eliminate the most significant assignablecauses. Random variability is generally present in thesystem and does not fluctuate. Sometimes, these areconsidered basic limitations associated with the machinery,materials, personnel skills or manufacturing methods.Assignable cause inconsistencies relate to time variations inyield, performance or reliability.

http://onsemi.com13

Figure 2. Impact of Assignable Causes on Process Predictable

Figure 3. Difference Between Process Control and Process Capability

??

?

??

?

???

Process “under control” − all assignable causes areremoved and future distribution is predictable.

PREDICTION

TIME

SIZESIZE

TIME

PREDICTION

SIZE

TIME

Out of control(assignable causes present)

In control assignable causes eliminated

SIZE

TIME

In control but not capable(variation from random variability

excessive)

LowerSpecification Limit

UpperSpecification Limit

In control and capable(variation from random

variability reduced)

? ?

Traditionally, assignable causes appear to be random dueto the lack of close examination or analysis. Figure 2 showsthe impact on predictability that assignable cause can have.Figure 3 shows the difference between process control andprocess capability.

A process capability study involves taking periodicsamples from the process under controlled conditions. Theperformance characteristics of these samples are chartedagainst time. In time, assignable causes can be identified andengineered out. Careful documentation of the process is keyto accurate diagnosis and successful removal of theassignable causes. Sometimes, the assignable causes willremain unclear requiring prolonged experimentation.

Elements which measure process variation control andcapability are Cp and Cpk respectively. Cp is thespecification width divided by the process width or Cp =(specification width) / 6σ. Cpk is the absolute value of theclosest specification value to the mean, minus the mean,divided by half the process width or Cpk = | closestspecification — X / 3σ.

At ON Semiconductor, for critical parameters, the processcapability is acceptable with a Cpk = 1.33. The desiredprocess capability is a Cpk = 2 and the ideal is a Cpk = 5.Cpk, by definition, shows where the current productionprocess fits with relationship to the specification limits. Offcenter distributions or excessive process variability willresult in less than optimum conditions.

SPC IMPLEMENTATION AND USE

The Discrete Group uses many parameters that showconformance to specification. Some parameters aresensitive to process variations while others remain constantfor a given product line. Often, specific parameters areinfluenced when changes to other parameters occur. It isboth impractical and unnecessary to monitor all parametersusing SPC methods. Only critical parameters that aresensitive to process variability are chosen for SPCmonitoring. The process steps affecting these criticalparameters must be identified also. It is equally important tofind a measurement in these process steps that correlateswith product performance. This is called a critical processparameter.

Once the critical process parameters are selected, a sampleplan must be determined. The samples used formeasurement are organized into RATIONALSUBGROUPS of approximately 2 to 5 pieces. Thesubgroup size should be such that variation among thesamples within the subgroup remain small. All samples mustcome from the same source e.g., the same mold pressoperator, etc.. Subgroup data should be collected atappropriate time intervals to detect variations in the process.As the process begins to show improved stability, theinterval may be increased. The data collected must becarefully documented and maintained for later correlation.Examples of common documentation entries would includeoperator, machine, time, settings, product type, etc..

http://onsemi.com14

Figure 4. Example of Process Control Chart Showing Oven Temperature Data

147

148

149

150

151

152

153

154

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75

0

1

2

3

4

5

6

7

UCL = 152.8

= 150.4

LCL = 148.0

UCL = 7.3

= 3.2

LCL = 0

X

R

Once the plan is established, data collection may begin.The data collected will generate X and R values that areplotted with respect to time. X refers to the mean of thevalues within a given subgroup, while R is the range orgreatest value minus least value. When approximately 20 ormore X and R values have been generated, the average ofthese values is computed as follows:

X = (X + X2 + X3 + ...)/KR = (R1 + R2 + R3 + ...)/K

where K = the number of subgroups measured.The values of X and R are used to create the process

control chart. Control charts are the primary SPC tool usedto signal a problem. Shown in Figure 4, process controlcharts show X and R values with respect to time andconcerning reference to upper and lower control limitvalues. Control limits are computed as follows:

R upper control limit = UCLR = D4 RR lower control limit LCLR = D3 RX upper control limit = UCLX = X + A2 RX lower control limit = LCLX = X − A

Where D4, D3 and A2 are constants varying by samplesize, with values for sample sizes from 2 to 10 shown inthe following partial table:

Control charts are used to monitor the variability ofcritical process parameters. The R chart shows basic

problems with piece to piece variability related to theprocess. The X chart can often identify changes in people,machines, methods, etc. The source of the variability can bedifficult to find and may require experimental designtechniques to identify assignable causes.

Some general rules have been established to helpdetermine when a process is OUT-OF-CONTROL. Figure5a shows a control chart subdivided into zones A, B, and Ccorresponding to 3 sigma, 2 sigma, and 1 sigma limitsrespectively. In Figure 5b through Figure 5e four of the teststhat can be used to identify excessive variability and thepresence of assignable causes are shown. As familiarity witha given process increases, more subtle tests may beemployed successfully.

Once the variability is identified, the cause of thevariability must be determined. Normally, only a few factorshave a significant impact on the total variability of theprocess. The importance of correctly identifying thesefactors is stressed in the following example. Suppose aprocess variability depends on the variance of five factors A,B, C, D and E. Each has a variance of 5, 3, 2, 1 and 0.4respectively.Since:

σ tot = σA2 + σB2 + σC2 + σD2 + σE2

σ tot = 52 + 32 + 22 + 12 + (0.4)2 = 6.3

n 2 3 4 5 6 7 8 9 10

D4 3.27 2.57 2.28 2.11 2.00 1.92 1.86 1.82 1.78

D3 * * * * * 0.08 0.14 0.18 0.22

A2 1.88 1.02 0.73 0.58 0.48 0.42 0.37 0.34 0.31

* For sample sizes below 7, the LCLR would technically be a negative number; in those cases there is no lower control limit;this means that for a subgroup size 6, six “identical” measurements would not be unreasonable.

http://onsemi.com15

Now if only D is identified and eliminated then;

s tot = 52 + 32 + 22 + (0.4)2 = 6.2This results in less than 2% total variability improvement.

If B, C and D were eliminated, then;

σ tot = 52 + (0.4)2 = 5.02This gives a considerably better improvement of 23%. If

only A is identified and reduced from 5 to 2, then;

σ tot = 22 + 32 + 22 + 12 + (0.4)2 = 4.3Identifying and improving the variability from 5 to 2 gives

us a total variability improvement of nearly 40%.Most techniques may be employed to identify the primary

assignable cause(s). Out-of-control conditions may becorrelated to documented process changes. The product maybe analyzed in detail using best versus worst partcomparisons or Product Analysis Lab equipment.Multi-variance analysis can be used to determine the familyof variation (positional, critical or temporal). Lastly,experiments may be run to test theoretical or factorialanalysis. Whatever method is used, assignable causes must

be identified and eliminated in the most expeditious mannerpossible.

After assignable causes have been eliminated, newcontrol limits are calculated to provide a more challengingvariability criteria for the process. As yields and variabilityimprove, it may become more difficult to detectimprovements because they become much smaller. When allassignable causes have been eliminated and the pointsremain within control limits for 25 groups, the process issaid to be in a state of control.

SUMMARYON Semiconductor is committed to the use of

STATISTICAL PROCESS CONTROLS. These principles,used throughout manufacturing, have already resulted inmany significant improvements to the processes. Continueddedication to the SPC culture will allow ON Semiconductorto reach the Six Sigma and zero defect capability goals. SPCwill further enhance the commitment to TOTALCUSTOMER SATISFACTION.

UCL

LCL

UCL

UCLUCL

UCL

LCL

LCLLCL

LCL

CENTERLINE

A

B

C

C

B

A

A

B

C

C

B

A

A

B

C

C

B

A

A

B

C

C

B

A

ZONE A (+ 3 SIGMA)

ZONE B (+ 2 SIGMA)

ZONE C (+ 1 SIGMA)

ZONE C (− 1 SIGMA)

ZONE B (− 2 SIGMA)

ZONE A (− 3 SIGMA)

Figure 5a. Control Chart Zones Figure 5b. One Point Outside Control LimitIndicating Excessive Variability

Figure 5c. Two Out of Three Points in Zone A orBeyond Indicating Excessive Variability

Figure 5d. Four Out of Five Points in Zone B orBeyond Indicating Excessive Variability

Figure 5e. Seven Out of Eight Points in Zone C orBeyond Indicating Excessive Variability

http://onsemi.com16

RELIABILITY STRESS TESTS

The following gives brief descriptions of the reliabilitytests commonly used in the reliability monitoring program.Not all of the tests listed are performed on each product.Other tests may be performed when appropriate. In additionsome form of preconditioning may be used in conjunctionwith the following tests.

AUTOCLAVE (aka, PRESSURE COOKER)

Autoclave is an environmental test which measuresdevice resistance to moisture penetration and the resultanteffects of galvanic corrosion. Autoclave is a highlyaccelerated and destructive test.

Typical Test Conditions: TA = 121°C, rh = 100%, p =1 atmosphere (15 psig), t = 24 to 96 hoursCommon Failure Modes: Parametric shifts, highleakage and/or catastrophicCommon Failure Mechanisms: Die corrosion orcontaminants such as foreign material on or within thepackage materials. Poor package sealing

HIGH HUMIDITY HIGH TEMPERATURE BIAS(H3TB or H3TRB)

This is an environmental test designed to measure themoisture resistance of plastic encapsulated devices. A biasis applied to create an electrolytic cell necessary toaccelerate corrosion of the die metallization. With time, thisis a catastrophically destructive test.

Typical Test Conditions: TA = 85°C to 95°C, rh = 85%to 95%, Bias = 80% to 100% of Data Book max. rating,t = 96 to 1750 hoursCommon Failure Modes: Parametric shifts, highleakage and/or catastrophicCommon Failure Mechanisms: Die corrosion orcontaminants such as foreign material on or within thepackage materials. Poor package sealingMilitary Reference: MIL-STD-750, Method 1042

HIGH TEMPERATURE REVERSE BIAS (HTRB)

The purpose of this test is to align mobile ions by meansof temperature and voltage stress to form a high-currentleakage path between two or more junctions.

Typical Test Conditions: TA = 85°C to 150°C, Bias =80% to 100% of Data Book max. rating, t = 120 to 1000hoursCommon Failure Modes: Parametric shifts in leakageCommon Failure Mechanisms: Ionic contamination onthe surface or under the metallization of the dieMilitary Reference: MIL-STD-750, Method 1039

HIGH TEMPERATURE STORAGE LIFE (HTSL)

High temperature storage life testing is performed toaccelerate failure mechanisms which are thermallyactivated through the application of extreme temperatures.

Typical Test Conditions: TA = 70°C to 200°C, no bias,t = 24 to 2500 hoursCommon Failure Modes: Parametric shifts in leakageCommon Failure Mechanisms: Bulk die and diffusiondefectsMilitary Reference: MIL-STD-750, Method 1032

INTERMITTENT OPERATING LIFE (IOL)

The purpose of this test is the same as SSOL in additionto checking the integrity of both wire and die bonds bymeans of thermal stressing.

Typical Test Conditions: TA = 25°C, Pd = Data Bookmaximum rating, Ton = Toff = Δ of 50°C to 100°C, t =42 to 30000 cyclesCommon Failure Modes: Parametric shifts andcatastrophicCommon Failure Mechanisms: Foreign material, crackand bulk die defects, metallization, wire and die bonddefectsMilitary Reference: MIL-STD-750, Method 1037

MECHANICAL SHOCK

This test is used to determine the ability of the device towithstand a sudden change in mechanical stress due toabrupt changes in motion as seen in handling, transportation,or actual use.

Typical Test Conditions: Acceleration = 1500 g’s,Orientation = X1, Y1, Y2 plane, t = 0.5 msec, Blows = 5Common Failure Modes: Open, short, excessiveleakage, mechanical failureCommon Failure Mechanisms: Die and wire bonds,cracked die, package defectsMilitary Reference: MIL-STD-750, Method 2015

MOISTURE RESISTANCE

The purpose of this test is to evaluate the moistureresistance of components under temperature/humidityconditions typical of tropical environments.

Typical Test Conditions: TA = −10°C to 65°C, rh = 80%to 98%, t = 24 hours/cycle, cycle = 10Common Failure Modes: Parametric shifts in leakageand mechanical failureCommon Failure Mechanisms: Corrosion orcontaminants on or within the package materials. Poorpackage sealingMilitary Reference: MIL-STD-750, Method 1021

http://onsemi.com17

SOLDERABILITY

The purpose of this test is to measure the ability of deviceleads/terminals to be soldered after an extended period ofstorage (shelf life).

Typical Test Conditions: Steam aging = 8 hours, Flux= R, Solder = Sn60, Sn63Common Failure Modes: Pin holes, dewetting,non-wettingCommon Failure Mechanisms: Poor plating,contaminated leadsMilitary Reference: MIL-STD-750, Method 2026

SOLDER HEAT

This test is used to measure the ability of a device towithstand the temperatures as may be seen in wave solderingoperations. Electrical testing is the endpoint criterion for thisstress.

Typical Test Conditions: Solder Temperature = 260°C,t = 10 secondsCommon Failure Modes: Parameter shifts, mechanicalfailureCommon Failure Mechanisms: Poor package designMilitary Reference: MIL-STD-750, Method 2031

STEADY STATE OPERATING LIFE (SSOL)

The purpose of this test is to evaluate the bulk stability ofthe die and to generate defects resulting from manufacturingaberrations that are manifested as time and stress-dependentfailures.

Typical Test Conditions: TA = 25°C, PD = Data Bookmaximum rating, t = 16 to 1000 hoursCommon Failure Modes: Parametric shifts andcatastrophicCommon Failure Mechanisms: Foreign material, crackdie, bulk die, metallization, wire and die bond defectsMilitary Reference: MIL-STD-750, Method 1026

TEMPERATURE CYCLING (AIR TO AIR)

The purpose of this test is to evaluate the ability of thedevice to withstand both exposure to extreme temperatures

and transitions between temperature extremes. This testingwill also expose excessive thermal mismatch betweenmaterials.

Typical Test Conditions: TA = −65°C to 200°C, cycle= 10 to 1000Common Failure Modes: Parametric shifts andcatastrophicCommon Failure Mechanisms: Wire bond, cracked orlifted die and package failureMilitary Reference: MIL-STD-750, Method 1051

THERMAL SHOCK (LIQUID TO LIQUID)

The purpose of this test is to evaluate the ability of thedevice to withstand both exposure to extreme temperaturesand sudden transitions between temperature extremes. Thistesting will also expose excessive thermal mismatchbetween materials.

Typical Test Conditions: TA = 0°C to 100°C, cycles= 10 to 1000Common Failure Modes: Parametric shifts andcatastrophicCommon Failure Mechanisms: Wire bond, cracked orlifted die and package failureMilitary Reference: MIL-STD-750, Method 1056

VARIABLE FREQUENCY VIBRATION

This test is used to examine the ability of the device towithstand deterioration due to mechanical resonance.

Typical Test Conditions: Peak acceleration = 20 g’s,Frequency range = 20 Hz to 20 kHz, t = 48 minutes.Common Failure Modes: Open, short, excessiveleakage, mechanical failureCommon Failure Mechanisms: Die and wire bonds,cracked die, package defectsMilitary Reference: MIL-STD-750, Method 2056

http://onsemi.com18

ZENER DIODE CHARACTERISTICS

INTRODUCTION

At first glance the zener diode is a simple deviceconsisting of one P-N junction with controlled breakdownvoltage properties. However, when considerations are givento the variations of temperature coefficient, zenerimpedance, thermal time response, and capacitance, all ofwhich are a function of the breakdown voltage (from 1.8 to400 V), a much more complicated picture arises. In additionto the voltage spectrum, a variety of power packages are onthe market with a variation of dice area inside theencapsulation.

This chapter is devoted to sorting out the importantconsiderations in a “typical” fashion. For exact details, thedata sheets must be consulted. However, much of theinformation contained herein is supplemental to the datasheet curves and will broaden your understanding of zenerdiode behavior.

Specifically, the following main subjects will be detailed:

Basic DC Volt-Ampere CharacteristicsImpedance versus Voltage and CurrentTemperature Coefficient versus Voltage and CurrentPower DeratingMountingThermal Time Response − Effective Thermal ImpedanceSurge CapabilitiesFrequency Response − Capacitance and Switching

Effects

BASIC ZENER DIODE DC VOLT-AMPERECHARACTERISTICS

Reverse and forward volt-ampere curves are representedin Figure 1 for a typical zener diode. The three areas −forward, leakage, and breakdown − will each be examined.

FORWARD DC CHARACTERISTICS

The forward characteristics of a zener diode areessentially identical with an “ordinary” rectifier and isshown in Figure 2. The volt-ampere curve follows the basicdiode equation of IF = IReqV/KT where KT/q equals about0.026 volts at room temperature and IR (reverse leakagecurrent) is dependent upon the doping levels of the P-Njunction as well as the area. The actual plot of VF versus IFdeviates from the theoretical due to slightly “fixed” seriesresistance of the lead wire, bonding contacts and some bulkeffects in the silicon.

Figure 1. Typical Zener Diode DC V-I Characteristics(Not to Scale)

FORWARD VOLTAGE

REVERSE VOLTAGE

BREAKDOWNREGION

LEAKAGE REGION

FORWARD CHARACTERISTIC

FO

RW

AR

D C

UR

RE

NT

RE

VE

RS

E C

UR

RE

NT

While the common form of the diode equation suggeststhat IR is constant, in fact IR is itself strongly temperaturedependent. The rapid increase in IR with increasingtemperature dominates the decrease contributed by theexponential term in the diode equation. As a result, theforward current increases with increasing temperature.Figure 2 shows a forward characteristic temperaturedependence for a typical zener. These curves indicate thatfor a constant current, an increase in temperature causes adecrease in forward voltage. The voltage temperaturecoefficient values are typically in the range of −1.4 to−2 mV/°C.

Figure 2. Typical Forward Characteristics ofZener Diodes

10.90.80.70.60.50.40.3

1000

VF, FORWARD VOLTAGE (VOLTS)

500

200

100

50

20

10

5

2

1

I ,

FO

RW

AR

D C

UR

RE

NT

(mA

)F

TJ = 150°C 100°C 25°C −55°C

http://onsemi.com19

LEAKAGE DC CHARACTERISTICS

When reverse voltage less than the breakdown is appliedto a zener diode, the behavior of current is similar to anyback-biased silicon P-N junction. Ideally, the reverse currentwould reach a level at about one volt reverse voltage andremain constant until breakdown is reached. There are boththeoretical and practical reasons why the typical V-I curvewill have a definite slope to it as seen in Figure 3.Multiplication effects and charge generation sites arepresent in a zener diode which dictate that reverse current(even at low voltages) will increase with voltage. Inaddition, surface charges are ever present across P-Njunctions which appear to be resistive in nature.

The leakage currents are generally less than onemicroampere at 150°C except with some large area devices.Quite often a leakage specification at 80% or so ofbreakdown voltage is used to assure low reverse currents.

Figure 3. Typical Leakage Currentversus Voltage

2016126400.1

1

10

100

1000

10000

I ,

RE

VE

RS

E L

EA

KA

GE

CU

RR

EN

T (n

A)

R

VR, REVERSE VOLTAGE (VOLTS)

TJ = 150°C

25°C

−55°C

VOLTAGE BREAKDOWN

At some definite reverse voltage, depending on the dopinglevels (resistivity) of the P-N junction, the current will beginto avalanche. This is the so-called “zener” or “breakdown”area and is where the device is usually biased during use. Atypical family of breakdown curves showing the effect oftemperature is illustrated in Figure 4.

Between the minimum currents shown in Figure 4 and theleakage currents, there is the “knee” region. The avalanchemechanism may not occur simultaneously across the entirearea of the P-N junction, but first at one microscopic site,then at an increasing number of sites as further voltage isapplied. This action can be accounted for by the“microplasma discharge” theory and correlates with severalbreakdown characteristics.

Figure 4. Typical Zener CharacteristicVariation with Temperature

32313029282726250.10.2

0.5

12

5

1020

50

100

200

5001000

VZ, ZENER VOLTAGE (VOLTS)

I ,

ZE

NE

R C

UR

RE

NT

(mA

)Z

T = −�55°C 100°C25°C 150°C

T = TJT = TA

An exaggerated view of the knee region is shown inFigure 5. As can be seen, the breakdown or avalanchecurrent does not increase suddenly, but consists of a seriesof smoothly rising current versus voltage increments eachwith a sudden break point.

Figure 5. Exaggerated V-I Characteristicsof the Knee Region

EXAGGERATED V-I

OF KNEE REGION

ZENER VOLTAGE

VOLTAGE

ZE

NE

R C

UR

RE

NT

CU

RR

EN

T

At the lowest point, the zener resistance (slope of thecurve) would test high, but as current continues to climb, theresistance decreases. It is as though each discharge site hashigh resistance with each succeeding site being in paralleluntil the total resistance is very small.

In addition to the resistive effects, the micro plasmas mayact as noise generators. The exact process of manufacturingaffects how high the noise will be, but in any event there willbe some noise at the knee, and it will diminish considerablyas current is allowed to increase.

Since the zener impedance and the temperaturecoefficient are of prime importance when using the zenerdiode as a reference device, the next two sections willexpand on these points.

http://onsemi.com20

ZENER IMPEDANCE

The slope of the VZ − IZ curve (in breakdown) is definedas zener impedance or resistance. The measurement isgenerally done with a 60 Hz (on modern, computerizedequipment this test is being done at 1 kHz) current variationwhose value is 10% in rms of the dc value of the current.(That is, ΔIZ peak to peak = 0.282 IZ.) This is really not asmall signal measurement but is convenient to use and givesrepeatable results.

The zener impedance always decreases as currentincreases, although at very high currents (usually beyond IZmax) the impedance will approach a constant. In contrast,the zener impedance decreases very rapidly with increasingcurrent in the knee region. On Semiconductor specifies mostzener diode impedances at two points: IZT and IZK. The termIZT usually is at the quarter power point, and IZK is anarbitrary low value in the knee region. Between these twopoints a plot of impedance versus current on a log-log scaleis close to a straight line. Figure 6 shows a typical plot of ZZversus IZ for a 20 volt−500 mW zener. The worst caseimpedance between IZT and IZK could be approximated byassuming a straight line function on a log-log plot; however,at currents above IZT or below IZK a projection of this linemay give erroneous values.

Figure 6. Zener Impedance versusZener Current

ZENER CURRENT (mA)

1001010.11

10

100

1000

ZE

NE

R IM

PE

DA

NC

E (

OH

MS

) APPROXIMATE MAXIMUM LINE

ZZT(MAX)

ZZK(MAX)

The impedance variation with voltage is much morecomplex. First of all, zeners below 6 volts or so exhibit “fieldemission” breakdown converting to “avalanche” at highercurrents. The two breakdowns behave somewhat differentlywith “field emission” associated with high impedance andnegative temperature coefficients and “avalanche” withlower impedance and positive temperature coefficients.

A V-I plot of several low voltage 500 mW zener diodes isshown in Figure 7. It is seen that at some given current(higher for the lower voltage types) there is a fairly suddendecrease in the slope of ΔV/ΔI. Apparently, this current is thetransition from one type of breakdown to the other. Above6 volts the curves would show a gradual decrease of ΔV/ΔIrather than an abrupt change, as current is increased.

Figure 7. Zener Current versus Zener Voltage(Low Voltage Region)

1

100

10

1

0.1

0.012 3 4 5 6 7 8

ZENER VOLTAGE (VOLTS)

ZE

NE

R C

UR

RE

NT

(mA

)

Possibly the plots shown in Figure 8 of zener impedanceversus voltage at several constant IZ’s more clearly pointsout this effect. It is obvious that zener diodes whosebreakdowns are about 7 volts will have remarkably lowimpedance.

20010070503020107532

1 mA

10 mA

20 mA

Figure 8. Dynamic Zener Impedance (Typical)versus Zener Voltage

2

3

57

10

20

30

5070

100

200Z

, D

YN

AM

IC IM

PE

DA

NC

E (

OH

MS

)Z


TA = 25°C

IZ(ac) = 0.1 IZ(dc)

However, this is not the whole picture. A zener diodefigure of merit as a regulator could be ZZ/VZ. This wouldgive some idea of what percentage change of voltage couldbe expected for some given change in current. Of course, alow ZZ/VZ is desirable. Generally zener current must bedecreased as voltage is increased to prevent excessive powerdissipation; hence zener impedance will rise even higher andthe “figure of merit” will become higher as voltageincreases. This is the case with IZT taken as the test point.However, if IZK is used as a comparison level in thosedevices which keep a constant IZK over a large range ofvoltage, the “figure of merit” will exhibit a bowl-shapedcurve − first decreasing and then increasing as voltage isincreased. Typical plots are shown in Figure 9. Theconclusion can be reached that for uses where wide swingsof current may occur and the quiescent bias current must behigh, the lower voltage zener will provide best regulation,

http://onsemi.com21

but for low power applications, the best performance couldbe obtained between 50 and 100 volts.

703010 50 90 110 130 150

0.85 mA

17 mA250 mA

10 W, ZZT(MAX)

1.7 mA

SEE NOTE BELOW

1.3 mA

12.5 mA

3.8 mA

28 mA

75 mA


ZE

NE

R IM

PE

DA

NC

E (

MA

X)/

ZE

NE

R V

OLT

AG

E

10

100

1

0.1

(NOTE: CURVE IS APPROXIMATE, ACTUAL

ZZ(MAX) IS ROUNDED OFF TO NEAREST

WHOLE NUMBER ON A DATA SHEET)

Figure 9. Figure of Merit: ZZ(Max)/VZ versus VZ(400 mW & 10 W Zeners)

400 mW, ZZK(MAX) AT 0.25 mA

10 W, ZZK(MAX) AT 1 mA

400 mW, ZZT(MAX)

TEMPERATURE COEFFICIENT

Below three volts and above eight volts the zener voltagechange due to temperature is nearly a straight line functionand is almost independent of current (disregardingself-heating effects). However, between three and eightvolts the temperature coefficients are not a simple affair. Atypical plot of TC versus VZ is shown in Figure 10.

2 3 4 5 6 7 8 9 11

0.01 mA

0.1 mA30 mA

1 mA

10 mA

Figure 10. Temperature Coefficient versusZener Voltage at 25°C Conditions Typical

7

6

5

4

3

2

1

0

−1

−2

−3


10 12

T

, TE

MP

ER

ATU

RE

CO

EF

FIC

IEN

TS

(m

V/�

C)

C°

VZ REFERENCE AT IZ = IZT

& TA = 25°C

Any attempt to predict voltage changes as temperaturechanges would be very difficult on a “typical” basis. (This,of course, is true to a lesser degree below three volts andabove eight volts since the curve shown is a typical one andslight deviations will exist with a particular zener diode.) Forexample, a zener which is 5 volts at 25°C could be from 4.9to 5.05 volts at 75°C depending on the current level.Whereas, a zener which is 9 volts at 25°C would be close to

9.3 volts at 75°C for all useful current levels (disregardingimpedance effects).

As was mentioned, the situation is further complicated bythe normal deviation of TC at a given current. For example,for 7.5 mA the normal spread of TC (expressed in %/°C) isshown in Figure 11. This is based on limited samples and inno manner implies that all On Semiconductor zenersbetween 2 and 12 volts will exhibit this behavior. At othercurrent levels similar deviations would occur.

+0.08

+0.06

+0.04

+0.02

0

−0.02

−0.08

−0.10

−0.04

−0.06

0 2 4 6 8 10 12 14


TYPICAL

MAX

Figure 11. Temperature Coefficient Spreadversus Zener Voltage

TE

MP

ER

ATU

RE

CO

EF

FIC

IEN

T (%

/�C

)°

MAX

MIN

MIN

TYPICAL

IZT = 7.5 mA

Obviously, all of these factors make it very difficult toattempt any calculation of precise voltage shift due totemperature. Except in devices with specified maximumT.C., no “worse case” design is possible. Informationconcerning the On Semiconductor temperaturecompensated or reference diodes is given in Chapter 4.

Typical temperature characteristics for a broad range ofvoltages is illustrated in Figure 12. This graphically showsthe significant change in voltage for high voltage devices(about a 20 volt increase for a 100°C increase on a 200 voltdevice).

NOTE: DV IS + ABOVE 5 VOLTS− BELOW 4.3 VOLTSBETWEEN 4.3 & 5 VOLTSVARIES ABOUT + 0.08 VOLTS


1 2 3 5 10 50 100 200 1,000

100

10

1

0.1

Figure 12. Typical TemperatureCharacteristics

V

(+2

5 C

TO

+12

5 C

)°

°Δ

Z

0.01

http://onsemi.com22

POWER DERATING AND MOUNTING

The zener diode like any other semiconductor has amaximum junction temperature. This limit is somewhatarbitrary and is set from a reliability viewpoint. Mostsemiconductors exhibit an increasing failure rate astemperature increases. At some temperature, the solder willmelt or soften and the failure rate soars. The 175°C to 200°Cjunction temperature rating is quite safe from solder failuresand still has a very low failure rate.

In order that power dissipated in the device will never causethe junction to rise beyond 175°C or 200°C (depending on thedevice), the relation between temperature rise and powermust be known. Of course, the thermal resistance (RθJA orRθJL) is the factor which relates power and temperature in thewell known “Thermal Ohm’s Law’’ relation:

ΔT = TJ − TA = RθJAPZ (1)and

ΔT = TJ − TL = RθJLPZ (2)where

TJTATLRθJARθJLPZ

= Junction temperature= Ambient temperature= Lead temperature= Thermal resistance junction to ambient= Thermal resistance junction to lead= Zener power dissipation

Obviously, if ambient or lead temperature is known andthe appropriate thermal resistance for a given device isknown, the junction temperature could be preciselycalculated by simply measuring the zener dc current andvoltage (PZ = IZVZ). This would be helpful to calculatevoltage change versus temperature. However, onlymaximum and typical values of thermal resistance are givenfor a family of zener diodes. So only “worst case” or typicalinformation could be obtained as to voltage changes.

The relations of equations 1 and 2 are usually expressedas a graphical derating of power versus the appropriatetemperature. Maximum thermal resistance is used togenerate the slope of the curve. An example of a 400milliwatt device derated to the ambient temperature and a 1watt device derated to the lead temperature are shown inFigures 13 and 14.

500

25 50 75 100 125 150 175 200

400

300

200

100

0

TA, AMBIENT TEMPERATURE (°C)

Figure 13. 400 mW Power TemperatureDerating Curve

P

, PO

WE

R D

ISS

IPA

TIO

N (

MIL

LIW

AT

TS

)D

1.25

0 20 40 60 80 100 120 140 160 180 200

L = LEAD LENGTH

TO HEAT SINK1

0.75

0.50

0.25

L = 1/8″L = 3/8″

L = 1″

TL, LEAD TEMPERATURE (°C)

Figure 14. Power TemperatureDerating Curve

P

, MA

XIM

UM

PO

WE

R D

ISS

IPA

TIO

N (

WA

TT

S)

D

A lead mounted device can have its power ratingincreased by shortening the lead length and “heatsinking”the ends of the leads. This effect is shown in Figure 15, forthe 1N4728, 1 watt zener diode.

Each zener has a derating curve on its data sheet andsteady state power can be set properly. However,temperature increases due to pulse use are not so easilycalculated. The use of “Transient Thermal Resistance”would be required. The next section expounds upontransient thermal behavior as a function of time and surgepower.

175

0

150

125

100

75

50

25

00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

L, LEAD LENGTH TO HEAT SINK (INCH)

Figure 15. Typical 1N4728 ThermalResistance versus Lead Length

R�

JL, J

UN

CT

ION

−TO

−LE

AD

TH

ER

MA

L R

ES

ISTA

NC

E (°C

/W)

THERMAL TIME RESPONSE

Early studies of zener diodes indicated that a “thermaltime constant” existed which allowed calculation oftemperature rise as a function of power pulse height, width,and duty cycle. More precise measurements have shown thattemperature response as a function of time cannot berepresented as a simple time constant. Although as shown inthe preceding section, the steady state conditions areanalogous in every way to an electrical resistance; a simple“thermal capacitance” placed across the resistor is not thetrue equivalent circuit. Probably a series of parallel R-C

http://onsemi.com23

networks or lumped constants representing a thermaltransmission line would be more accurate.

Fortunately a concept has developed in the industrywherein the exact thermal equivalent circuit need not befound. If one simply accepts the concept of a thermalresistance which varies with time in a predictable manner,the situation becomes very practical. For each family ofzener diodes, a “worst case” transient thermal resistancecurve may be generated.

The main use of this transient RθJL curve is when the zeneris used as a clipper or a protective device. First of all, thepower wave shape must be constructed. (Note, even thoughthe power-transient thermal resistance indicates reasonablejunction temperatures, the device still may fail if the peakcurrent exceeds certain values. Apparently a currentcrowding effect occurs which causes the zener to short. Thisis discussed further in this section.)

TRANSIENT POWER-TEMPERATURE EFFECTS

A typical transient thermal resistance curve is shown inFigure 16. This is for a lead mounted device and shows theeffect of lead length to an essentially infinite heatsink.

To calculate the temperature rise, the power surge waveshape must be approximated by its rectangular equivalent asshown in Figure 17. In case of an essentially non-recurrentpulse, there would be just one pulse, and ΔT = RθT1 Pp. Inthe general case, it can be shown that

whereDRθT1

RθT

RθT1 + T

RθJA(ss) or RθJL(ss) = Steady state value of thermalRθJA(ss) or RθJL(ss) = resistance

= Duty cycle in percent= Transient thermal resistance at the time

equal to the pulse width= Transient thermal resistance at the time

equal to pulse interval= Transient thermal resistance at the time

equal to the pulse interval= plus one more pulse width.

ΔT = [DRθJA (ss) + (1 − D) RθT1 + T + RθT1 − RθT] PP

PW, PULSE WIDTH (ms)

100

3 5 10 20 50 100 200 500 1000 2000 5000 10k 30k

70

50

30

20

10

7

5

3

2

1

L = 1/32″

L = 1″

FOR θJL(t) VALUES AT PULSE WIDTHSLESS THAN 3.0 ms, THE ABOVECURVE CAN BE EXTRAPOLATEDDOWN TO 10 μs AT A CONTINUINGSLOPE OF 1/2

Figure 16. Typical Transient ThermalResistance (For Axial Lead Zener)

L L

HEAT SINKÉÉÉÉ

ÉÉÉÉ

TH

ER

MA

L R

ES

ISTA

NC

E (

C/W

)°

RJL

(t),

JUN

CT

ION

-TO

-LE

AD

TR

AN

SIE

NT

θ

Figure 17. Relation of Junction Temperature toPower Pulses

T

T1T1

T

PEAKTEMPERATURE RISE

AVERAGETEMPERATURE RISE

AMBIENTTEMPERATURE

PEAK POWER (PP)

AVERAGE POWER = PP

This method will predict the temperature rise at the end ofthe power pulse after the chain of pulses has reachedequilibrium. In other words, the average power will havecaused an average temperature rise which has stabilized, buta temperature “ripple” is present.

Example: (Use curve in Figure 16)PP = 5 watt (Lead length 1/32″)D = 0.1T1 = 10 msT = 100 msRθJA(ss) = 12°C/W (for 1/32″ lead length)

ThenRθT1 = 1.8°C/WRθT = 5.8°C/WRθT1 + T = 6°C/W

AndΔT = [0.1 x 12 + (1 − 0.1) 6 + 1.8 − 5.8] 5ΔT = 13°C

Or at TA = 25°, TJ = 38°C peak

SURGE FAILURES

If no other considerations were present, it would be asimple matter to specify a maximum junction temperatureno matter what pulses are present. However, as has beennoted, apparently other fault conditions prevail. The samegroup of devices for which the transient thermal curves weregenerated were tested by subjecting them to single shotpower pulses. A failure was defined as a significant shift ofleakage or zener voltage, or of course opens or shorts. Eachdevice was measured before and after the applied pulse.Most failures were shifts in zener voltage. The results areshown in Figure 18.

Attempts to correlate this to the transient thermalresistance work quite well on a typical basis. For example,assuming a value for 1 ms of 90 watts and 35 watts at 10 ms,the predicted temperature rise would be 180°C and 190°C.But on a worst case basis, the temperature rises would beabout one half these values or junction temperatures, on theorder of 85°C to 105°C, which are obviously low.Apparently at very high power levels a current restrictionoccurs causing hot spots. There was no apparent correlation

http://onsemi.com24

of zener voltage or current on the failure points since eachgroup of failures contained a mixture of voltages.

1000

0.00001 0.0001 0.001 0.01 0.1 1

WORSE CASE

100

10

1

PO

WE

R (

WA

TT

S)

Figure 18. One Shot Power Failure AxialLead Zener Diode

TIME OF PULSE (SECONDS)

TYPICAL

VOLTAGE VERSUS TIME

Quite often the junction temperature is only of academicinterest, and the designer is more concerned with the voltagebehavior versus time. By using the transient thermalresistance, the power, and the temperature coefficient, thedesigner could generate VZ versus time curves. TheOn Semiconductor zener diode test group has observeddevice voltages versus time until the thermal equilibriumwas reached. A typical curve is shown for a lead mountedlow wattage device in Figure 19 where the ambienttemperature was maintained constant. It is seen that voltagestabilizes in about 100 seconds for 1 inch leads.

Since information contained in this section may not befound on data sheets it is necessary for the designer tocontact the factory when using a zener diode as a surgesuppressor. Additional information on transient suppressionapplication is presented elsewhere in this book.

166

0.01

165

164

163

162

161

1600.1 1 10 100

TIME (SECONDS)

ZE

NE

R V

OLT

AG

E (

VO

LTS

)

Figure 19. Zener Voltage (Typical) versusTime for Step Power Pulses

(500 mW Lead Mounted Devices)

FREQUENCY AND PULSE CHARACTERISTICS

The zener diode may be used in applications whichrequire a knowledge of the frequency response of the device.Of main concern are the zener resistance (usually specifiedas “impedance”) and the junction capacitance. Thecapacitance curves shown in this section are typical.

ZENER CAPACITANCE

Since zener diodes are basically PN junctions operated inthe reverse direction, they display a capacitance thatdecreases with increasing reverse voltage. This is so becausethe effective width of the PN junction is increased by theremoval of charges (holes and electrons) as reverse voltageis increased. This decrease in capacitance continues until thezener breakdown region is entered; very little furthercapacitance change takes place, owing to the now fixedvoltage across the junction. The value of this capacitance isa function of the material resistivity, ρ, (amount of doping −which determines VZ nominal), the diameter, D, of junctionor dice size (determines amount of power dissipation), thevoltage across the junction VC, and some constant, K. Thisrelationship can be expressed as:

KD4pVC

�CC =n KD4

ρVC

After the junction enters the zener region, capacitanceremains relatively fixed and the AC resistance thendecreases with increasing zener current.

TEST CIRCUIT CONSIDERATIONS: A capacitivebridge was used to measure junction capacitance. In thismethod the zener is used as one leg of a bridge that isbalanced for both DC at a given reverse voltage and for AC(the test frequency 1 MHz). After balancing, the variablecapacitor used for balancing is removed and its valuemeasured on a test instrument. The value thus indicated isthe zener capacitance at reverse voltage for which bridgebalance was obtained. Figure 20 shows capacitance testcircuit.

Figure 21 is a plot of junction capacitance for diffusedzener diode units versus their nominal operating voltage.Capacitance is the value obtained with reverse bias set atone-half the nominal VZ. The plot of the voltage range from6.8 V to 200 V, for three dice sizes, covers mostOn Semiconductor diffused-junction zeners. Consultspecific data sheets for capacitance values.

Figures 22, 23, and 24 show plots of capacitance versusreverse voltage for units of various voltage ratings in eachof the three dice sizes. Junction capacitance decreases asreverse voltage increases to the zener region. This change incapacitance can be expressed as a ratio which follows aone-third law, and C1/C2 = (V2/V1)1/3. This law holds onlyfrom the zener voltage down to about 1 volt, where the curvebegins to flatten out. Figure 25 shows this for a group of lowwattage units.

http://onsemi.com25

Figure 20. Capacitance Test Circuit

C1

C2

VDC1 k1% 1 k

1%

1 k

1%

1 k1%

DC

POWER

SUPPLY

HP

NO. 712A

100 W

IV

VAC

1 MHz

SIGNAL

GEN

TEK

NO. 190A

10/50

pF

XZENER

UNDER

TEST

HI-GAIN

DIFF SCOPE

NULL IND

TEK

TYPE D

R = ZR R

CAP

DECADE

0−.09 mF

100 pF

STEPS

ΔC 1%

BAL READ

S1

10/150pF

L/C

METER

TEK

140

0.1 μF

1,000

1 10 100


100

10

C

, CA

PAC

ITA

NC

E (

PIC

OFA

RA

DS

)Z

100 VOLTS

50 VOLTS

20 VOLTS

10 VOLTS

10,000

1 10 100 1,000

1,000

100

10C

, CA

PAC

ITA

NC

E (

PIC

OFA

RA

DS

) @

VZ

/ 2

VZ, NOMINAL UNIT VOLTAGE (VOLTS)

Z

HIGH WATTAGE

LOW WATTAGE

MEDIUM WATTAGE

Figure 21. Capacitance versus Voltage Figure 22. Capacitance versus Reverse Voltage

LOW WATTAGE UNITS

AVG. FOR 10 UNITS EACH

10,000

1 10


100

C

, CA

PAC

ITA

NC

E (

PIC

OFA

RA

DS

)Z

100

1,000

10,000

1 10 1,000


100

10

C

, CA

PAC

ITA

NC

E (

PIC

OFA

RA

DS

)Z

100

1,000

Figure 23. Capacitance versusReverse Voltage

Figure 24. Capacitance versusReverse Voltage

100 VOLTS

50 VOLTS

20 VOLTS

10 VOLTS

100 VOLTS

50 VOLTS

20 VOLTS

10 VOLTSMEDIUM WATTAGE UNITS


HIGH WATTAGE UNITS


http://onsemi.com26

C

, CA

PAC

ITA

NC

E (

PIC

OFA

RA

DS

)Z

100

1,000

101 10


1000.1

LOW WATTAGEUNITS

Figure 25. Flattening of Capacitance Curve atLow Voltages

100 VOLTS

50 VOLTS

10 VOLTS

Figure 26. Impedance Test Circuit

1 k 1 k

R2

1 Ω

RX = ZZ

DC

SUPPLY

HP

712A

mA

DC

E1

E2

Rx =E1 − E2

E2

0.1 μF

600READ S1 A

READ

SET

SET

10M

100 pF

S1 B

SIGNAL

GEN

HP

650A

AC

VTVM

HP

400H

DC

VTVM

HP

412A

ZENER IMPEDANCE

Zener impedance appears primarily as composed of acurrent-dependent resistance shunted by avoltage-dependent capacitor. Figure 26 shows the testcircuit used to gather impedance data. This is avoltage-impedance ratio method of determining theunknown zener impedance. The operation is as follows:

(1) Adjust for desired zener IZDC by observing IR dropacross the 1-ohm current-viewing resistor R2.

(2) Adjust IZAC to 100 μA by observing AC IR drop acrossR2.

(3) Measure the voltage across the entire network byswitching S1. The ratio of these two AC voltages isthen a measure of the impedance ratio. This can beexpressed simply as RX = [(E1 − E2)/E2] R2.

Section A of S1 provides a dummy load consisting of a10-M resistor and a 100 pF capacitor. This network isrequired to simulate the input impedance of the AC VTVMwhile it is being used to measure the AC IR drop across R2.

This method has been found accurate up to about threemegahertz; above this frequency, lead inductances and strapcapacitance become the dominant factors.

http://onsemi.com27

Figure 27 shows typical impedance versus frequencyrelationships of 6.8 volt 500 mW zener diodes at various DCzener currents. Before the zener breakdown region isentered, the impedance is almost all reactive, being providedby a voltage-dependent capacitor shunted by a very highresistance. When the zener breakdown region is entered, thecapacitance is fixed and now is shunted bycurrent-dependent resistance. For comparison, Figure 27also shows the plot for a 680 pF capacitor XC, a 1K 1%nonreactive resistor, R, and the parallel combination of thesetwo passive elements, ZT.

ZTIZMA

2

10

20

1K & 680 pF

R, 1K 1% DC

10,000

10 100 1 kHz 10 kHz 100 kHz 1 MHz 10 MHz

FREQUENCY (Hz)

1,000

100

10

1

Z

, ZE

NE

R IM

PE

DA

NC

E (

OH

MS

)Z

X , 680 pFC

Figure 27. Zener Impedanceversus Frequency

1.00

2.50

5

0.25

0.50

HIGH FREQUENCY AND SWITCHINGCONSIDERATIONS

At frequencies about 100 kHz or so and switching speedsabove 10 microseconds, shunt capacitance of zener diodesbegins to seriously effect their usefulness. The upper photoof Figure 28 shows the output waveform of a symmetricalpeak limiter using two zener diodes back-to-back. Thecapacitive effects are obvious here. In any application wherethe signal is recurrent, the shunt capacitance limitations canbe overcome, as lower photo of Figure 28 shows. This isdone by operating fast diodes in series with the zener. Uponapplication of a signal, the fast diode conducts in the forwarddirection charging the shunt zener capacitance to the levelwhere the zener conducts and limits the peak. When thesignal swings the opposite direction, the fast diode becomesback-biased and prevents fast discharge of shuntcapacitance. The fast diode remains back-biased when thesignal reverses again to the forward direction and remainsoff until the input signal rises and exceeds the charge levelof the capacitor. When the signal exceeds this level, the fastdiode conducts as does the zener. Thus, between successivecycles or pulses the charge in the shunt capacitor holds offthe fast diode, preventing capacitive loading of the signaluntil zener breakdown is reached. Figures 29 and 30 showthis method applied to fast-pulse peak limiting.

5 V/cm

0.5 μs/cm

0.5 μs/cm

5 V/cm

Figure 28. Symmetrical Peak Limiter

RS

RS

ei eo

ei eo

http://onsemi.com28

2 V/cm

20 ns/cm

ei

eo

2 V/cm

eo

ei

20 ns/cm

Figure 29. Shunt Clipper

Figure 30. Shunt Clipper with Clamping Network

200 Ω

50 Ωei eo10 V

Z

200 Ω

50 Ωei

0.001

eo

http://onsemi.com29

Figure 31 is a photo of input-output pulse waveformsusing a zener alone as a series peak clipper. The smalleroutput waveform shows the capacitive spike on the leading

edge. Figure 32 clearly points out the advantage of theclamping network.

2 V/cm

20 ns/cm

eo

ei

2 V/cmeo

ei

20 ns/cm

0

Figure 31. Basic Series Clipper

Figure 32. Series Clipper with Clamping Network

10 VZ

50 Ω

50 Ω

ei eo

ei eo

200 Ω

10 VZ

200 Ω

.001

http://onsemi.com30

TEMPERATURE COMPENSATED ZENERS

INTRODUCTION

A device which provides reference voltages in a specialmanner is a reference diode.

As was discussed in the preceding chapters, the zenerdiode has the unique characteristic of exhibiting either apositive or a negative temperature coefficient, or both. Byproperly employing this phenomenon in conjunction withother semiconductor devices, it is possible to manufacture azener reference element exhibiting a very low temperaturecoefficient when properly used. This type of lowtemperature coefficient device is referred to as a referencediode.

INTRODUCTION TO REFERENCE DIODES

The temperature characteristics of the zener diode arediscussed in a previous chapter, where it was shown thatchange in zener voltage with temperature can be significantunder severe ambient temperature changes (for example, a100 V zener can change 12.5 volts from 0 to 125°C). Thereference diode (often called the temperature compensatedzener or the TC zener) is specially designed to minimizethese specific temperature effects.

Design of temperature compensated zeners make possibledevices with voltage changes as low as 5 mV from −55 to+100°C, consequently, the advantages of the temperaturecompensated zener are obvious. In critical applications, asa voltage reference in precision dc power supplies, in highstability oscillators, in digital voltmeters, in frequencymeters, in analog-to-digital converters, or in other precisionequipment, the temperature compensated zener is anecessity.

Conceivably temperature compensated devices can bedesigned for any voltage but present devices with optimumvoltage temperature characteristics are limited to specificvoltages. Each family of temperature compensated zeners isdesigned by careful selection of its integral parts with specialattention to the use conditions (temperature range andcurrent). A distinct operating current is associated with eachdevice. Consequently, changes from the specified operatingcurrent can only degrade the voltage-temperaturerelationships. This will be discussed in more detail later.

The device “drift” or voltage-time stability is critical insome reference applications. Typically zeners and TCzeners offer stability of better than 500 parts per million per1000 hours.

TEMPERATURE CHARACTERISTICS OF THE P-NJUNCTION AND COMPENSATION

The voltage of a forward biased P-N junction, at a specificcurrent, will decrease with increasing temperature. Thus, adevice so biased displays a negative temperature coefficient(Figure 1). A P-N junction in avalanche (above 5 voltsbreakdown) will display a positive temperature coefficient;that is, voltage will increase as temperature increases. Dueto energy levels of a junction which breaks down below5 volts, the temperature coefficient is negative.

It follows that various combinations of forward biasedjunctions and reverse biased junctions may be arranged toachieve temperature compensation. From Figure 2 it can beseen that if the absolute value of voltage change (ΔV) is thesame for both the forward biased diode and the zener diodewhere the temperature has gone from 25°C to 100°C, thenthe total voltage across the combination will be the same atboth temperatures since one ΔV is negative and the otherpositive. Furthermore, if the rate of increase (or decrease) isthe same throughout the temperature change, voltage willremain constant. The non-linearity associated with thevoltage temperature characteristics is a result of this rate ofchange not being a perfect match.

VREF = VZ + ΔVZ + VD − ΔVD

THE METHODS OF TEMPERATURECOMPENSATION

The effect of temperature is shown in Figure 1. Theforward characteristic does not vary significantly withreverse voltage breakdown (zener voltage) rating. A changein ambient temperature from 25° to 100°C produces a shiftin the forward curve in the direction of lower voltage (anegative temperature coefficient — in this case about150 mV change), while the same temperature changeproduces approximately 1.9 V increase in the zener voltage(a positive coefficient). By combining one or more silicondiodes biased in the forward direction with the P-N biasedzener diode as shown in Figure 3, it is possible tocompensate almost completely for the zener temperaturecoefficient. Obviously, with the example shown, 13junctions would be needed. Usually reference diodes are lowvoltage devices, using zeners with 6 to 8 volts breakdownand one or two forward diodes.

http://onsemi.com31

Figure 1. Effects of Temperature on Zener Diode Characteristics

FORWARD

CHARACTERISTIC

TYPICAL

(ALL TYPES)

VZ (VOLTS)

VF (VOLTS)

30 20 10

15

30

45

I� (

mA

)Z

0.5 1 1.5

450

300

150

1.9 V

100°C

25°C

100°C

150 mV

25°C

I� (

mA

)F

Figure 2. Principle of Temperature Compensation

Figure 3. Zener Temperature Compensation with Silicon Forward Junctions

DIRECTION OF CURRENT FLOW

PACKAGE

OUTLINE

FORWARD-BIASEDPN JUNCTION

REVERSE-BIASEDZENER JUNCTION

+

7.5 mA

100°C

25°C

+V

−ΔV

+ΔV +ΔV

100°C 25°C

−V

100°C 25°C

100°C

25°C7.5 mA

−

−

−

SILICON JUNCTIONDIODES

ZENER DIODES

+

−

http://onsemi.com32

In ac regulator and clipper circuits where zener diodes arenormally connected cathode to cathode, the forward biaseddiode during each half cycle can be chosen with the correctforward temperature coefficient (by stacking, etc.) tocorrectly compensate for the temperature coefficient of thereverse-biased zener diode. It is possible to compensate forvoltage drift with temperature using this method to theextent that zener voltage stabilities on the order of0.001%/°C are quite feasible.

This technique is sometimes employed where higherwattage devices are required or where the zener iscompensated by the emitter base junction of a transistorstage. Consider the example of using discrete components,1N4001 rectifier and ON Semiconductor 5 Watt zener, toobtain compensated voltage-temperature characteristics.Examination of the curve in Figure 4 indicates that a 10 voltzener diode exhibits a temperature coefficient ofapproximately +5.5 mV/°C. At a current level of 100 mA atemperature coefficient of approximately −2.0 mV/°C ischaracteristic of the 1N4001 rectifiers. A series connectionof three silicon 1N4001 rectifiers produces a totaltemperature coefficient of approximately −6 mV/°C and atotal forward drop of approximately 2.17 volts at 25°C. Thecombination of three silicon rectifiers and the 10 volt zenerdiode produces a device with a coefficient of approximately−0.5 mV/°C and a total breakdown voltage at 100 mA ofapproximately 12.2 volts. Calculation shows this to be atemperature stability of −0.004%/°C.

�� 0.5 mV� C12.2 V

�� 100�

The temperature compensated zener employs thetechnique of specially selected dice. This provides optimumvoltage temperature characteristics by close control of diceresistivities.

6

5

4

3

2

1

0

−1

−2

−30 1 2 3 4 5 6 7 8 9 10 11 12 13

ZENER VOLTAGE (10 mA AT 25°C)7

ALLOY-DIFFUSEDJUNCTION

THREEFORWARDS

ONEFORWARD

TWOFORWARDS

DIFFUSED JUNCTION

VOLTS

Figure 4.

mV

/ C°

TEMPERATURE COEFFICIENT STABILITY

Figure 5 shows the voltage-temperature characteristics ofthe TC diode. It can be seen that the voltage drops slightlywith increasing temperature.

VO

LTA

GE

(V

OLT

S)

mV

CH

AN

GE

FR

OM

25

C V

OLT

AG

E°

6.326

6.324

6.322

6.320

6.318

6.316

6.314

TEMPERATURE (°C)

−55 −10 25 62 100

Figure 5. Voltage versus Temperature,Typical for ON Semiconductor 1N827

Temperature Compensated Zener Diode

65

4

3

2

1

0

− 1

− 2

− 3

− 4

− 5

− 6

This non-linearity of the voltage temperaturecharacteristic leads to a definition of a representative designparameter ΔVZ. For each device type there is a specifiedmaximum change allowable. The voltage temperaturestability measurement consists of voltage measurement atspecified temperatures (for the 1N821 Series thetemperatures are −55, 0, +25, +75, and +100°C). The voltagereadings at each of the temperatures is compared withreadings at the other temperatures and the largest voltagechange between any of the specified temperaturesdetermines the exact device type. For devices registeredprior to complete definition of the voltage temperaturestability measurement, the allowable maximum voltagechange over the temperature range is derived from thecalculation converting %/°C to mV over the temperaturerange. Under this standard definition, %/°C is merely anomenclature and the meaningful allowable voltagedeviation to be expected becomes the designed parameter.

http://onsemi.com33

CURRENT

Thus far, temperature compensated zeners have beendiscussed mainly with regard to temperature and voltage.However, the underlying assumption throughout theprevious discussion was that current remained constant.

There is a significant change in the temperaturecoefficient of a unit depending on how much above or belowthe test current the device is operated.

A particular unit with a 0.01%/°C temperature coefficientat 7.5 mA over a temperature range of −55°C to +100°Ccould possibly have a 0.0005%/°C temperature coefficientat 11 mA. In fact, there is a particular current which can bedetermined for each individual unit that will give the lowestTC.

Manufacturing processes are designed so that the yields oflow TC units are high at the test specification for current. Aunit with a high TC at the test current can have a low TC atsome other current. A look at the volt-ampere curves atdifferent temperatures illustrates this point clearly (seeFigure 6).

VOLTAGE (VOLTS)−6.6 −6.5 −6.4 −6.3 −6.2 −6.1 −6 −5.9

IB

IA

IC

CU

RR

EN

T

25°C

−55+100

ΔVB

B

A

CΔVC

Figure 6. Voltage-Ampere Curves ShowingCrossover at A

If the three curves intersect at A, then operation at IAresults in the least amount of voltage deviation due totemperature from the +25°C voltage. At IB and IC there aregreater excursions (ΔVB and ΔVC) from the +25°C voltageas temperature increases or decreases.

THE EFFECTS OF POOR CURRENTREGULATION

If current shifts (randomly or as a function oftemperature), then an area of operation can be defined for thetemperature compensated zener.

Once again the curves are drawn, this time a shaded areais shown on the graph. The upper and lower extremitiesdenote the maximum current values generated by the currentsupply while the voltage extremes at each current are shownby the left and right sides of the area, shown in Figure 7.

VOLTAGE (VOLTS)

−6.6 −6.5 −6.4 −6.3 −6.2 −6.1 −6 −5.9

CU

RR

EN

T

25°C

−55 +100

ΔIMAXΔVMAX

Figure 7. Effects of Poorly Regulated Current

The three volt-ampere curves do not usually cross over atexactly the same point. However, this does not take awayfrom the argument that current regulation is probably themost critical consideration when usingtemperature-compensated units.

ZENER IMPEDANCE AND CURRENTREGULATION

Zener impedance is defined as the slope of the V-I curveat the test point corresponding to the test current. It ismeasured by superimposing a small ac current on the dc testcurrent and then measuring the resulting ac voltage. Thisprocedure is identical with that used for regular zeners.

Impedance changes with temperature, but the variation isusually small and it can be assumed that the amount ofcurrent regulation needed at +25°C will be the same for othertemperatures.

As an example, one might want to determine the amountof current regulation necessary for the device describedbelow when the maximum deviation in voltage due tocurrent variation is ±5 millivolts.

VZT = 6.32 VIZT = 7.5 mAZZT = 15 Ω @ +25°CΔV = ΔI⋅VZZT0.005 = ΔI⋅V15

ΔI =0.005

15= 0.33 mA

Therefore, the current cannot vary more than 0.33 mA.The amount of current regulation necessary is:

0.337.5

x 100% = 4.5% regulation.

If the device of Figure 5 is considered to be the device usedin the preceding discussion, it becomes apparent that on theaverage more voltage variation is due to current fluctuationthan is due to temperature variation. Therefore, to obtain atruly stable reference source, the device must be driven froma constant current source.

http://onsemi.com34

BASIC VOLTAGE REGULATION USING ZENER DIODES

BASIC CONCEPTS OF REGULATION

The purpose of any regulator circuit is to minimize outputvariations with respect to variations in input, temperature,and load requirements. The most obvious use of a regulatoris in the design of a power supply, but any circuit thatincorporates regulatory technique to give a controlledoutput or function can be considered as a regulator. Ingeneral, to provide a regulated output voltage, electroniccircuitry will be used to pass an output voltage that issignificantly lower than the input voltage and block allvoltage in excess of the desired output. Allocations shouldalso be made in the regulation circuitry to maintain thisoutput voltage for variation in load current demand.

There are some basic rules of thumb for the electricalrequirements of the electronic circuitry in order for it toprovide regulation. Number one, the output impedanceshould be kept as low as possible. Number two, a controllingreference needs to be established that is relativelyinsensitive to the prevailing variables. In order to illustratethe importance of these rules, an analysis of some simpleregulator circuits will point out the validity of thestatements. The circuit of Figure 1 can be considered aregulator. This circuit will serve to illustrate the importanceof a low output impedance.

The resistors RS and RR can be considered as the sourceand regulator impedances, respectively.

The output of the circuit is:

�RS

RRRL

RR � RL�

(1)

VO = VI xRRRL

RR + RLRS +

RRRL

RR + RL=

VI

RS

RL

RS

RR+ + 1

Figure 1. Shunt Resistance Regulator

+

−

RS

RR RL

+

−

VI VO

For a given incremental change in VI, the changes in VOwill be:

(2)�RRRRL

RR ��ΔVO = ΔVI

1

RS

RL

RS

RR+ + 1

Assuming RL fixed at some constant value, it is obviousfrom equation (2) that in order to minimize changes in VOfor variations in VI, the shunt resistor RR should be made assmall as possible with respect to the source resistor RS.Obviously, the better this relation becomes, the larger VI is

going to have to be for the same VO, and not until the ratioof RS to RR reaches infinity will the output be held entirelyconstant for variation in VI. This, of course, is animpossibility, but it does stress the fact that the regulationimproves as the output impedance becomes lower and lower.Where the output impedance of Figure 1 is given by

(3)RO =RSRR

RS + RR

It is apparent from this relation that as regulation isimproving with RS increasing and RR decreasing the outputimpedance RO is decreasing, and is approximately equal toRR as the ratio is 10 times or greater. The regulation of thiscircuit can be greatly improved by inserting a referencesource of voltage in series with RR such as Figure 2.

Figure 2. Regulator with Battery Reference Source

+

−

RS

VR

RL

RR

+

−

VI VO

The resistance RR represents the internal impedance of thebattery. For this circuit, the output is

(4)VO = VR + VIV

RS

RL

RS

RR+ + 1

Then for incremental changes in the input VI, the changesin VO will be dependent on the second term of equation(4), which again makes the regulation dependent on theratio of RS to RR. Where changes in the output voltageor the regulation of the circuit in Figure 1 were directlyand solely dependent upon the input voltage and outputimpedance, the regulation of circuit 2 will have an outputthat varies about the reference source VR in accordancewith the magnitude of battery resistance RR and itsfluctuations for changes in VI. Theoretically, if a perfectbattery were used, that is, VR is constant and RR is zero,the circuit would be a perfect regulator. In other words,in line with the basic rules of thumb the circuit exhibitsoptimum regulation with an output impedance of zero, anda constant reference source.

For regulator application, a zener diode can be usedinstead of a battery with a number of advantages. A battery’sresistance and nominal voltage will change with age andload demand; the ON Semiconductor zener diodecharacteristics remain unchanged when operating within its

http://onsemi.com35

specified limits. Any voltage value from a couple of volts tohundreds of volts is available with zener diodes, whereconventional batteries are limited in the nominal valuesavailable. Also, the zener presents a definite size advantage,and is less expensive than a battery because it is permanentand need not be regularly replaced. The basic zener diodeshunt regulator circuit is shown in Figure 3.

Figure 3. Basic Zener Diode Shunt Regulator

+

−

RS

RL

RZ

VZ

ZENER

DIODE

+

−

VI VO

Depending upon the operating conditions of the device, azener diode will exhibit some relatively low zenerimpedance RZ and have a specified breakover voltage of VZthat is essentially constant. These inherent characteristicsmake the zener diode suited for voltage regulatorapplications.

DESIGNING THE ZENER SHUNT REGULATOR

For any given application of a zener diode shunt regulator,it will be required to know the input voltage variations andoutput load requirements. The calculation of componentvalues will be directly dependent upon the circuitrequirements. The input may be constant or have maximumand minimum values depending upon the natural regulationor waveform of the supply source. The output voltage willbe determined by the designer’s choice of VZ and the circuitrequirements. The actual value of VZ will be dependentupon the manufacturer’s tolerance and some small variationfor different zener currents and operating temperatures.

For all practical purposes, the value of VZ as specified onthe manufacturer’s data sheet can be used to approximateVO in computing component values. The requirement forload current will be known and will vary within some givenrange of IL(min) to IL(max).

The design objective of Figure 3 is to determine the propervalues of the series resistance, RS, and zener powerdissipation, PZ. A general solution for these values can bedeveloped as follows, when the following conditions areknown:

VI (input voltage) from VI(min) to VI(max)

VO (output voltage) from VZ(min) to VZ(max)

IL (load current) from IL(min) to IL(max)

The value of RS must be of such a value so that the zenercurrent will not drop below a minimum value of IZ(min).This minimum zener current is mandatory to keep the

device in the breakover region in order to maintain thezener voltage reference. The minimum current can be eitherchosen at some point beyond the knee or found on themanufacturer’s data sheet (IZK). The basic voltage loopequation for this circuit is:

(5)VI = (IZ + IL)RS + VZ

The minimum zener current will occur when VI isminimum, VZ is maximum, and IL is maximum, thensolving for RS, we have:

(6)VI(min) − VZ(max)

IZ(min) + IL(max)RS =

Having found RS, we can determine the maximum powerdissipation PZ for the zener diode.

PZ(max) = VZ(max)VI(max) � VZ(min)RS

� IL(min)

(7)

(8)

(9)

PZ(max) = IZ(max) VZ(max)

Where:

IZ(max) =VI(max) − VZ(min)

RS− IL(min)

Therefore:

VI(max) − VZ(min)

RS− IL(min)

Once the basic regulator components values have beendetermined, adequate considerations will have to be given tothe variation in VO. The changes in VO are a function of fourdifferent factors; namely, changes in VI, IL, temperature, andthe value of zener impedance, RZ. These changes in VO canbe expressed as:

(10)RSRZ

RS + RZ

ΔVI

RS

RZ

RS

RL+

1 +ΔVO = − ΔIL + TCΔTVZ

The value of ΔVO as calculated with equation (10) willquite probably be slightly different from the actual valuewhen measured empirically. For all practical purposesthough, this difference will be insignificant for regulatordesigns utilizing the conventional commercial line of zenerdiodes.

Obviously to precisely predict ΔVO with a given zenerdiode, exact information would be needed about the zenerimpedance and temperature coefficient throughout thevariation of zener current. The “worst case” change can onlybe approximated by using maximum zener impedance andwith typical temperature coefficient.

The basic zener shunt regulator can be modified tominimize the effects of each term in the regulation equation(10). Taking one term at a time, it is apparent that theregulation or changes in output ΔVO will be improved if themagnitude of ΔVI is reduced. A practical and widely usedtechnique to reduce input variation is to cascade zener shuntregulators such as shown in Figure 4.

http://onsemi.com36

Figure 4. Cascaded Zener Shunt Regulators ReduceΔVO by Reducing ΔVI to the Succeeding Stages

−

+RS2

Z1 RL

RS1

−

Z2RL′

+

VI VOVO′ = VI′ = VZ1

This, in essence, is a regulator driven with a pre-regulatorso that the over all regulation is the product of both. Theregulation or changes in output voltage is determined by:

Where:

RL′ = RS2 +RLRZ2

RL + RZ2and IL′ = IL + IZ2

(11)

RS2RZ2

RS2 + RZ2

ΔVZI

RS2

RL+1 +

ΔVO = −

ΔIL + TC2 ΔTVZ2

RS2

RZ2

(12)

RS1RZ1

RS1 + RZ1

ΔVI

RS1

RL′ +1 +

ΔVZ1 = ΔVO′ = −

ΔIL′ + TC1 ΔTVZ1

RS1

RZ1

The changes in output with respect to changes in input forboth stages assuming the temperature and load are constantis

(13)ΔVOΔVZ1

ΔVOΔVO′= = Regulation of second stage

(14)

(15)

= Regulation of first stage

ΔVO′ΔVI

= Combined regulationxΔVOΔVO′=

ΔVOΔVI

ΔVO′ΔVI

Obviously, this technique will vastly improve overallregulation where the input fluctuates over a relatively widerange. As an example, let’s say the input varies by ±20% andthe regulation of each individual stage reduces the variationby a factor of 1/20. This then gives an overall outputvariation of ±20% × (1/20)2 or ±0.05%.

The next two factors in equation (10) affecting regulationare changes in load current and temperature excursions. Inorder to minimize changes for load current variation, theoutput impedance RZRS/(RZ + RS) will have to be reduced.This can only be done by having a lower zener impedancebecause the value of RS is fixed by circuit requirements.There are basically two ways that a lower zener impedancecan be achieved. One, a higher wattage device can be usedwhich allows for an increase in zener current of which willreduce the impedance. The other technique is to series lowervoltage devices to obtain the desired equivalent voltage, sothat the sum of the impedance is less than that for a single

high voltage device. So to speak, this technique will kill twobirds with one stone, as it can also be used to minimizetemperature induced variations of the regulator.

In most regulator applications, the single most detrimentalfactor affecting regulation is that of variation in junctiontemperature. The junction temperature is a function of boththe ambient temperature and that of self heating. In order toillustrate how the overall temperature coefficient isimproved with series lower voltage zener, a mathematicalrelationship can be developed. Consider the diagram ofFigure 5.

Figure 5. Series Zener Improve Dynamic Impedanceand Temperature Coefficient

+

−

RS

RL

Z1

Z2

Zn

+

−

VI VO

With the temperature coefficient TC defined as the % changeper °C, the change in output for a given temperature rangewill equal some overall TC x ΔT x Total VZ. Such as

(16)ΔVO(ΔT) = TC ΔT (VZ1 + VZ2 + . . . + VZN)

Obviously, the change in output will also be equal to thesum of the changes as attributed from each zener.

(17)ΔVO(ΔT) = ΔT(TC1VZ1 + TC2VZ2 + . . . + TCNVZN)

Setting the two equations equal to each other and solvingfor the overall TC, we get

(18)

(19)

TCΔT(VZ1 + VZ2 + . . . + VZN) = ΔT(TC1VZ1+ TC2VZ2 + . . . + TCNVZN)

TC =TC1 VZ1 + TC2VZ2 + . . . + TCNVZN

VZ1 + VZ2 + . . . + VZN

For equation (19) the overall temperature coefficient forany combination of series zeners can be calculated. Say forinstance several identical zeners in series replace a singlehigher voltage zener. The new overall temperaturecoefficient will now be that of one of the low voltagedevices. This allows the designer to go to the manufacturer’s

http://onsemi.com37

data sheet and select a combination of low TC zener diodesin place of the single higher TC devices. Generally speaking,the technique of using multiple devices will also yield alower dynamic impedance. Advantages of this technique arebest demonstrated by example. Consider a 5 watt diode witha nominal zener voltage of 10 volts exhibits approximately0.055% change in voltage per degree centrigrade, a 20 voltunit approximately 0.075%/°C, and a 100 volt unitapproximately 0.1%/°C. In the case of the 100 volt diode,five 20 volt diodes could be connected together to providethe correct voltage reference, but the overall temperaturecoefficient would remain that of the low voltage units, i.e.0.075%/°C. It should also be noted that the same seriescombination improves the overall zener impedance inaddition to the temperature coefficient. A 20 volt, 5 wattON Semiconductor zener diode has a maximum zenerimpedance of 3 ohms, compared to the 90 ohms impedancewhich is maximum for a 100 volt unit. Although theseimpedances are measured at different current levels, theseries impedance of five 20 volt zener diodes is still muchlower than that of a single 100 volt zener diode at the testcurrent specified on the data sheet.

For the ultimate in zener shunt regulator performance, theaforementioned techniques can be combined with the properselection of devices to yield an overall improvement inregulation. For instance, a multiple string of low voltagezener diodes can be used as a preregulator, with a seriescombination of zero TC reference diodes in the final stagesuch as Figure 6.

The first stage will reduce the large variation in VI to somerelatively low level, i.e. ΔVZ. This ΔVZ is optimized byutilizing a series combination of zeners to reduce the overallTC and ΔVZ. Because of this small fluctuation of input to thesecond stage, and if RL is constant, the biasing current of theTC units can be maintained at their specified level. This willgive an output that is very precise and not significantlyaffected by changes in input voltage or junction temperature.

Figure 6. Series Zeners Cascaded With SeriesReference Diodes for Improved Zener

Shunt Regulation

RS1 RS2

Z1

Z2

Z3

Z4

RL

TC1

TC2

+

−

+

−

VI VO

The basic zener shunt regulator exhibits some inherentlimitations to the designer. First of all, the zener is limited toits particular power dissipating rating which may be lessthan the required amount for a particular situation. The totalmagnitude of dissipation can be increased to some degree byutilizing series or parallel units. Zeners in series present few

problems because individual voltages are additive and thedevices all carry the same current and the extent that thistechnique can be used is only restricted by the feasibility ofcircuit parameters and cost. On the other hand, caution mustbe taken when attempting to parallel zener diodes. If thedevices are not closely matched so that they all break overat the same voltage, the low voltage device will go intoconduction first and ultimately carry all the current. In orderto avoid this situation, the diodes should be matched forequal current sharing.

EXTENDING POWER AND CURRENT RANGE

The most common practice for extending the powerhandling capabilities of a regulator is to incorporatetransistors in the design. This technique is discussed in detailin the following sections of this chapter. The seconddisadvantage to the basic zener shunt regulator is thatbecause the device does not have a gain function, a feedbacksystem is not possible with just the zener resistorcombination. For very precise regulators, the design willnormally be an electronic circuit consisting of transistordevices for control, probably a closed loop feedback systemwith a zener device as the basic referencing element.

The concept of regulation can be further extended andimproved with the addition of transistors as the powerabsorbing elements to the zener diodes establishing areference. There are three basic techniques used thatcombine zener diodes and transistors for voltage regulation.The shunt transistor type shown in Figure 7 will extend thepower handling capabilities of the basic shunt regulator, andexhibit marked improvement in regulation.

Figure 7. Basic Transistor Shunt Regulator

RL

RS

ZIZIC

RB

IB

VBE

Q1IL

−

+

−

+

VI VO

In this configuration the source resistance must be largeenough to absorb the overvoltage in the same manner as inthe conventional zener shunt regulator. Most of the shuntregulating current in this circuit will pass through thetransistor reducing the current requirements of the zenerdiode by essentially the dc current gain of the transistor hFE.Where the total regulating shunt current is:

(20)

IS = IZ + IC = IZ + IB hFE

where

therefore

IZ = IB + IRB and IB >> IRB

IS ≈ IZ + IZ hFE = IZ (1 + hFE)

http://onsemi.com38

The output voltage is the reference voltage VZ plus theforward junction drop from base to emitter VBE of thetransistor.

(21)VO = VZ + VBE

The values of components and their operating conditionis dictated by the specific input and output requirements andthe characteristics of the designer’s chosen devices, asshown in the following relations:

(22)RS =VI(min) − VO(max)

IZ(min) [1+hFE(min)] + IL(max)

RB =VI(min) − VZ(max)

IZ(min)

PDZ = IZ(max) VZ(max)

when

IZ(max) =VI(max) � V O(min)

RS� IL(min) � 1

1 � hFE(min)�

hence

VI(max) − VO(min)

RS− IL(min)

11 + hFE(min)

PDZ =VI(max) � VO(min)

RS� IL(min) � 1

1 � hFE(min)�VI(max) − VO(min)

RS− IL(min)

VZ(max)

1 + hFE(min)

PDQ =VI(max) � V O(min)

RS� IL(min) �1 888888 �VI(max) − VO(min)

RS− IL(min) VO(max)

(23)

(24)

(25)

(26)

(27)

Regulation with this circuit is derived in essentially thesame manner as in the shunt zener circuit, where the outputimpedance is low and the output voltage is a function of thereference voltage. The regulation is improved with thisconfiguration because the small signal output impedance isreduced by the gain of Q1 by 1/hFE.

One other highly desirable feature of this type of regulatoris that the output is somewhat self compensating fortemperature changes by the opposing changes in VZ andVBE for VZ ≈ 10 volts. With the zener having a positive2 mV/°C TC and the transistor base to emitter being anegative 2 mV/°C TC, therefore, a change in one is cancelledby the change in the other. Even though this circuit is a veryeffective regulator it is somewhat undesirable from anefficiency standpoint. Because the magnitude of RS isrequired to be large, and it must carry the entire inputcurrent, a large percentage of power is lost from input tooutput.

EMITTER FOLLOWER REGULATOR

Another basic technique of transistor-zener regulation isthat of the emitter follower type shown in Figure 8.

Figure 8. Emitter Follower Regulator

RS

RB

Z1

IRB

IZ

+

−VBE

−

+

IL

RL

Q1IC

+

−

+

−

VI VO

This circuit has the desirable feature of using a seriestransistor to absorb overvoltages instead of a large fixedresistor, thereby giving a significant improvement inefficiency over the shunt type regulator. The transistor mustbe capable of carrying the entire load current andwithstanding voltages equal to the input voltage minus theload voltage. This, of course, imposes a much more stringentpower handling requirement upon the transistor than wasrequired in the shunt regulator. The output voltage is afunction of the zener reference voltage and the base toemitter drop of Q1 as expressed by the equation (28).

(28)VO = VZ − VBE

The load current is approximately equal to the transistorcollector current, such as shown in equation (29).

(29)IL(max) ≈ IC(max)

The designer must select a transistor that will meet thefollowing basic requirements:

(30)

PD ≅ (VI(max) − VO)IL(max)

IC(max) ≈ IL(max)

BVCES ≥ (VI(max) − VO)

Depending upon the designer’s choice of a transistor andthe imposed circuit requirements, the operation conditionsof the circuit are expressed by the following equations:

(31)

VZ = VO + VBE

= VO + IL(max)/gFE(min) @ IL(max)

RS =VI(min) − VZ − VCE(min) @ IL(max)

IL(max)

Where VCE(min) is an arbitrary value of minimumcollector to emitter voltage and gFE is the transconductance.

This is sufficient to keep the transistor out of saturation,which is usually about 2 volts.

http://onsemi.com39

(32)RB =

VCE(min) @ IL(max)

IL(max)/hFE(min) @ IL(max) + IZ(min)

IZ(max) =VI(max) −VZ

RB + RZ

PDZ = IZ(max)VZ

Actual PDQ = (VI(max) − VO) IL(max)

(33)

(34)

(35)

There are two primary factors that effect the regulationmost in a circuit of this type. First of all, the zener currentmay vary over a considerable range as the input changesfrom minimum to maximum and this, of course, may havea significant effect on the value of VZ and therefore VO.Secondly, VZ and VBE will both be effected by temperaturechanges which are additive on their effect of output voltage.This can be seen by altering equation (28) to show changesin VO as dependent on temperature, see equation (36).

(36)VO(ΔT) = ΔT[(+TC) VZ − (−TC) VBE]

The effects of these detrimental factors can be minimizedby replacing the bleeder resistor RB with a constant currentsource and the zener with a reference diode in series with aforward biased diode (see Figure 9).

Figure 9. Improved Emitter Follower Regulator

+

−

+

−RS

RL

Q1

FORWARDBIAS DIODE

TC ZENER

IB = K

CONSTANT

CURRENT

SOURCE

VI VO

The constant current source can be either a current limiterdiode or a transistor source. The current limiter diode isideally suited for applications of this type, because it willsupply the same biasing current irregardless of collector tobase voltage swing as long as it is within the voltage limitsof the device. This technique will overcome changes in VZfor changes in IZ and temperature, but changes in VBE dueto load current changes are still directly reflected upon theoutput. This can be reduced somewhat by combining a

transistor with the zener for the shunt control element asillustrated in Figure 10.

Figure 10. Series Pass Regulator

+

−

RL

Z1

Q2

Q1

RB

RS

IC2

CONSTANT

CURRENT

SOURCE

VI VO

This is the third basic technique used for transistor-zenerregulators. This technique or at least a variation of it, findsthe widest use in practical applications. In this circuit thetransistor Q1 is still the series control device operating as anemitter follower. The output voltage is now established bythe transistor Q2 base to emitter voltage and the zenervoltage. Because the zener is only supplying base drive toQ2, and it derives its bias from the output, the zener currentremains essential constant, which minimizes changes in VZdue to IZ excursions. Also, it may be possible (VZ ≈ 10 V)to match the zener to the base-emitter junction of Q2 for anoutput that is insensitive to temperature changes. Theconstant current source looks like a very high loadimpedance to the collector of Q2 thus assuming a very highvoltage gain. There are three primary advantages gainedwith this configuration over the basic emitter follower:

1. The increased voltage gain of the circuit with theaddition of Q2 will improve regulation for changes inboth load and input.

2. The zener current excursions are reduced, therebyimproving regulation.

3. For certain voltages the configuration allows goodtemperature compensation by matching thetemperature characteristics of the zener to thebase-emitter junction of Q2.

The series pass regulator is superior to the other transistorregulators thus far discussed. It has good efficiency, betterstability and regulation, and is simple enough to beeconomically practical for a large percentage ofapplications.

http://onsemi.com40

Figure 11. Block Diagram of Regulator with Feedback

REGULATING

POWER

ELEMENT

CONTROL

UNITAMPLIFIER

REFERENCE

AND

ERROR

DETECTION

LOADINPUT OUTPUT

EMPLOYING FEEDBACK FOROPTIMUM REGULATION

The regulators discussed thus far do not employ anyfeedback techniques for precise control and compensationand, therefore, find limited use where an ultra preciseregulator is required. In the more sophisticated regulatorssome form of error detection is incorporated and amplifiedthrough a feedback network to closely control the powerelements as illustrated in the block diagram of Figure 11.

Regulating circuits of this type will vary in complexityand configuration from application to application. Thistechnique can best be illustrated with a couple of actualcircuits of this type. The feedback regulators will generallybe some form of series pass regulator, for optimumperformance and efficiency. A practical circuit of this typethat is extensively utilized is shown in Figure 12.

In this circuit, the zener establishes a reference level forthe differential amplifier composed of Q4 and Q5 which willset the base drive for the control transistor Q3 to regulate theseries high gain transistor combination of Q1 and Q2. Thedifferential amplifier samples the output at the voltagedividing network of R8, R9, and R10. This is compared to thereference voltage provided by the zener Z1. The difference,if any, is amplified and fed back to the control elements. Byadjusting the potentiometer, R9, the output level can be setto any desired value within the range of the supply. (Theoutput voltage is set by the relation VO = VZ[(RX +RY)/RX].) By matching the transistor Q4 and Q5 forvariations in VBE and gain with temperature changes andincorporating a temperature compensated diode as thereference, the circuit will be ultra stable to temperatureeffects. The regulation and stability of this circuit is verygood, and for this reason is used in a large percentage ofcommercial power supplies.

Figure 12. Series Pass Regulator with Error Detection and FeedbackAmplification Derived from a Differential Amplifier

R1

R4

Q2Q4

Q3 C1

Q1

R2

R3

Q5

RL

Z1

R5 R6

R7

R9

R10

+

− −

R8RX

RY

+

VI VO

http://onsemi.com41

Figure 13. Series Pass Regulator with Temperature Compensated Reference Amplifier

+

−

+

−

R1

R2RL

R3

R5

R4

R8

R6

R7

D1

Q1

Q2

Q3

REFERENCE

AMPLIFIER

VI VO

Another variation of the feedback series pass regulator isshown in Figure 13. This circuit incorporates a stabletemperature compensated reference amplifier as the primarycontrol element.

This circuit also employs error detection and amplifiedfeedback compensation. It is an improved version over thebasic series pass regulator shown in Figure 10. The serieselement is composed of a Darlington high gainconfiguration formed by Q1 and Q2 for an improvedregulation factor. The combined gain of the referenceamplifier and Q3 is incorporated to control the series unit.This reduced the required collector current change of thereference amplifier to control the regulator so that the biascurrent remains close to the specified current for lowtemperature coefficient. Also the germanium diode D1 willcompensate for the base to emitter change in Q3 and keep thereference amplifier collector biasing current fairly constantwith temperature changes. Proper biasing of the zener andtransistor in the reference amplifier must be adhered to if theoutput voltage changes are to be minimized.

CONSTANT CURRENT SOURCES FORREGULATOR APPLICATIONS

Several places throughout this chapter emphasize theneed for maintaining a constant current level in the variousbiasing circuits for optimum regulation. As was mentionedpreviously in the discussion on the basic series passregulator, the current limiter diode can be effectively usedfor the purpose.

Aside from the current limiter diode a transistorizedsource can be used. A widely used technique is shownincorporated in a basic series pass regulator in Figure 14.

The circuit is used as a preregulated current source tosupply the biasing current to the transistor Q2. The constantcurrent circuit is seldom used alone, but does find wide usein conjunction with voltage regulators to supply biasingcurrent to transistors or reference diodes for stableoperation. The Zener Z2 establishes a fixed voltage acrossRE and the base to emitter of Q3. This gives an emittercurrent of IE = (VZ − VBE)/RE which will vary only slightlyfor changes in input voltage and temperature.

Figure 14. Constant Current Source Incorporated in a Basic Regulator Circuit

Q1

Q2

Q3

RB1

Z1

RL

REZ2

RB2

CONSTANT

CURRENT

SOURCE

−−

++

VI VO

http://onsemi.com42

IMPEDANCE CANCELLATION

One of the most common applications of zener diodes isin the general category of reference voltage supplies. Thefunction of the zener diode in such applications is to providea stable reference voltage during input voltage variations.This function is complicated by the zener diode impedance,which effectively causes an incremental change in zenerbreakdown voltage with changing zener current.

Figure 15. Impedance Cancellation with AnUncompensated Zener

R2R1

Z1 TC1

+

−

VI

It is possible, however, by employing a bridge type circuitwhich includes the zener diode and current regulatingresistance in its branch legs, to effectively cancel the effectof the zener impedance. Consider the circuit of Figure 15 asan example. This is the common configuration for a zenerdiode voltage regulating system. The zener impedance at 20mA of a 1N4740 diode is typically 2 ohms. If the supplyvoltage now changes from 30 V to 40 V, the diode currentdetermined by R1 changes from 20 to 30 mA; the averagezener impedance becomes 1.9 ohms; and the referencevoltage shifts by 19 mV. This represents a reference changeof .19%, an amount far too large for an input change of 30%in most reference supplies.

The effect of zener impedance change with current isrelatively small for most input changes and will be neglectedfor this analysis. Assuming constant zener impedance, thezener voltage is approximated by

(37)V′Z = VZ + Z(I′Z − IZ)

where V′Z is the new zener voltageVZ is the former zener voltageI′Z is the new zener currentIZ is the new zener current flowing at VZRZ is the zener impedance

Then

Let the input voltage VI in Figure 15 increase by anamount ΔVI

Then ΔI =ΔVI − ΔVZ

R1

ΔVZ

RZAlso ΔI =

Solving ΔVIRZ − ΔVZRZ − ΔVZR1 = 0

ΔVZ

ΔVIOr =

RZ

R1 + RZ

(39)

(38)

(40)

ΔVZ = ZΔIZ

Equation 40 merely states that the change in referencevoltage with input tends to zero when the zener impedancetends also to zero, as expected.

The figure of merit equation can be applied to the circuitsof Figure 16 and 17 to explain impedance cancellation. TheChange Factor equations for each leg and the referencevoltage VR are:

(41)

(43)

(42)

CFVZ =ΔVZ

ΔVI

RZ

R1 + RZ= = RA

R3

R2 + R3

ΔV2

ΔVICFV2 = = = RB

RZ

R1 + RZ

ΔVR

ΔVICFVR = = = RA − RB

R3

R2 + R3=

Figure 16. Standard VoltageRegulation Circuit

R1

VI′ ΔVI

I

VZ

Figure 17. Impedance Cancellation Bridge

VI′ ΔVI

R3

R1R2

VR

V2VZ

http://onsemi.com43

Since the design is to minimize CFVR, RB can be set equalto RA. The Input Regulation Factors are:

(44)

(46)

(45)

γVZ =ΔVZ

ΔVI� VI

ViZ� =

1VI

VZ1 +

VZ

VI� VI

ViZ�R1

RZ

γV2 =ΔV2

ΔVI� VI

ViZ� = 1

VI

V2

γVR =ΔVR

ΔVI� VI

ViZ� =

1VI

VR1 + � VI

ViZ�R1

RZ� VI

ViZ�VZ

VI� VI

Vi Z�1

1 −RB

RA

It is seen that γVR can be minimized by setting RB = RA.Note that it is not necessary to match R3 to RZ and R2 to

R1. Thus R3 and R2 can be large and hence dissipate lowpower. This discussion is assuming very light load currents.

http://onsemi.com44

ZENER PROTECTIVE CIRCUITS AND TECHNIQUES:BASIC DESIGN CONSIDERATIONS

INTRODUCTION

The reliability of any system is a function of the ability ofthe equipment to operate satisfactorily during moderatechanges of environment, and to protect itself duringotherwise damaging catastrophic changes. The silicon zenerdiode offers a convenient, simple but effective means ofachieving this result. Its precise voltage sensitivebreakdown characteristic provides an accurate limitingelement in the protective circuit. The extremely highswitching speed possible with the zener phenomenon allowsthe circuit to react faster by orders of magnitude thatcomparable mechanical and magnetic systems.

By shunting a component, circuit, or system with a zenerdiode, the applied voltage cannot exceed that of theparticular device’s breakdown voltage. (See Figure 1.)

A device should be chosen so that its zener voltage issomewhat higher than the nominal operating voltage butlower than the value of voltage that would be damaging ifallowed to pass. In order to adequately incorporate the zenerdiode for circuit protection, the designer must considerseveral factors in addition to the required zener voltage. Thefirst thing the designer should know is just what transientcharacteristics can be anticipated, such as magnitude,duration, and the rate of reoccurrence. For short durationtransients, it is usually possible to suppress the voltage spikeand allow the zener to shunt the transient current away fromthe load without a circuit shutdown. On the other hand, if theover-voltage condition is for a long duration, the protectivecircuit may need to be complimented with a disconnectelement to protect the zener from damage created byexcessive heating. In all cases, the end circuit will have to bedesigned around the junction temperature limits of thedevice.

The following sections illustrate the most common zenerprotective circuits, and will demonstrate the criteria to befollowed for an adequate design.

BASIC PROTECTIVE CIRCUITSFOR SUPPLY TRANSIENTS

The simple zener shunt protection circuit shown inFigure 1 is widely used for supply voltage transientprotection where the duration is relatively short. The circuitapplies whether the load is an individual component or acomplete circuit requiring protection. Whenever the inputexceeds the zener voltage, the device avalanches intoconduction clamping the load voltage to VZ. The totalcurrent the diode must carry is determined by the magnitude

of the input voltage transient and the total circuit impedanceminus the load current. The worst case occurs when loadcurrent is zero and may be expressed as follows:

IZ(max) =VI(max) − VZ

RS(1)

Figure 1. Basic Shunt Zener TransientProtection Circuit

LOADZPOWER

SUPPLY

RS+

−

The maximum power dissipated by the zener is

PZ(max) = IZ(max) VZ(max)=VI(max) − VZ

RS(2)VZ(max)

Also, more than one device can be used, i.e., a seriesstring, which will reduce the percentage of total power to bedissipated per device by a factor equal to the number ofdevices in series. The number of diodes required can befound from the following expression:

Number =PZ(max)

PZ (allowable per device)(3)

Any fraction of a zener must be taken as the next highestwhole number. This design discussion has been based uponthe assumption that the transient is of a single shot,non-recurrent type. For all practical purposes it can beconsidered non-recurrent if the “off period” betweentransients is at least four times the thermal time constant ofthe device. If the “off period” is shorter than this, then thedesign calculations must include a factor for the duty cycle.This is discussed in detail in Chapter 4. In Chapter 4 there arealso some typical curves relating peak power, pulse durationand duty cycle that may be appropriate for some designs.

Obviously, the factor that limits the feasibility of the basiczener shunt protective circuit is the pulse durations “t”. Asthe duration increases, the allowable peak power for a givenconfiguration decreases and will approach a steady statecondition.

http://onsemi.com45

Figure 2. Overvoltage Protection with Zener Diodes and Fuses

LOAD

R CVS

RS FUSE

POWERSUPPLY

+

−

When the anticipated transients expected to prevail for aspecific situation are of long duration, a basic zener shuntbecomes impractical, in such a case the circuit can beimproved by using a complementary disconnect element.The most common overload protective element is without adoubt the standard fuse. The common fuse adequatelyprotects circuit components from over-voltage surges, but atthe same time must be chosen to eliminate “nuisance fusing”which results when the maximum current rating of the fuseis too close to the normal operational current of the circuit.

AN EXAMPLE PROBLEM: SELECTING AFUSE-ZENER COMBINATION

Consider the case illustrated in Figure 2. Here the loadcomponents are represented by a parallel combination of Rand C, equivalent to many loads found in practice. Themaximum capacitor voltage rating is usually thecircuit-voltage limiting factor due to the cost of high voltagecapacitors. Consequently, a protective circuit must bedesigned to prevent voltage surges greater than 1.5 timesnormal working voltage of the capacitor. It is common,however, for the supply voltage to increase to 135% normalfor long periods. Examination of fuse manufacturers’melting time-current curves shows the difficulty of trying toselect a fuse which will melt rapidly at overload (within oneor two cycles of the supply frequency to prevent capacitordamage), and will not melt when subjected to voltages closeto overload for prolonged periods.

By connecting a zener diode of correct voltage ratingsacross the load as shown, a fuse large enough to withstandnormal current increases for long periods may be chosen.The sudden current increase when zener breakdown occursmelts the fuse rapidly and protects the load from largesurges. In Figure 3, fuse current was plotted against supplyvoltage to illustrate the improvement in load protectionobtained with zener-fuse combinations. Fuse current “A”would be selected to limit current resulting from voltagesurges above 112 V to 90 mA, which would melt the fuse in100 ms. It is a simple matter, however, to select a fuse whichmelts in 30 ms at 200 mA but is unaffected by 100 mAcurrents. The zener connection allows fuse current “B” to beselected, eliminating this design problem and providing afaster, more reliable protective circuit. If the same fuse wasused without the zener diode, a supply voltage of 210 volts

would be reached before the fuse would begin to protect theload.

VS, SUPPLY VOLTAGE (VOLTS)

60 70 80 90 100 110 120 130 14060

80

100

120

140

160

180

200

220

�A"

�B"

FU

SE

CU

RR

EN

T (m

A)

Figure 3. Fuse Current versusSupply Voltage

RESISTIVELOAD ONLY

ZENER DIODE WITH,RESISTIVE LOAD

ZENER BREAKDOWN,

VOLTAGENORMAL LOAD VOLTAGE

Selection of the correct power rating of zener diodes to beused for surge protection depends upon the magnitude andduration of anticipated surges. Often in circuits employingboth fuses and zener diodes, the limiting surge duration willbe the melting time of the fuse. This, in turn, depends on thenature of the load protected and the length of time it willtolerate an overload.

As a first solution to the example problem, consider azener diode with a nominal breakdown voltage of 110 voltsmeasured at a test current (IZT) of 110 mA. Since the fuserequires about 200 mA to melt and 100 mA are drawnthrough the load at this voltage, the load voltage will neverexceed the zener breakdown voltage on slowly rising inputs.Transients producing currents of approximately 200 mA butof shorter duration than 30 ms will simply be clipped byzener action and diverted from the load. Transients of veryhigh voltage will produce larger currents and, hence, willmelt the fuse more rapidly. In the limiting case wheretransient power might eventually destroy the zener diode,the fuse always melts first because of the slower thermaltime constant inherent in the zener diode’s larger geometry.

The curves in Figure 4 illustrate the change in zenervoltage as a function of changing current for a typical devicetype.

http://onsemi.com46

LOG IZ/IZK

CH

AN

GE

, IN

VZ

VZ = V1

V2

V3

Figure 4. Change in VZ for Changes in IZ

If an actual curve for the device being used is notavailable, the zener voltage at a specific current above orbelow the test current may be approximated by equation 4.

Where: V = VZ + ZZT (I−IZT)VZ = zener voltage at test current IZTZZT = zener impedance at test current IZTIZT = test currentV = zener voltage at current I

(4)

For a given design, the maximum zener voltage to expectfor the higher zener current should be determined to makesure the limits of the circuit are met. If the maximum limitis excessive for the original device selection, the next lowervoltage rating should be used.

The previous discussion on design consideration forprotective circuits incorporating fuses is applicable to anyprotective element that permanently disconnects the supplywhen actuated. Rather than a fuse, a non-resetting magneticcircuit breaker could have been used, and the samereasoning would have applied.

LOAD CURRENT SURGES

In many actual problems the designer must choose aprotective circuit to perform still another task. Not only mustthe equipment be protected from the voltage surges in thesupply, but the supply itself often requires protection fromshorts or partial shorts in the load. A direct short in the loadis fairly easy to handle, as the drastic current change permitsthe use of fuses with ratings high enough to avoid problemswith supply surges. More common is the partial short, as

illustrated in Figure 5. If a short circuit occurs in thecapacitive section of the load (represented by C) theresulting fault current is limited by the resistive section(represented by R) to a value which may not be great enoughto melt the fuse. The fault current could be sufficient,however, to damage the supply and other components in theload.

The problem is resolved by employing a zener diode toprotect against supply surges as described in the previoussection, and by selecting a separate fuse to protect from loadfaults. The load fuse in Figure 5 is chosen close to the normaloperating current. Abnormal supply surges do not affect itand equipment operates reliably but with ample protectionfor the supply against load changes.

ZENER DIODES AND RECLOSINGDISCONNECT ELEMENTS

An interesting application of zener diodes as overvoltageprotectors, which offers the possibility of designing for bothlong and short duration surges, is shown in Figure 6.

Figure 6. Zener Diode Reclosing Circuit BreakerProtective Circuit

R

IZ

LOAD

POWER

SUPPLY

RECLOSING

CIRCUIT

BREAKER

In the event of a voltage overload exceeding a chosenzener voltage, a large current will be drawn through thediode. The reclosing disconnect element opens after aninterval determined by its time constant, and the supply isdisconnected. After another interval, again depending on theswitch characteristics, the supply is reconnected and thevoltage “sampled” by the zener diode. This leads to an“on-off” action which continues until the supply voltagedrops below the predetermined limit. At no time can the loadvoltage or current exceed that set by the zener. The chiefadvantage in this type of circuit is the elimination of fusereplacement in similar fusing circuits, while providingessentially the same load protection.

Figure 5. Supply and Load with Zener Diode; Fuse Circuitry

POWERSUPPLY

RSUPPLYFUSE

LOADFUSE R

C

LOAD

http://onsemi.com47

Figure 7. (Typical) Voltage, Current and Temperature Waveformsfor a Thermal Breaker

TIME

TIME

TIME

TIME

VOLTS

SUPPLY

VOLTAGE

°C

THERMAL

BREAKER

TEMPERATURE

AMPS

ZENER

DIODE

CURRENT

ZENER

DIODE

JUNCTION

TEMPERATURE

SURGE VOLTAGE

OVER VOLTAGE

NORMAL OPERATING VOLTAGE

BREAK TEMPERATURE

MAKE

TEMPERATURE

MAXIMUM TJ°C

It is difficult to define a set design procedure in this case,because of the wide variety of reclosing, magnetic andthermal circuit breakers available. Care should be taken toensure that the power dissipated in the zener diode during theconduction time of the disconnect element does not exceedits rating. As an example, assume the disconnect elementwas a thermal breaker switch. The waveforms for a typicalover-voltage situation are shown in Figure 7.

It is apparent that the highest zener diode junctiontemperature is reached during the first conduction period. Atthis time the thermal breaker is cold and requires the greatesttime to reach its break temperature. The breaker then cyclesthermally between the make and break temperatures as longas the supply voltage is greater than the zener voltage, asshown in Figure 7.

The zener diode current and junction temperaturevariation are shown in the last two waveforms of Figure 7.Overvoltage durations longer than the trip time of thethermal breaker do not affect the diode as the supply isdisconnected. An overvoltage of much higher level simply

causes the thermal breaker to open sooner. In effect, thezener diode rating must be high enough to ensure thatmaximum junction temperature is not reached during thelongest interval that the thermal switch will be closed.

Manufacturers of thermally operated circuit breakerspublish current-time curves for their devices similar to thatshown in Figure 8. By estimating the peak supplyovervoltage and determining the maximum overvoltagetolerated by the load, an estimation of peak zener current canbe made. The maximum breaker trip time may then be readfrom Figure 8. (After the initial current surge, the durationof “of” time is determined entirely by the breakercharacteristics and will vary widely with manufacture.) Thezener diode junction temperature rise during conductionmay be calculated now from the thermal time constant of thedevice and the heatsink used.

Because the reclosing circuit breaker is continuallycycling on and off, the zener current takes on thecharacteristics of a repetitive surge, as can be seen inFigure 7.

http://onsemi.com48

30

20

10

00 1 2 3

CURRENT (AMPS)

TR

IP T

IME

(S

EC

ON

DS

)

Figure 8. Trip Time versus Current forThermal Breaker

TRANSISTOR OVERVOLTAGE PROTECTION

In many electronic circuits employing transistors, highinternal voltages can be developed and, if applied to thetransistors, will destroy them. This situation is quitecommon in transistor circuits that are switching highlyinductive loads. A prime example of this would be intransistorized electronic ignition systems such as shown inFigures 9a and 9b.

The zener diode is an important component to assure solidstate ignition system reliability. There are two basic methods

of using a zener diode to protect an ignition transistor. Theseare shown in Figures 9a and 9b. In Figure 9b the transistoris protected by a zener diode connected between base andcollector and in Figure 9a, the zener is connected betweenemitter and collector. In both cases the voltage level of thezener must be selected carefully so that the voltage stress onthe transistor is in a region where the safe operating area isadequate for reliable circuit operation.

Figure 10 illustrates “safe” and “unsafe” selection of azener diode for collector-base protection of a transistor in theignition coil circuit. It can be seen that the safe operating areaof a transistor must be known if an adequate protective zeneris to be selected.

The zener diode must be able to take the stress of peakpulse current necessary to clamp the voltage rise across thetransistor to a safe value. In a typical case, a 5 watt, 100 voltzener transient suppressor diode is required to operate withan 80 μs peak pulse current of 8 amperes when connectedbetween the collector-emitter of the transistor. Thewaveform of this pulse current approaches a sine wave inshape (Figure 11). The voltage rise across a typical transientsuppressor diode due to this current pulse is shown in Figure12. This voltage rise of approximately 8 volts indicates aneffective zener impedance of approximately 1 ohm.However, a good share of this voltage rise is due to thetemperature coefficient and thermal time constant of thezener. The temperature rise of the zener diode junction isindicated by the voltage difference between the rise and fallof the current pulse.

Figure 9. Transistor Ignition Systems with ZenerOvervoltage Surge Protection

10 Ω1/2 W

1 Ω

10 Ω10 W

1 Ω100 W

560 pF

2 μF

5 Ω

1N5374B

1N6295

10 Ω

2N5879

+12 V

H.V. TODIST.

PRESTO-LITE

201

MALLORY

COIL

28100

H.V. TO

DIST.

200 VPAPER

2N6031

+12 V

(A) (B)

http://onsemi.com49

COLLECTOR-BASE

ZENER CLAMP

10

0

9

8

7

6

5

4

3

2

1

0

SAFE

UNSAFE

IC

10 30 40 50 60 70 80 90 100 110 120

VCE

COLLECTOR-

EMITTER

ZENER

CLAMP

TYPICAL TRANSISTOR SAFE AREA LIMIT

Figure 10. Safe Zener Protection

20

LOADLINE

SAFE

TIME 10 μs/div

2 A

/div

ZE

NE

R C

UR

RE

NT

Figure 11. Zener Diode Current Pulse

100 VZENER VOLTAGE 1 V/div

1 A

/div

ZE

NE

R C

UR

RE

NT

0

Figure 12. Voltage-Current Representation on100 V Zener

In order to assure safe operation, the change in zenerjunction temperature for the peak pulse conditions must beanalyzed. In making the calculation, the method describedin Chapter 3 should be used, taking into account duty cycle,pulse duration, and pulse magnitude.

When the zener diode is connected between the collectorand emitter of the transistor, additional power dissipationwill result from the clipping of the ringing voltage of theignition coil by the forward conduction of the zener diode.This power dissipation by the forward diode current will

result in additional zener voltage rise. It is not uncommon toobserve a 15-volt rise above the zener device voltage ratingdue to temperature coefficient and impedance under thesepulse current conditions.

The zener diode should be connected as close as possibleto the terminals of the transistor the zener is intended toprotect. This insures that induced voltage transients, causedby current changes in long lead lengths, are clamped by thezener and do not appear across the transistor.

http://onsemi.com50

Figure 13. DC-DC Converter with Surge Protecting Diodes

+

Another example of overvoltage protection of transistoroperating in an inductive load switch capacity is illustratedin Figure 13. The DC-DC converter circuit shows aconnection from collector to emitter of two zener diodes ascollector overvoltage protectors. Without some type oflimiting device, large voltage spikes may appear at thecollectors, due to the switching transients produced withnormal circuit operation. When this spike exceeds thecollector breakdown rating of the transistor, transistor life isconsiderably shortened. The zener diodes shown are chosenwith zener breakdowns slightly below transistor breakdownvoltage to provide the necessary clipping action. Since thespikes are normally of short duration (0.5 to 5 μs) and dutycycle is low, normal chassis mounting provides adequateheatsinking.

METER PROTECTION

The silicon zener diode can be employed to preventoverloading sensitive meter movements used in low rangeDC and AC voltmeters, without adversely affecting themeter linearity. The zener diode has the advantage overthermal protective devices of instantaneous action and, ofcourse, will function repeatedly for an indefinite time (ascompared to the reset time necessary with thermal devices).While zener protection is presently available for voltages aslow as 2.4 volts, forward diode operation can be used formeter protection where the voltage drop is much smaller. Atypical protective circuit is illustrated in Figure 14. Here themeter movement requires 100 μAmps for full scaledeflection and has 940 ohms resistance. For use in a

voltmeter to measure 25 V, approximately 249 thousandohms are required in series.

Figure 14. Meter Protection with Zener Diode

70K

179K

1N4746

(18 VOLT

ZENER DIODE)

+

−

25 V

μA

The protection provided by the addition of an 18 volt zeneris illustrated in Figure 15. With an applied voltage of25 volts, the 100 μAmps current in the circuit produces adrop of 17.9 volts across the series resistance of 179thousand ohms. A further increase in voltage causes thezener diode to conduct, and the overload current is shuntedaway from the meter. Since ON Semiconductor zener diodeshave zener voltages specified within 5 and 10%, a safedesign may always be made with little sacrifice in meterlinearity by assuming the lowest breakdown voltage withinthe tolerance. The shunting effect on the meter of the reversebiased diode is generally negligible below breakdownvoltage (on the order of 0.5° full scale). For very precisework, the zener diode breakdown voltage must be accuratelyknown and the design equations solved for the correctresistance values.

http://onsemi.com51

200

0 50 100 150 200

TOTAL CURRENT (μA)

150

100

50

0

ME

TE

R C

UR

RE

NT

( A

)μ

WITH PROTECTIVE CIRCUIT

Figure 15. Meter Protection withZener Diodes

ZENER DIODES USED WITH SCRS FORCIRCUIT PROTECTION

An interesting aspect of circuit protection incorporatingthe reliable zener diode is the protective circuits shown inFigures 16 and 17.

In a system that is handling large amounts of power, it maybecome impractical to employ standard zener shuntprotection because of the large current it would be requiredto carry. The SCR crowbar technique shown in Figure 16 canbe effectively used in these situations. The zener diode is stillthe transient detection component, but it is only required tocarry the gate current for SCR turn on, and the SCR willcarry the bulk of the shunt current. Whenever the incomingvoltage exceeds the zener voltage, it avalanches, supplyinggate drive to the SCR which, when fired, causes a currentdemand that will trip the circuit breaker. The resistors shownare for current limiting so that the SCR and zener ratings arenot exceeded.

The circuit of Figure 17 is designed to disconnect thesupply in the event a specified load current is exceeded. Thisis done by means of a series sense resistor and a compatiblezener to turn the shunt SCR on. When the voltage across theseries resistor, which is a function of the load current,becomes sufficient to break over the zener, the SCR is fired,causing the circuit breaker to trip.

Figure 16. SCR Crowbar Over-Voltage Protection Circuit for AC Circuit Operation

Figure 17. SCR Longterm Current Overload Protection

R2

R3

R5

R6R1

R4

AC

ZENER

ZENER

SCR SCR

CIRCUITBREAKER

CIRCUITBREAKER

http://onsemi.com52

ZENER TRANSIENT SUPPRESSORS

The transient suppressor is used as a shunt element inexactly the same manner as a conventional zener. It offersthe same advantages such as low insertion loss, immediaterecovery after operation, a clamping factor approachingunity, protection against fast rising transients, and simplecircuitry. The primary difference is that the transientsuppressor extends these advantages to higher power levels.

Even in the event of transients with power contents far inexcess of the capacity of the zeners, protection is stillprovided the load. When overloaded to failure, the zener willapproximate a short. The resulting heavy drain will aid inopening the fuse or circuit breaker protecting the loadagainst excess current. Thus, even if the suppressor isdestroyed, it still protects the load.

The design of the suppressor-fuse combination for therequired level of protection follows the techniques forconventional zeners discussed earlier in this chapter.

TRANSIENT SUPPRESSION CHARACTERISTICS

Zener diodes, being nearly ideal clippers (that is, theyexhibit close to an infinite impedance below the clippinglevel and close to a short circuit above the clipping level), areoften used to suppress transients. In this type of application,it is important to know the power capability of the zener forshort pulse durations, since they are intolerant of excessivestress.

Some ON Semiconductor data sheets such as the ones fordevices shown in Table 1 contain short pulse surgecapability. However, there are many data sheets that do notcontain this data and Figure 18 is presented here tosupplement this information.

Table 1. Transient Suppressor Diodes

SeriesNumbers

Steady StatePower Package Description

1N4728A 1 W DO-41 Double SlugGlass

1N6267A 5 W Case 41A Axial LeadPlastic

1N5333B 5 W Case 102 Surmetic 40

1N746A/957B/4370A

500 mW DO-35 Double SlugGlass

1N5221B 500 mW DO-35 Double SlugGlass

Some data sheets have surge information which differsslightly from the data shown in Figure 18. A variety ofreasons exist for this:

1. The surge data may be presented in terms of actualsurge power instead of nominal power.

2. Product improvements have occurred since the datasheet was published.

3. Large dice are used, or special tests are imposed on theproduct to guarantee higher ratings than those shownin Figure 18.

4. The specifications may be based on a JEDECregistration or part number of another manufacturer.

The data of Figure 18 applies for non-repetitive conditionsand at a lead temperature of 25°C. If the duty cycle increases,the peak power must be reduced as indicated by the curvesof Figure 19. Average power must be derated as the lead orambient temperature rises above 25°C. The average powerderating curve normally given on data sheets may benormalized and used for this purpose.

100

0.01 0.02 0.05 0.1 0.2 0.5 1 2 5 10

PULSE WIDTH (ms)

50

20

10

5

2

1

0.5

0.2

0.1

0.05

0.020.01

P PK

(nom

), N

OM

INA

L P

EA

K P

OW

ER

(kW

)

Figure 18. Peak Power Ratings ofZener Diodes

1N6267 SERIES

5 WATT TYPES

250 mW TO 1 W TYPESGLASS DO-35 & GLASS DO-41

1 TO 3 W TYPES

PLASTIC DO-41

0.1 0.2 0.5 1 2 5 10 5020 100

D, DUTY CYCLE (%)

1

0.7

0.5

0.3

0.2

0.1

DE

RAT

ING

FA

CT

OR

0.07

0.05

0.03

0.02

0.01

Figure 19. Typical Derating Factor forDuty Cycle

PULSE WIDTH

10 ms

1 ms

100 μs

10 μs

When it is necessary to use a zener close to surge ratings,and a standard part having guaranteed surge limits is notsuitable, a special part number may be created having a surgelimit as part of the specification. Contact your nearestON Semiconductor OEM sales office for capability, price,delivery, and minimum order quantities.

http://onsemi.com53

MATHEMATICAL MODEL

Since the power shown on the curves is not the actualtransient power measured, but is the product of the peakcurrent measured and the nominal zener voltage measuredat the current used for voltage classification, the peak currentcan be calculated from:

IZ(PK) =P(PK)

VZ(nom)(5)

The peak voltage at peak current can be calculated from:

(6)VZ(PK) = FC x VZ(nom)

where FC is the clamping factor. The clamping factor isapproximately 1.20 for all zener diodes when operated attheir pulse power limits. For example, a 5 watt, 20 voltzener can be expected to show a peak voltage of 24 voltsregardless of whether it is handling 450 watts for 0.1 msor 50 watts for 10 ms. This occurs because the voltage isa function of junction temperature and IR drop. Heatingof the junction is more severe at the longer pulse width,causing a higher voltage component due to temperaturewhich is roughly offset by the smaller IR voltagecomponent.For modeling purposes, an approximation of the zenerresistance is needed. It is obtained from:

(7)RZ(nom) =VZ(nom)(FC−1)

PPK(nom) / VZ(nom)

The value is approximate because both the clampingfactor and the actual resistance are a function of temperature.

CIRCUIT CONSIDERATIONS

It is important that as much impedance as circuitconstraints allow be placed in series with the zener diode andthe components to be protected. The result will be a lowerclipping voltage and less zener stress. A capacitor in parallel

with the zener is also effective in reducing the stress imposedby very short duration transients.

To illustrate use of the data, a common application will beanalyzed. The transistor in Figure 20 drives a 50 mHsolenoid which requires 5 amperes of current. Without somemeans of clamping the voltage from the inductor when thetransistor turns off, it could be destroyed.

The means most often used to solve the problem is toconnect an ordinary rectifier diode across the coil; however,this technique may keep the current circulating through thecoil for too long a time. Faster switching is achieved byallowing the voltage to rise to a level above the supply beforebeing clamped. The voltage rating of the transistor is 60 V,indicating that approximately a 50 volt zener will berequired.

The peak current will equal the on-state transistor current(5 amperes) and will decay exponentially as determined bythe coil L/R time constant (neglecting the zener impedance).A rectangular pulse of width L/R (0.01 s) and amplitude ofIPK (5 A) contains the same energy and may be used to selecta zener diode. The nominal zener power rating thereforemust exceed (5 A × 50) = 250 watts at 10 ms and a duty cycleof 0.01/2 = 0.5%. From Figure 19, the duty cycle factor is0.62 making the single pulse power rating required equal to250/0.62 = 403 watts. From Figure 18, one of the 1N6267series zeners has the required capability. The 1N6287 isspecified nominally at 47 volts and should provesatisfactory.

Although this series has specified maximum voltagelimits, equation 7 will be used to determine the maximumzener voltage in order to demonstrate its use.

RZ =47(1.20 − 1)

500/479.4

10.64= = 0.9 Ω

At 5 amperes, the peak voltage will be 4.5 volts abovenominal or 51.5 volts total which is safely below the 60volt transistor rating.

Figure 20. Circuit Example

10 ms

2 s

50 mH, 5 Ω26 Vdc

USED TO SELECT A ZENER DIODE HAVING THE PROPERVOLTAGE AND POWER CAPABILITY TO PROTECT THE TRANSISTOR

http://onsemi.com54

ZENER VOLTAGE SENSING CIRCUITSAND APPLICATIONS

BASIC CONCEPTS OF VOLTAGE SENSING

Numerous electronic circuits require a signal or voltagelevel to be sensed for circuit actuation, function control, orcircuit protection. The circuit may alter its mode ofoperation whenever an interdependent signal reaches aparticular magnitude (either higher or lower than a specifiedvalue). These sensing functions may be accomplished byincorporating a voltage dependent device in the systemcreating a switching action that controls the overalloperation of the circuit.

The zener diode is ideally suited for most sensingapplications because of its voltage dependentcharacteristics. The following sections are some of the morecommon applications and techniques that utilize the zener ina voltage sensing capacity.

Figure 1. Basic Transistor-Zener DiodeSensing Circuits

R2

R1

R3

R1

R2Z1

R3

VIN

VINZ1 Q1

Q1

VO1

VO2

(a)

(b)

TRANSISTOR-ZENER SENSING CIRCUITS

The zener diode probably finds its greatest use in sensingapplications in conjunction with other semiconductordevices. Two basic widely used techniques are illustrated inFigures 1a and 1b.

In both of these circuits the output is a function of the inputvoltage level. As the input goes from low to high, the outputwill switch from either high to low (base sense circuit) orlow to high (emitter sense circuit), (see Figure 2).

The base sense circuit of Figure 1a operates as follows:When the input voltage is low, the voltage dropped across R2is not sufficient to bias the zener diode and base emitterjunction into conduction, therefore, the transistor will notconduct. This causes a high voltage from collector toemitter. When the input becomes high, the zener is biasedinto conduction, the transistor turns on, and the collector toemitter voltage, which is the output, drops to a low value.

Figure 2. Outputs of Transistor-Zener VoltageSensing Circuits

R2 + R1

R2VZ + VBE(SAT)VIN =

SENSING LEVEL

TIME

TIME

TIME

VIN BOTH CIRCUITS

VOUT BASE SENSE

VOUT EMITTER SENSE

VO2

VO1

VIN

(OUTPUT OF FIGURE 1b)

(OUTPUT OF FIGURE 1a)

http://onsemi.com55

The emitter sense circuit of Figure 1b operates as follows:When the input is low the voltage drop across R3 (the output)is negligible. As the input voltage increases the voltage dropacross R2 biases the zener into conduction and forwardbiases the base-emitter junction. A large voltage drop acrossR3 (the output voltage) is equal to the product of the collectorcurrent times the resistance, R3. The following relationshipsindicate the basic operating conditions for the circuits inFigure 1.

Circuit Output

1a

1b

HighVOUT = VIN − ICOR3 ≅ VIN

LowVOUT = VIN − ICR3 = VCE(sat)

LowVOUT = VIN − VZ − VCE(off) = ICOR3

HighVOUT = VIN − VCE(sat) = ICR3

In addition, the basic circuits of Figure 1 can be rearrangedto provide inverse output.

AUTOMOTIVE ALTERNATOR VOLTAGEREGULATOR

Electromechanical devices have been employed for manyyears as voltage regulators, however, the regulation setting

of these devices tend to change and have mechanical contactproblems. A solid state regulator that controls the charge rateby sensing the battery voltage is inherently more accurateand reliable. A schematic of a simplified solid state voltageregulator is shown in Figure 3.

The purpose of an alternator regulator is to control thebattery charging current from the alternator. The chargelevel of the battery is proportional to the battery voltagelevel. Consequently, the regulator must monitor the batteryvoltage level allowing charging current to pass when thebattery voltage is low. When the battery has attained theproper charge the charging current is switched off. In thecase of the solid state regulator of Figure 3, the chargingcurrent is controlled by switching the alternator field currenton and off with a series transistor switch, Q2. The switchingaction of Q2 is controlled by a voltage sensing circuit that isidentical to the base sense circuit of Figure 1a. Whenunder-charged, the zener Z1 does not conduct keeping Q1off. The collector-emitter voltage of Q1 supplies a forwardbias to the base-emitter of Q2, turning it on. With Q2 turnedon, the alternator field is energized allowing a chargingcurrent to be delivered to the battery. When the batteryattains a proper charge level, the zener conducts causing Q1to turn on, and effectively shorting out the base-emitterjunction of Q2. This short circuit cuts off Q2, turns off thecurrent flowing in the field coil which consequently, reducesthe output of the alternator. Diode D1 acts as a fieldsuppressor preventing the build up of a high induced voltageacross the coil when the coil current is interrupted.

Figure 3. Simplified Solid State Voltage Regulator

R3

R1

R2

R4

B+

Z1 Q1

Q2

ALTERNATOROUTPUT

D1ALTERNATORFIELD

http://onsemi.com56

In actual operation, this switching action occurs manytimes each second, depending upon the current drain fromthe battery. The battery charge, therefore, remainsessentially constant and at the maximum value for optimumoperation.

A schematic of a complete alternator voltage regulator isshown in Figure 4.

It is also possible to perform the alternator regulationfunction with the sensing element in the emitter of thecontrol transistor as shown in Figure 5.

In this configuration, the sensing circuit is composed of Z1and Q1 with biasing components. It is similar to the sensingcircuit shown in Figure 1b. The potentiometer R1 adjusts theconduction point of Q1 establishing the proper charge level.When the battery has reached the desired level, Q1 begins toconduct. This draws Q2 into conduction, and thereforeshorts off Q3 which is supplying power to the alternatorfield. This type of regulator offers greater sensitivity with anincrease in cost.

Figure 4. Complete Solid State Alternator Voltage Regulator

Figure 5. Alternator Regulator with Emitter Sensor

B+

100 Ω

15 Ω

30 Ω

30 Ω

70 Ω

1N961BSENSINGZENERDIODE

RT*THER-MISTOR

0.05 μFFEEDBACKCAPACITOR

1N4001FIELDSUPPRESSIONDIODE

TO ALTERNATORFIELD COIL

2N5879

1N3493BIASDIODE

ALTERNATOROUTPUT

0.05 Ω

2N4234

*THE VALUE OF RT DEPENDS ON THE SLOPE OF THE VOLTAGE REGULATIONVERSUS TEMPERATURE CURVE.

B+ ALTERNATOROUTPUT

D2

D1

Q3

R5

R4

Q2

R6

Z1

Q1

R3

R1

R2ALTERNATORFIELD

http://onsemi.com57

UNIJUNCTION-ZENER SENSE CIRCUITS

Unijunction transistor oscillator circuits can be madeGO-NO GO voltage sensitive by incorporating a zener diodeclamp. The UJT operates on the criterion: under properbiasing conditions the emitter-base one junction willbreakover when the emitter voltage reaches a specific valuegiven by the equation:

Vp = ηVBB + VD (1)

where:VpηVBBVD

= peak point emitter voltage= intrinsic stand-off ratio for the device= interbase voltage, from base two to base one= emitter to base one diode forward junction drop

Obviously, if we provide a voltage clamp in the circuitsuch that the conditions of equation 1 are met only withrestriction on the input, the circuit becomes voltagesensitive. There are two basic techniques used in clampingUJT relaxation oscillators. They are shown in Figure 6 andFigure 7.

The circuit in Figure 6 is that of a clamped emitter type.As long as the input voltage VIN is low enough so that Vp

does not exceed the Zener voltage VZ, the circuit willgenerate output pulses. At some given point, the required Vpfor triggering will exceed VZ. Since Vp is clamped at VZ, thecircuit will not oscillate. This, in essence, means the circuitis GO as long as VIN is below a certain level, and NO GOabove the critical clamp point.

The circuit of Figure 7, is a clamped base UJT oscillator.In this circuit VBB is clamped at a voltage VZ and the emittertied to a voltage dividing network by a diode D1. When theinput voltage is low, the voltage drop across R2 is less thanVp. The forward biased diode holds the emitter below thetrigger level. As the input increases, the R2 voltage dropapproaches Vp. The diode D1 becomes reversed biased and,the UJT triggers. This phenomenon establishes theoperating criterion that the circuit is NO GO at a low inputand GO at an input higher than the clamp voltage. Therefore,the circuits in Figures 6 and 7 are both input voltagesensitive, but have opposite input requirements for a GOcondition. To illustrate the usefulness of the clamped UJTrelaxation oscillators, the following two sections show thembeing used in practical applications.

Figure 6. UJT Oscillator, GO — NO GO Output,GO for Low VIN — NO GO for High VIN

+

−

RT

C ZUJT

RB2

RB1

VP = ηVBB + VD

VOUT

VIN

VE

VBB

Figure 7. UJT — NO GO Output, NO GO for Low VIN — GO for High VIN

Z

R1

R2

+

−

RT RB2

RB1

CT

D1

VIN

VOUT

UJTVE

http://onsemi.com58

BATTERY VOLTAGE SENSITIVE SCR CHARGER

A clamped emitter unijunction sensing circuit of the typeshown in Figure 6 makes a very good battery charger(illustrated in Figure 8). This circuit will not operate until thebattery to be charged is properly connected to the charger.The battery voltage controls the charger and will dictate itsoperation. When the battery is properly charged, the chargerwill cease operation.

The battery charging current is obtained through thecontrolled rectifier. Triggering pulses for the controlledrectifier are generated by unijunction transistor relaxationoscillator (Figure 9). This oscillator is activated when thebattery voltage is low.

While operating, the oscillator will produce pulses in thepulse transformer connected across the resistance, RGC(RGC represents the gate-to-cathode resistance of thecontrolled rectifier), at a frequency determined by theresistance, capacitance, R.C. time delay circuit.

Since the base-to-base voltage on the unijunctiontransistor is derived from the charging battery, it willincrease as the battery charges. The increase in base-to-basevoltage of the unijunction transistor causes its peak pointvoltage (switching voltage) to increase. These waveformsare sketched in Figure 9 (this voltage increase will tend tochange the pulse repetition rate, but this is not important).

Figure 8. 12 Volt Battery Charger Control

Figure 9. UJT Relaxation Oscillator Operation

RECTIFIED A.C.VOLTAGE FROMCHARGER

R3

C1

UJT

E B2R2

A C

GSCR

V

12 V

−

T1

B1

R1 � 3.9K, 1/2 W

R2 � 1K, POT.

R3 � 5.1K, 1/2 W

C1 � .25 μf

Z1 � 1N753, 6.2 V

SCR � MCR3813

VB1 B2

BATTERY

CHARGING

BATTERY

CHARGED

Z1

R1

R2

C1

TIME

TIME

TIME

VC1

VRGC

ZENER

VOLTAGE

UJT PEAK

POINT VOLTAGE

SCR

CONDUCTS

SCR

NONCONDUCTING

UJT B2

B1

RGCT1

VBATT.

+

UJT � 2N2646

T1 � PR1, 30T, no. 22

SEC, 45T, NO. 22

CORE: FERROX CUBE

203F181-303

A+

R1

Z1

http://onsemi.com59

When the peak point voltage (switching voltage) of theunijunction transistor exceeds the breakdown voltage of theZener diode, Z1, connected across the delay circuitcapacitor, C1, the unijunction transistor ceases to oscillate.If the relaxation oscillator does not operate, the controlledrectifier will not receive trigger pulses and will not conduct.This indicates that the battery has attained its desired chargeas set by R2.

The unijunction cannot oscillate unless a voltagesomewhere between 3 volts and the cutoff setting is presentat the output terminals with polarity as indicated. Therefore,the SCR cannot conduct under conditions of a short circuit,an open circuit, or a reverse polarity connection to thebattery.

ALTERNATOR REGULATOR FORPERMANENT MAGNET FIELD

In alternator circuits such as those of an outboard engine,the field may be composed of a permanent magnet. Thisincreases the problem of regulating the output by limitingthe control function to opening or shorting the output.Because of the high reactance source of most alternators,opening the output circuit will generally stress the bridgerectifiers to a very high voltage level. It is, therefore,apparent that the best control function would be shorting theoutput of the alternator for regulation of the charge to thebattery.

Figure 10 shows a permanent magnet alternator regulatordesigned to regulate a 15 ampere output. The two SCRs areconnected on the ac side of the bridge, and short out thealternator when triggered by the unijunction voltagesensitive triggering circuit. The sensing circuit is of the typeshown in Figure 7. The shorted output does not appreciablyincrease the maximum output current level.

A single SCR could be designed into the dc side of thebridge. However, the rapid turn-off requirement of this typeof circuit at high alternator speeds makes this circuitimpractical.

The unijunction circuit in Figure 10 will not oscillate untilthe input voltage level reaches the voltage determined by theintrinsic standoff ratio. The adjustable voltage divider willcalibrate the circuit. The series diode in the voltage dividercircuit will compensate for the emitter-base-one diodetemperature change, consequently, temperaturecompensation is necessary only for the zener diodetemperature changes.

Due to the delay in charging the unijunction capacitor,when the battery is disconnected the alternator voltage willproduce high stress voltage on all components before theSCRs will be fired. The 1N971B Zener was included in thecircuit to provide a trigger pulse to the SCRs as soon as thealternator output voltage level approaches 30 volts.

Figure 10. Permanent Magnet Field Alternator Regulator

SE

C. 1

SE

C. 2

PR

I.

MCR2304-2

ALT.OUT MCR

2304-2

MDA2500

1N960B

200 Ω 5K Ω 27 Ω

0.1 μF

2N2646

1N971B

T1

1N4001200 Ω

27 Ω

BATTERY+

T1

CORE: ARNOLD no. 4T5340 D1 DD1

PRIMARY 125 TURNSAWG 36

SEC no. 1 125 TURNSAWG 36

SEC no. 2 125 TURNSAWG 36

TRIFILAR WOUND

+

http://onsemi.com60

Figure 11. Zener-Resistor Voltage Sensitive Circuit

Figure 12. Improving Meter Resolution

+

−

Z

R

BASE CIRCUIT

VZ

+ +

VIN VOUT VIN VOUT

(LEVEL DETECTION) (MAGNITUDE REDUCTION)

Z

R

+

−

+

−

VZ = 20 V

VOLTMETER10 V � FULL SCALE

TYPICAL OUTPUTS

VIN VOUT

VIN = 24 V − 28 V

ZENER-RESISTOR VOLTAGE SENSING

A simple but useful sense circuit can be made from just aZener diode and resistor such as shown in Figure 11.

Whenever the applied signal exceeds the specific Zenervoltage VZ, the difference appears across the droppingresistor R. This level dependent differential voltage can beused for level detection, magnitude reduction, waveshaping, etc. An illustrative application of the simple seriesZener sensor is shown in Figure 12, where the resistor dropis monitored with a voltmeter.

If, for example, the input is variable from 24 to 28 volts,a 30 voltmeter would normally be required. Unfortunately,a 4 volt range of values on a 30 volt scale utilizes only 13.3%of the meter movement — greatly limiting the accuracy with

which the meter can be read. By employing a 20 volt zener,one can use a 10 voltmeter instead of the 30 volt unit, therebyutilizing 40% of the meter movement instead of 13.3% witha corresponding increase in accuracy and readability. Forultimate accuracy a 24 volt zener could be combined with a5 voltmeter. This combination would have the disadvantageof providing little room for voltage fluctuations, however.

In Figure 13, a number of sequentially higher-voltageZener sense circuits are cascaded to actuate transistorswitches. As each goes into avalanche its respectiveswitching transistor is turned on, actuating the indicatorlight for that particular voltage level. This technique can beexpanded and modified to use the zener sensors to actuatesome form of logic system.

Figure 13. Sequential Voltage Level Indicator

Z1

Q1 Q2 Q3

R1 R2 R3RE2RE1 RE3

Z2 Z3LIGHT

(1)

LIGHT

(2)

LIGHT

(3)

INPUT OUTPUT

http://onsemi.com61

MISCELLANEOUS APPLICATIONSOF ZENER TYPE DEVICES

INTRODUCTION

Many of the commonly used applications of zener diodeshave been illustrated in some depth in the precedingchapters. This chapter shows how a zener diode may be usedin some rarer applications such as voltage translators, toprovide constant current, wave shaping, frequency controland synchronized SCR triggers.

The circuits used in this chapter are not intended asfinished designs since only a few component values aregiven. The intent is to show some general broad ideas andnot specific designs aimed at a narrow use.

FREQUENCY REGULATION OF ADC TO AC INVERTER

Zener diodes are often used in control circuits, usually tocontrol the magnitude of the output voltage or current. In thisunusual application, however, the zener is used to control theoutput frequency of a current feedback inverter. The circuitis shown in Figure 1.

Figure 1. Frequency Controlled CurrentFeedback Inverter

Z1

Q1

Q2

T1

NC

N6

N6

NC

B1

B2

+

A

−

N1

N1

N2

T2

LOAD

The transformer T1 functions as a current transformerproviding base current IB = (NC/NB)IC. Without the zener

diode, the voltage across NB windings of the timingtransformer T1 is clamped to VBE of the ON device, givingan inverter frequency of

f =VBE x 108

4BS1A1NB

where BS1A1 is the flux capacity of T1 transformer core.The effect on output frequency of VBE variations due tochanging load or temperature can be reduced by using azener diode in series with VBE as shown in Figure 1. Forthis circuit, the output frequency is given by

f =(VBE + VZ) x 108

4BS1A1NB

If VBE is small compared to the zener voltage VZ, goodfrequency accuracy is possible. For example, with VZ =9.1 volts, a 40 Watt inverter using 2N3791 transistors(operating from a 12 volt supply), exhibited frequencyregulation of ±2% with ±25% load variation.

Care should be taken not to exceed V(BR)EBO of thenon-conducting transistor, since the reverse emitter-basevoltage will be twice the introduced series voltage, plus VBEof the conducting device.

Transformer T2 should not saturate at the lowest inverterfrequency.

Inverter starting is facilitated by placing a resistor frompoint A to B1 or a capacitor from A to B2.

SIMPLE SQUARE WAVE GENERATOR

The zener diode is widely used in wave shaping circuits;one of its best known applications is a simple square wavegenerator. In this application, the zener clips sinusoidalwaves producing a square wave such as shown in Figure 2a.In order to generate a wave with reasonably vertical sides,the ac voltage must be considerably higher than the zenervoltage.

Clipper diodes with opposing junctions built into thedevice are ideal for applications of the type shown inFigure 2b.

http://onsemi.com62

(a) Single Zener Diode Square Wave Generator

(b) Opposed Zener Diodes Square Wave Generator

Figure 2. Zener Diode Square Wave Generator

Z1

Z

Z2

R

R

ZENER

VOLTAGE

FORWARD

DROP

VOLTAGE

ZENER Z1

VOLTAGE

ZENER Z2

VOLTAGE

A.C. INPUT OUTPUT

A.C. INPUT OUTPUT

http://onsemi.com63

TRANSIENT VOLTAGE SUPPRESSION

INTRODUCTIONElectrical transients in the form of voltage surges have

always existed in electrical distribution systems, and prior tothe implementation of semiconductor devices, they were ofminor concern. The vulnerability of semiconductors tolightning strikes was first studied by Bell Laboratories in1961.1 A later report tried to quantify the amount of energycertain semiconductors could absorb before they sufferedlatent or catastrophic damage from electrostatic discharge.2

Despite these early warnings, industry did not begin toaddress the issue satisfactorily until the late 1970s. Listedbelow are the seven major sources of overvoltage.

• Lightning• Sunspots• Switching of Loads in Power Circuits• Electrostatic Discharge• Nuclear Electromagnetic Pulses• Microwave Radiation• Power Cross

Most electrical and electronic devices can be damaged byvoltage transients. The difference between them is theamount of energy they can absorb before damage occurs.Because many modern semiconductor devices, such assmall signal transistors and integrated circuits can bedamaged by disturbances that exceed the voltage ratings atonly 20 volts or so, their survivability is poor in unprotectedenvironments.

In many cases, as semiconductors have evolved theirruggedness has diminished. The trend to produce smallerand faster devices, and the advent of MOSFET and galliumarsenide FET technologies has led to an increasedvulnerability. High impedance inputs and small junctionsizes limit the ability of these devices to absorb energy andto conduct large currents. It is necessary, therefore, tosupplement vulnerable electronic components with devicesspecially designed to cope with these hazards. Listed beloware the four primary philosophies for protecting againsttransients.

• Clamping, or “clipping” is a method of limiting theamplitude of the transient.

• Shunting provides a harmless path for the transient,usually to ground by way of an avalanche or a crowbarmechanism.

• Interrupting opens the circuit for the duration of thetransient.

• Isolating provides a transient barrier between hostileenvironments and vulnerable circuits through the use oftransformers or optoisolators.

Selection of the proper protective method should be madebased upon a thorough investigation of the potential sourcesof the overvoltage hazard. Different applications andenvironments present different sources of overvoltage.

LIGHTNINGAt any given time there are about 1800 thunderstorms in

progress around the world, with lightning striking about 100times each second. In the U.S., lightning kills about 150people each year and injures another 250. In flat terrain withan average lightning frequency, each 300 foot structure willbe hit, on average, once per year. Each 1200 foot structure,such as a radio or TV tower, will be hit 20 times each year,with strikes typically generating 600 million volts.

Each cloud-to-ground lightning flash really contains fromthree to five distinct strokes occurring at 60 ms intervals,with a peak current of some 20,000 amps for the first strokeand about half that for subsequent strokes. The final strokemay be followed by a continuing current of around 150 ampslasting for 100 ms.

The rise time of these strokes has been measured at around200 nanoseconds or faster. It is easy to see that thecombination of 20,000 amps and 200 ns calculates to a valueof dI/dt of 1011 amps per second! This large value means thattransient protection circuits must use RF design techniques,particularly considerate of parasitic inductance andcapacitance of conductors.

While this peak energy is certainly impressive, it is reallythe longer-term continuing current which carries the bulk ofthe charge transferred between the cloud and ground. Fromvarious field measurements, a typical lightning model hasbeen constructed, as shown in Figure 1.

Figure 1. Typical Lightning Model, with and withoutContinuing Current

Flash with Continuing Current

Flash with No Continuing Current

40 μs

6060

TIME (ms)

TIME (ms)

60 0.3 100

150 A

http://onsemi.com64

Depending on various conditions, continuing current mayor may not be present in a lightning strike. A severe lightningmodel has also been created, which gives an indication of thestrength which can be expected during worst case conditionsat a point very near the strike location. Figure 2 shows thismodel. Note that continuing current is present at more thanone interval, greatly exacerbating the damage which can beexpected. A severe strike can be expected to ignitecombustible materials.

Figure 2. Severe Lightning Model

460 8607006405805201601106010

400 A

200 A

TIME (ms)

A direct hit by lightning is, of course, a dramatic event. Infact, the electric field strength of a lightning strike nearbymay be enough to cause catastrophic or latent damage tosemiconductor equipment. It is a more realistic venture to tryto protect equipment from these nearby strikes than toexpect survival from a direct hit.

With this in mind, it is important to be able to quantify theinduced voltage as a function of distance from the strike.Figure 3 shows that these induced voltages can be quite high,explaining the destruction of equipment from relativelydistant lightning flashes.

Figure 3. Voltage Induced by NearbyLightning Strike

10000

1000

100

10

11010.1

DISTANCE FROM STRIKE (km)

IND

UC

ED

VO

LTA

GE

IN 1

m O

F W

IRE

(V

)Burying cables does not provide appreciable protection as

the earth is almost transparent to lightning radiated fields. Infact, underground wiring has a higher incidence of strikesthan aerial cables.3

SUNSPOTSThe sun generates electromagnetic waves which can

disrupt radio signals and increase disturbances on residentialand business power lines. Solar flares, which run in cyclesof 11 years (1989 was a peak year) send out electromagneticwaves which disrupt sensitive equipment.

Although not quantified, the effects of sunspot activityshould be considered. Sunspots may be the cause ofsporadic, and otherwise unexplainable problems in suchsensitive areas.

SWITCHING OF LOADS IN POWER CIRCUITSInductive switching transients occur when a reactive load,

such as a motor or a solenoid coil, is switched off. Therapidly collapsing magnetic field induces a voltage acrossthe load’s winding which can be expressed by the formula:

V = −L (dI/dt)where L is inductance in henrys and dI/dt is the rate ofchange of current in amps per second.

Such transients can occur from a power failure or thenormal opening of a switch. The energy associated with thetransient is stored within the inductance at powerinterruption and is equal to:

W = 1/2 Li2

where W is energy in joules and i is instantaneous current inamps at the time of interruption.

As an example, a 1.4 to 2.5 kV peak transient can beinjected into a 120 vac power line when the ignition systemof an oil furnace is fired. It has also been shown that there aretransients present on these lines which can reach as high as6 kV. In locations without transient protection devices, themaximum transient voltage is limited to about 6 kV by theinsulation breakdown of the wiring.

Inductive switching transients are the silent killers ofsemiconductors as they often occur with no outwardindication. A graphic example is the report of a largeelevator company indicating the failure of 1000 voltrectifiers during a power interruption. In another area, powerinterruption to a 20 HP pump motor in a remote area wasdirectly related to failure of sensitive monitoring equipmentat that same site.4

http://onsemi.com65

Figure 4. Switching Transient Definition for Aircraftand Military Buses, per Boeing Document D6-16050

28 VdcB A

WAVEFORM AT POINT A

TIME (μs)

V

t1t2

V = 600 V pk-pkNo. of Repetitions = 5 to 100

t1 = 0.2 to 10 μst2 = 50 to 1000 μs

After characterizing electrical overstress on aircraftpower buses, Boeing published Document D6-16050 asshown in Figure 4.

The military has developed switching transientdefinitions within several specifications including:

DOD-STD-1399 for shipboardMIL-STD-704 for aircraftMIL-STD-1275 for ground vehicles

The International Electrotechnical Commission (IEC) isnow promoting their specification IEC 801-4 throughoutthe European community. This describes an inductiveswitching transient voltage threat having 50 ns wide spikeswith amplitudes from 2 kV to 4 kV occurring in 300 ms widebursts.5

Besides these particular military specifications, many areapplication specific and functional tests exist. A supplier oftransient voltage suppressor components will be expected toperform to a wide variety of them.

ELECTROSTATIC DISCHARGE (ESD)ESD is a widely recognized hazard during shipping and

handling of many semiconductor devices, especially thosethat contain unprotected MOSFETs, semiconductors for useat microwave frequencies and very high speed logic withswitching times of 2 ns or less. In response to this threat,most semiconductors are routinely shipped in containersmade of conductive material.

In addition to various shipping precautions, electronicassembly line workers should be grounded, usegrounded-tip soldering irons, ionized air blowers and othertechniques to prevent large voltage potentials to begenerated and possibly discharged into the semiconductorsthey are handling.

Once the assembled device is in normal operation, ESDdamage can still occur. Any person shuffling his feet on acarpet and then touching a computer keyboard can possiblycause a software crash or, even worse, damage the keyboardelectronics.

The electrical waveform involved in ESD is a brief pulse,with a rise time of about 1 ns, and a duration of 100−300 μs.The peak voltage can be as large as 30 kV in dry weather, butis more commonly 0.5−5.0 kV.6 The fastest rise times occurfrom discharges originating at the tip of a hand-held tool,while discharges from the finger tip and the side of the handare slightly slower.7 A typical human with a bodycapacitance of 150 pF, charged to 3 microcoulombs, willdevelop a voltage potential of 20 kV, according to theformula:

V = Q / C

where V is voltage, Q is charge and C is capacitance. Theenergy delivered upon discharge is:

W = 1/2 CV2

where W is energy in joules, C is capacitance and V isvoltage.

It is interesting to note that most microcircuits can bedestroyed by a 2500 volt pulse, but a person cannot feel astatic spark of less than 3500 volts!

NUCLEAR ELECTROMAGNETIC PULSES (NEMP)When a nuclear weapon is detonated, a very large flux of

photons (gamma rays) is produced. These rays act toproduce an electromagnetic field known as a nuclearelectromagnetic pulse or NEMP. When a nuclear detonationoccurs above the atmosphere, a particulary intense pulseilluminates all objects on the surface of the earth, and allobjects in the lower atmosphere within line of sight of theburst. A burst 300−500 km above Kansas would illuminatethe entire continental U.S.

A typical NEMP waveform is a pulse with a rise time ofabout 5 ns and a duration of about 1 μs. Its peak electric fieldis 50−100 kV/m at ground level. After such a pulse iscoupled into spacecraft, aircraft and ground supportequipment, it produces a waveform as described inMIL-STD-461C. The insidious effect of NEMP is its broadcoverage and its potential for disabling military defensesystems.

MICROWAVE RADIATIONMicrowaves can be generated with such high power that

they can disable electronic hardware upon which manymilitary systems depend. A single pulse flux of 10−8 MJ/cm2

burns out receiver diodes, and a flux of 10−4 MJ/cm2 causesbit errors in unshielded computers.8 With automobilesutilizing MPU controls in more applications, it is importantto protect against the effects of driving by a microwavetransmitter. Likewise, a nearby lightning strike could alsohave detrimental effects to these systems.

POWER CROSSYet another source of electrical overstress is the accidental

connection of signal lines, such as telephone or cabletelevision, to an ac or dc power line. Strictly speaking, thisphenomenon, known as a power cross, is a continuous state,not a transient. However, the techniques for ensuring the

http://onsemi.com66

survival of signal electronics after a power cross are similarto techniques used for protection against transientovervoltages.

STANDARDIZED WAVEFORMSFortunately, measurements of these hazards have been

studied and documented in several industry specifications.For example, Bellcore Technical AdvisoryTA-TSY-000974 defines the generic measurementwaveform for any double exponential waveform, which isthe basis for most of the specific applications norms.

The predominant waveform for induced lightningtransients, set down by Rural Electrification AdministrationDocument PE-60, is shown in Figure 5. This pulse test,performed at the conditions of 100 V/μs rise, 10/1000 μs, Ip= 1 kV, is one of the two most commonly specified in theindustry. The other is the 8/20 μs waveform, shown inFigure 6.

Figure 5. Pulse Waveform (10/1000 μs)

Figure 6. Pulse Waveform (8/20 μs)

3020100

lp.9 lp

.5 lp

.1 lp

TIME, μs

1 2 3

.1 Ip

.5 Ip

.9 IpIp

0

TIME, ms

TRANSIENT VOLTAGE SUPPRESSIONAND TELECOM

TRANSIENT VOLTAGE SURGE SUPPRESSIONCOMPONENTS ON DATA AND TELEPHONE LINES

Lines carrying data and telephone signals are subject to anumber of unwanted and potentially damaging transientsprimarily from two sources: lightning and “power crosses.”A power cross is an accidental connection of a signal line toa powerline. Transients from lightning can impress voltageswell above a kilovolt on the line but are of short duration,usually under a millisecond. Lightning transients aresuppressed by using Transient Voltage Surge Suppressor

(TVS) devices. TVS devices handle high peak currentswhile holding peak voltage below damaging levels, but haverelatively low energy capability and cannot protect againsta power cross fault. The first TVS used by telephonecompanies is the carbon block, but its peak let-throughvoltage was too high for modern equipment usingunprotected solid state circuitry. A number of othercomponents fill today’s needs.

The power cross condition causes a problem withtelephone lines. Fast acting fuses, high speed circuitbreakers and positive temperature coefficient thermistorshave been successfully used to limit or interrupt currentsurges exceeding a millisecond.

Over the years, telecommunications switching equipmenthas been transitioning from electromechanical relays tointegrated circuits and MOSFET technology. The newerequipment operates at minimal electrical currents andvoltages, which make it very efficient. It is therefore quitesensitive to electrical overloads caused by lightning strikesand other transient voltage sources, and by power crosses.

Because of the deployment of new technology, both innew installations and in the refurbishment of older systems,the need for transient protection has grown rapidly. It iswidely recognized that any new equipment must includeprotection devices for reasons of safety, reliability and longterm economy.

The major telecom companies, in their never ending questfor the elimination of electromechanical technology havebeen looking at a number of novel methods andimplementations of protection. These methods provide forsolutions to both the primary and secondary protectioncategories.

A number of studies have been conducted to determine thetransient environment on telephone lines. Very little hasbeen done with data lines because a typical situation does notexist. However, information gathered from telephone linestudies can serve as a guide for data lines.

Past studies on telephone lines coupled with the highcurrent capability of arc type arrestors and the conservativenature of engineers seem to have produced specificationswhich far exceed the real need. A recent study by Bell SouthServices9 reported that the highest level of transient energyencountered was well below standards and specifications incommon use. Now, solid state devices perform adequatelyfor many applications. However the stringent specificationsof some regulatory agencies promote arc-type arrestors,though solid state devices would be a better choice.

PRIMARY PROTECTION

Primary protection is necessary to protect against highvoltage transients which occur in the outdoor environment.These transients include induced lightning surges and accross conditions.

As such, primary protection is located at the point wherewiring enters the building or terminal box. It is the first linedefense against outside hazards. TVS devices located wherelines enter a building are called primary protectors.

http://onsemi.com67

Protectors connected to indoor lines are referred to assecondary protectors. Both primary and secondaryprotectors are required to provide complete equipmentprotection.

Today, primary protection is most generally accomplishedthrough the use of surge protector modules. For telecom,these are designed specifically for the environment and thestandards dictated by the telecom applications. Theytypically contain a two or three element gas arrestor tube anda mechanically-triggered heat coil. Some also include airgap carbon block arrestors which break over at voltagesabove about 1500 volts.

Some modules contain high speed diodes for clampresponse in the low nanoseconds. This provides protectionuntil the gas tube fires, generally in about one microsecond.The diodes may be connected between the tip, ring andground conductors in various combinations. The 5ESSelectronic switching system norms dictate design andperformance requirements of TVS modules in use today.Test methods are spelled out in REA PE-80, a publicationof the Rural Electrification Administration.

In the U.S. alone, 58 million primary protection modulesare sold annually, about 40 million for central offices and 18million for station locations, such as building entrances.Eighty percent of these use gas tubes, 16% use air-gapcarbon blocks, and only 2% (so far) are solid state.

SECONDARY PROTECTION

Secondary protection is necessary for the equipmentinputs, and as such, is located between the primary protectorand the equipment. Secondary protection is generallyaccomplished with one or more TVS components, asopposed to the modules used for primary. It is often placedon a circuit board along with other components handlingother duties, such as switching. Secondary protection isapplied to lines associated with long branch circuits whichhave primary protection a significant distance away, tointernal data lines, and to other locations requiringadditional local hazard-proofing.

While not as open to external transients as the primary,secondary can still see peak open circuit voltages in excessof 1000 volts and short circuit currents of hundreds of amps.These transients may be locally generated, or they may beresiduals from the primary protectors upstream.

STANDARDS

Transient voltage waveforms are commonly described interms of a dual exponential wave as defined in Figure 7. Thestandard chosen for power lines is a 1.2/50 μs voltage wavewhich causes an 8/20 μs current wave. Although the sourceof the most severe transients on telecom lines is the same asfor power lines and lightning, the higher impedance per unitlength of the telephone line stretches the waves as theypropagate through the lines.

Figure 7. Definition of Double ExponentialImpulse Waveform

P0.9 P

0.5 P

0.1 P

T0

a b T1 T2TIME

WAVEFORM IS DEFINED AS tr/tdWHERE

tr : FRONT TIME = 1.25 (b - a)

= (T1-T0)

td : DURATION = (T2 -T0)

The 10/1000 μs wave approximates the worst casewaveform observed on data and telecom lines. TVS devicesintended for this service are usually rated and characterizedusing a 10/1000 waveform. The Bell South study revealedthat the worst transient energy handled by primaryprotectors on lines entering a central office was equivalentto only 27 A peak of a 10/1000 wave. This level isconsiderably less than that required by secondary protectorsin most of the standards in use today. This finding isparticularly significant because the Bell South service areaincludes Central Florida, the region experiencing the highestlightning activity in the U.S.

The United States Federal Communications Commission(FCC) has defined mandatory requirements for equipmentwhich is to be connected to the U.S. telephone network. Insome cases, U.S. equipment must meet standards developedby the Rural Electrification Agency (REA). Many nationsdemand compliance to standards imposed by theConsultative Committee, International Telegraph andTelephone (CCITT). In addition, most equipment usersdemand safety certification from U.L., which has its ownstandards.

The FCC Standards are based on a worst case residue froma carbon block primary protector installed where the phoneline enters the building. The CCITT standard is applicablefor situations lacking primary protection, other than wiringflashover. Companies entering the telephone equipment orprotector market will need to obtain and become familiarwith the appropriate governing standards.

TRANSIENT VOLTAGE PROTECTIONCOMPONENTS

GENERAL TVS CHARACTERISTICS

A number of transient voltage suppressor (TVS) devicesare available. Each finds use in various applications basedupon performance and cost. All types are essentiallytransparent to the line until a transient occurs; however,some devices have significant capacitance which loads theline for ac signals. A few of the these are described inTable 1.

Based upon their response to an overvoltage, TVS devicesfit into two main categories, clamps and crowbars. A clampconducts when its breakdown voltage is exceeded andreverts back to an open circuit when the applied voltage

http://onsemi.com68

drops below breakdown. A crowbar switches into a lowvoltage, low impedance state when its breakover voltage isexceeded and restores only when the current flowingthrough it drops below a “holding” level.

CLAMP DEVICES

All clamp devices exhibit the general V-I characteristic ofFigure 8. There are variations; however, some clamps areasymmetric. In the non clamping direction, some devicessuch as the zener TVS exhibit the forward characteristic ofa diode while others exhibit a very high breakdown voltageand are not intended to handle energy of “reverse” polarity.Under normal operating conditions, clamp devices appearvirtually as an open circuit, although a small amount ofleakage current is usually present. With increasing voltagea point is reached where current increases rapidly withvoltage as shown by the curved portion of Figure 8. Therapidly changing curved portion is called the “knee region.”Further increases in current places operation in the“breakdown” region.

Figure 8. Static Characteristics of a Clamp Device

0

V

I O

In the knee region the V-I characteristic of clampingdevices can be approximated by the equation:

I = K Vs (1)

where K is a constant of proportionality and s is an exponentwhich defines the “sharpness factor” of the knee. Theexponent s is 1 for a resistor and varies from 5 to over 100for the clamping devices being used in TVS applications. Ahigh value of s i.e., a sharp knee, is beneficial. A TVS devicecan be chosen whose breakdown voltage is just above theworst case signal amplitude on the line without concern ofloading the line or causing excessive dissipation in the TVS.

As the current density in the clamp becomes high, theincremental resistance as described by Equation 1 becomesvery small in comparison to the bulk resistance of thematerial. The incremental resistance is therefore ohmic inthe high current region.

Unfortunately, a uniform terminology for all TVS deviceshas not been developed; rather, the terms were developed inconjunction with the appearance of each device in themarketplace. The key characteristics normally specifieddefine operation at voltages below the knee and at currentsabove the knee.

Leakage current is normally specified below the knee ata voltage variously referred to as the stand-off voltage, peakworking voltage or rated dc voltage. Some devices are ratedin terms of an RMS voltage, if they are bidirectional. Normalsignal levels must not exceed this working voltage if thedevice is to be transparent.

Breakdown voltage is normally specified at a fairly lowcurrent, typically 1 mA, which places operation past theknee region. Worst case signal levels should not exceed thebreakdown voltage to avoid the possibility of circuitmalfunction or TVS destruction.

The voltage in the high current region is called theclamping voltage, VC. It is usually specified at the maximumcurrent rating for the device. To keep VC close to thebreakdown voltage, s must be high and the bulk resistancelow. A term called clamping factor, (FC) is sometimes usedto describe the sharpness of the breakdown characteristic.FC is the ratio of clamping voltage to the breakdown voltage.As the V-I characteristic curve of the TVS approaches a rightangle, the clamping factor approaches unity. Clampingfactor is not often specified, but it is useful to describe clampdevice behavior in general terms.

Table 1. Comparison of TVS Components

TypeProtection

TimeProtection

VoltagePower

DissipationReliable

PerformanceExpected

LifeOther

Considerations

GAS TUBE > 1 μs 60−100 V Nil No Limited Only 50−2500 surges.Can short power line.

MOV 10−20 ns > 300 V Nil No Degrades Fusing required. Degrades.Voltage level too high.

AVALANCHE TVS 50 ps 3−400 V Low Yes Long Low power dissipation.Bidirectional requires dual.

THYRISTOR TVS < 3 ns 30−400 V Nil Yes Long High capacitance.Temperature sensitive.

http://onsemi.com69

Clamp devices generally react with high speed and as aresult find applications over a wide frequency spectrum. Nodelay is associated with restoration to the off state afteroperation in the breakdown region.

CROWBAR DEVICES

Crowbar TVS devices have the general characteristicsshown in Figure 9. As with clamp devices, asymmetriccrowbars are available which may show a diode forwardcharacteristic or a high voltage breakdown in one direction.

Figure 9. Static Characteristics of aBidirectional Crowbar Device

VD V3O2

VT

I2

I1

3

21

V3O1VT

The major difference between a crowbar and a clamp isthat, at some current in the breakdown region, the deviceswitches to a low voltage on-state. In the clamping regionfrom I1 to I2, the slope of the curve may be positive as shownby segment 1, negative (segment 2) or exhibit bothcharacteristics as shown by segment 3. A slightly positiveslope is more desirable than the other two curves because anegative resistance usually causes a burst of high frequencyoscillation which may cause malfunction in associatedcircuitry. However, a number of performance andmanufacturing trade-offs affect the shape of the slope in theclamping region.

A crowbar TVS has an important advantage over a clampTVS in that it can handle much larger transient surge currentdensities because the voltage during the surge isconsiderably lower. In a telephone line application, forexample, the clamping level must exceed the ring voltagepeak and will typically be in the vicinity of 300 volts duringa high current surge. The on-state level of the crowbar maybe as low as 3 volts for some types which allows about twoorders of magnitude increase in current density for the samepeak power dissipation.

However, a crowbar TVS becomes “latched” in theon-state. In order to turn off its current flow the drivingvoltage must be reduced below a critical level called theholding or extinguishing level. Consequently, in anyapplication where the on-state level is below the normalsystem voltage, a follow-on current occurs. In a dc circuitcrowbars will not turn off unless some means is provided tointerrupt the current. In an ac application crowbars will turnoff near the zero crossing of the ac signal, but a time delayis associated with turn-off which limits crowbars to

relatively low frequency applications. In a data line ortelecom application the turn-off delay causes a loss ofintelligence after the transient surge has subsided.

A telephone line has both ac and dc signals present.Crowbars can be successfully used to protect telecom linesfrom high current surges. They must be carefully chosen toensure that the minimum holding current is safely above themaximum dc current available from the lines.

TVS DEVICESA description of the various types of TVS devices follows

in the chronological order in which they became available.Used appropriately, sometimes in combination, anytransient protection problem can be suitably resolved. Theirsymbols are shown in Figure 10.

Figure 10. TVS Devices and Their Symbols

Air-GapCarbon Block

2- and 3-ElementGas Tubes

Heat CoilSwitch

Metal OxideVaristor (MOV)

ZenerRegulator

UnidirectionalAvalanche TVS De-

vice

BidirectionalAvalanche TVS Device

Thyristor TVS Devices Dual Thyristor TVS Devices

AIR GAP ARRESTORS

The air gap is formed by a pair of metal points rigidly fixedat a precise distance. The air ionizes at a particular voltagedepending upon the gap width between the points. As the airionizes breakover occurs and the ionized air provides a lowimpedance conductive path between the points.

The breakover threshold voltage is a function of the air’srelative humidity; consequently, open air gaps are usedmainly on high voltage power lines where preciseperformance is not necessary. For more predictablebehavior, air gaps sealed in glass and metal packages are alsoavailable.

Because a finite time is required to ionize the air, the actualbreakover voltage of the gap depends upon the rate of rise ofthe transient overvoltage. Typically, an arrestor designed fora 120 V ac line breaks over at 2200 volts.

Air gaps handle high currents in the range of 10,000amperes. Unfortunately, the arc current pits the tips whichcauses the breakover voltage and on-state resistance toincrease with usage.

http://onsemi.com70

CARBON BLOCK ARRESTORS

The carbon block arrestor, developed around the turn ofthe century to protect telephone circuits, is still in place inmany older installations. The arrestor consists of two carbonblock electrodes separated by a 3 to 4 mil air gap. The gapbreaks over at a fairly high level − approximately 1 kV − andcannot be used as a sole protection element for moderntelecom equipment. The voltage breakdown level isirregular. With use, the surface of the carbon block is burnedwhich increases the unit’s resistance. In addition, the burnedmaterial forms carbon tracks between the blocks causing aleakage current path which generates noise. Consequently,many of the carbon blocks in service are being replaced bygas tubes and are seldom used in new installations.

SILICON CARBIDE VARISTORS

The first non-linear resistor to be developed was called a“varistor.” It was made from specially processed siliconcarbide and found wide use in high power, high voltage TVSapplications. It is not used on telecom lines because itsclamping factor is too high: s is only about 5.

GAS SURGE ARRESTORS

Gas surge arrestors are a sophisticated modification of theair gap more suited to telecom circuit protection. Most oftenused is the “communication” type gap which measuresabout 3/8 inch in diameter and 1/4 inch thick. A cross sectionis shown in Figure 11. They consist of a glass or ceramicenvelope filled with a low pressure inert gas with specializedelectrodes at each end. Most types contain a minute quantityof radioactive material to stabilize breakover voltage.Otherwise, breakover is sensitive to the level of ambientlight.

Because of their small size and fairly wide gap,capacitance is very low, only a few picofarads. When notactivated, their off-state impedance or insulation resistanceis virtually infinite.

ÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇ

ÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇÇ

Figure 11. Gas Arrestor Cross Section

ELECTRODES

DISCHARGE

REGION

INSULATOR

(GLASS OR

CERAMIC MATERIAL)

ACTIVATING

COMPOUNDIGNITION AID

Key electrical specifications for this TVS type includebreakover voltage (dc & pulse), maximum holdovervoltage, arc voltage, and maximum surge current.

The breakover voltage is rated at a slow rate of rise,5000 V/s, essentially dc to a gas arrestor. Typical dc voltageratings range from 75 V through 300 V to provide for mostcommunication systems protection requirements. Themaximum pulse voltage rating is that level at which thedevice fires and goes into conduction when subjected to afairly rapid rate of voltage rise, (dv/dt) usually 100 V/μs.Maximum rated pulse voltages typically range from 400 Vto 600 V, depending on device type.

A typical waveform of a gas surge arrestor responding toa high voltage pulse is shown in Figure 12. From thewaveform, it can be seen that the dv/dt of the wave is 100v/μs and the peak voltage (the breakover voltage) is 520 V.

Figure 12. Voltage Waveform of Gas Surge ArrestorResponding to a Surge Voltage

AR

RE

STO

R V

OLT

AG

E (

VO

LTS

)

TIME (500 ns/DIVISION)

0

200

400

600

0 1 2 3

Gas surge arrestors fire faster but firing voltage increasesas the transient wave fronts increase in slope as illustrated inFigure 13. The near vertical lines represent the incidenttransient rise time. Note that the response time is greater than0.1s at slow rise times but decreases to less than 0.1 μs forrisetimes of 20 kV/μs. However, the firing voltage hasincreased to greater than 1000 V for the gas tube whichbreaks over at 250 Vdc.

The driving circuit voltage must be below the holdovervoltage for the gap to extinguish after the transient voltagehas passed. Holdover voltage levels are typically 60% to70% of the rated dc breakdown voltage.

Arc voltage is the voltage across the device duringconduction. It is typically specified at 5 to 10 V under a lowcurrent condition, but can exceed 30 V under maximumrated pulse current.

The maximum surge current rating for a 8/20 μswaveform is typically in the 10 kA to 20 kA range forcommunication type devices. For repetitive surges with a10/1000 wave, current ratings are typically 100 A,comfortably above the typical exposure levels in a telephonesubscriber loop.

http://onsemi.com71

1.2

1.0

0.8

0.6

0.4

0.2

0

1.4

8 7 6 5 4 3 2 1 1 10

RESPONSE TIME (SECONDS)

SP

AR

KO

VE

R V

OLT

AG

E (

kV)

10

0 V

/ sμ

10

0 V

/ms

1

00 V

/SE

C

Figure 13. Typical Response Time of a Gas Surge Arrestor

470 Vdc

350 Vdc

250 Vdc

150 Vdc

500

V/�

�sμ

1KV

/��sμ

5KV

/��sμ

10K

V/�

�sμ

20K

V/�

�sμ

Gas tubes normally provide long life under typicaloperating conditions, however; wear-out does occur.Wear-out is characterized by increased leakage current andfiring voltage. An examination of gas tubes in service for sixto eight years revealed that 15% were firing outside of theirspecified voltage limits.9 Because firing voltage increaseswith use, protectors often use an air gap backup in parallelwith the gas tube. End-of-life is often specified bymanufacturers as an increase of greater than 50% ofbreakover or firing voltage. Other limits include a decreasein leakage resistance to less than 1 mW.

The features and limitations of gas tube surge arrestors arelisted below.

Advantages:• High current capability• Low capacitance• Very high off-state impedanceDisadvantages:• Slow response time• Limited life• High let-through voltage• Open circuit failure mode

Principally because of their high firing voltage, gas surgearrestors are not suitable for use as the sole element toprotect modern equipment connected to a data or telecomline. However, they are often part of a protection networkwhere they are used as the primary protector at the buildinginterface with the outside world.

SELENIUM CELLS

Polycrystalline diodes formed from a combination ofselenium and iron were the forerunners of monocrystalline

semiconductor diodes. The TVS cells are built by depositingthe polycrystalline material on a metal plate to increase theirthermal mass thereby raising energy dissipation. The cellsexhibit typical diode characteristics and a non-linear reversebreakdown which is useful for transient suppression. Cellscan be made which are “self-healing”; that is, the damagewhich occurs when subjecting them to excessive transientcurrent is repaired with time.

Selenium cells are still used in high power ac lineprotection applications because of their self-healingcharacteristic; however, their high capacitance and poorclamping factor (s » 8) rule them out for data or telecom lineapplications.

METAL OXIDE VARISTORS

The metal oxide varistor (MOV) is composed of zincoxide granules in a matrix of bismuth and other metaloxides. The interface between the zinc oxide and the matrixmaterial exhibits characteristics similar to that of a p-njunction having a voltage breakdown of about 2.6 V. Withthis structure the electrical equivalent is that of groups ofdiodes in parallel which are stacked in series with similarparallel groups to provide the desired electrical parameters.The taller the stack, the higher the breakdown and operatingvoltage. Larger cross-sections provide higher currentcapability. The structure of an MOV is shown in Figure 14.

MICROVARISTOR

ZINC OXIDE

INTERGRANULAR

Figure 14. MOV Cross Section

http://onsemi.com72

MOV (27 V)Vert: 10 V/divHoriz: 0.5 ms/divTransient Source Impedance: 0.55 ΩVpeak = 62.5 V

Figure 15. MOV Clamping Voltage Waveform

MOVs, formed from a ceramic-like material, are usuallyproduced in the shape of discs with most widely used MOVshaving diameters of 7 mm, 14 mm, and 20 mm. The discsurfaces are coated with a highly conductive metal such assilver to assure uniform conduction through the crosssectional area of the device. After terminal attachment theparts are coated with a durable plastic material.

The typical voltage spectrum of MOVs ranges from 8 Vthrough 1000 V for individual elements. Pulse currentcapability (8/20 μs) ranges from a few amperes to severalthousands of amperes depending on the element’s size. TheV-I characteristic of MOVs is similar to Figure 8. Theirclamping factor is fairly good; s is in the vicinity of 25.

Key electrical specifications include: operating voltage,breakdown voltage, peak current maximum clampingvoltage, and leakage current.

The maximum operating voltage specified is chosen to bebelow the breakdown voltage by a margin sufficient toproduce negligible heating under normal operatingconditions. Breakdown voltage is the transition point atwhich a small increase in voltage results in a significantincrease in current producing a clamping action. Maximumlimits for breakdown voltage are typically specified at1 mA with upper end limits ranging from 20% to 40%greater than the minimum breakdown voltage.

Maximum peak current is a function of element area andranges from tens of amperes to tens of thousands of amperes.MOVs are typically pulse rated with an 8/20 μs waveformsince they are intended primarily for use across power lines.

The clamping characteristics of a 27 V ac rated MOV, witha 4 joule maximum pulse capability is shown in Figure 15.The transient energy is derived from an exponentiallydecreasing pulse having a peak amplitude of 90 V. The pulsegenerator source impedance is 0.55 W. Peak clampingvoltage is 62.5 V while the developed current is 50 A. Theclamping factor calculates to be 2.3.

Leakage currents are listed for MOVs intended for use insensitive protection applications but are not normally listed

for devices most often used on power lines. Leakage currentbehavior is similar to that of a p-n junction. It roughlydoubles for every 10°C increase in temperature and alsoshows an exponential dependence upon applied voltage. Ata voltage of 80% of breakdown, leakage currents are severalmicroamperes at a temperature of 50°C.

Although the theory of MOV operation is not fullydeveloped, behavior is similar to a bidirectional avalanchediode. Consequently its response time is very fast.

Life expectancy is an important characteristic generatedunder pulse conditions. A typical example is shown inFigure 16. The data applies to 20 mm diameter disc typeshaving rated rms voltage from 130 V to 320 V. Lifetimerating curves are usually given for each device family.

IMPULSE DURATION � μs

RA

TE

D P

ULS

E C

UR

RE

NT

− A

MP

ER

ES

10,0005000

2000

1000500

200100

50

2010

5

21

Figure 16. MOV Pulse Life Curve

10,000100010020

INDEFINITE

105

104

103

12

10

102

106

For a single 8/20 μs pulse, the device described in Figure16 is rated at 6500 A; however, it must be derated by morethan two orders of magnitude for large numbers of pulses.Longer duration pulses also require further derating. Forexample, for a 10/1000 μs duration pulse, this family ofdevices has a maximum pulse rating of about 100 A on asingle shot basis and devices must be derated to less than10A for long lifetimes in excess of 100,000 pulses.

http://onsemi.com73

End-of-life for an MOV is defined as the voltagebreakdown degrading beyond the limits of ± 10%. As MOVsare pulsed, they degrade incrementally as granularinterfaces are overheated and changed to a highlyconductive state. Failure occurs in power line applicationswhen the breakdown voltage has degraded to the pointwhere the MOV attempts to clip the powerline peaks. Intelecom applications, their breakdown must be above thepeaks of the impressed ac line during a ring cycle or a powercross; otherwise an immediate catastrophic failure willoccur.

When MOVs fail catastrophically they initially fail short.However, if a source of high energy is present as might occurwith a power cross, the follow-on current may cause the partto rupture resulting in an open circuit.

The advantages and shortcomings of using an MOV forgeneral purpose protection in microprocessor basedcircuitry include the following:

Advantages:• High current capability• Broad voltage spectrum• Broad current spectrum• Fast response• Short circuit failure modeDisadvantages:• Gradual decrease of breakdown voltage• High capacitance

The capacitance of MOVs is fairly high because a largedevice is required in order to achieve a low clamping factor;consequently, they are seldom used across telecom lines.

ZENER TVS

Zener TVS devices are constructed with large area siliconp-n junctions designed to operate in avalanche and handlemuch higher currents than their cousins, zener voltageregulator diodes. Some manufacturers use small area mesachips with metal heatsinks to achieve high peak powercapability. However, ON Semiconductor has determinedthat large area planar die produce lower leakage current andclamping factor. The planar construction cross section isshown in Figure 17 and several packages are shown inFigure 18.

Figure 17. Zener TVS Cross-Section

ÈÈÈÈÈÈÈÈ

PLASTICENCAPSULATION

PLANAR DIE SOLDER

Figure 18. Typical Insertion and Surface Mount SiliconTVS Packages — Zeners and Thyristors

59-0317-02 41A

318-07 403A403

Key electrical parameters include maximum operatingvoltage, maximum reverse breakdown voltage, peak pulsecurrent, peak clamping voltage, peak pulse power, andleakage current.

The normal operating or working voltage is usually calledthe reverse standoff voltage in specification sheets. Devicesare generally available over the range of 5 V through 250 V.Standoff voltage defines the maximum peak ac or dc voltagewhich the device can handle. Standoff voltage is typically10% to 15% below minimum reverse breakdown voltage. Alisting of TVS products available from ON Semiconductoris shown in Table 2.

Table 2. ON Semiconductor Zener TVS Series

DEVICESERIES

VZRANGE

PULSE POWERRATING

(100/1000PULSE) PACKAGE

*SA5.0A-SA170A

6.8-200 500 W Axial

*P6KE6.8A -P6KE200A

6.8-200 600 W Axial

*1.5KE6.8A -1.5KE250A

6.8-250 1500 W Axial

1SMB5.0AT3 -1SMB170AT3

6.8-200 600 W SMB

P6SMB6.8AT3 -P6SMB200T3

6.8-200 600 W SMB

1SMC5.0AT3 -1SMC78AT3

6.8-91 1500 W SMC

1.5SMC6.8AT31.5SMC91AT3

6.8-91 1500 W SMC

MMBZ15VDLT1 15ESD Protection

>15 kVSOT-23

* Available in bidirectional configurations

http://onsemi.com74

The reverse breakdown voltage is specified at a bias levelat which the device begins to conduct in the avalanche mode.Test current levels typically are 1 mA for diodes whichbreakdown above 10 V and 10 mA for lower voltagedevices. Softening of the breakdown knee, that is, lower s,for lower voltage p-n junction devices requires a higher testcurrent for accurate measurements of reverse breakdownvoltage. Diodes that break down above 10 V display a verysharp knee; s is over 100.

Peak pulse current is the maximum upper limit at whichthe device is expected to survive. Silicon p-n junctions arerated for constant power using a particular transientwaveform; consequently, current is a function of the peakclamping voltage. For example, a 6.8 V device handlesabout 28 times the pulse current that a 220 V device willwithstand; however, both the 6.8 V and 220 V types dissipatethe same peak power under the same pulse waveformconditions. Most Zener TVS diodes are rated for 10/1000 μswaveform pulses which are common in the telecom industry.

The clamping voltage waveform of a 27 V Zener TVShaving a 1.5 joule capability is illustrated in Figure 19. Itspeak voltage is 30.2 V. The transient energy source is thesame as applied to the MOV whose response is shown inFigure 10. However, the current through the Zener TVS isover 100 A, much higher than occurs with the MOV becausethe clamping voltage is significantly lower. Despite thehigher pulse current, the Zener displays much betterclamping action; its clamping factor is 1.1.

Figure 14

Figure 19. Zener TVS Clamping Voltage Waveform

TIME (500 μs/DIVISION)

0

10

20

30

40

50

60

V

, CLA

MP

ING

VO

LTA

GE

(V

OLT

S)

C

Peak pulse power is the instantaneous power dissipated atthe rated pulse condition. Common peak pulse power ratingsare 500 W, 600 W, and 1500 W for 10/1000 μs waveforms.As the pulse width decreases, the peak power capabilityincreases in a logarithmic relationship. An example of acurve depicting peak pulse power versus pulse width isshown in Figure 20. The graph applies to the 1.5 kW series(10/1000 pulse) of TVS diodes and can be interpolated todetermine power ratings over a broad range of pulse widths.At 50 μs, the maximum rated power shown in the curve is

6 kW, which is four times greater than the rating at 1ms. Thecurrent handling capability is also increased roughly by thissame factor of four.

Figure 20. Peak Pulse Power Rating for a Popular Zener TVS Family

1�μs 10�μs 100�μs 1 ms 10 ms

100

10

1

tP, PULSE WIDTH

P P, P

EA

K P

OW

ER

(kW

)

0.1�μs

To increase power capability devices are stacked in series.For example, doubling the power capability requirement fora 100 V, 1.5 kW Zener TVS is easily done by placing two 50V devices in series. Clamping factor is not significantlyaffected by this arrangement.

Although leakage current limits are relatively high for theindustry low voltage types (500 μA to 1000 μA), droppingoff to 5 μA or less for voltages above 10 V, the planar die inuse by ON Semiconductor exhibit considerably less leakagethan the specified limits of the industry types.

Capacitance for the popular 1500 W family exceeds10,000 pF at zero bias for a 6.8 V part, droppingexponentially to less than 100 pF for a 200 V device.Capacitance drops exponentially with a linear increase inbias. The capacitance of a 6.8 V device is 7000 pF, while the200 V part measures under 60 pF, at their respective standoffvoltages.

Capacitance loads the signal line at high frequencies. Forhigh speed data transmission circuits, low capacitance isachieved by placing two diodes in a series stack as shown inFigure 21. Under normal operation the top diode (DS)operates at essentially zero bias current. Since its powerdissipation requirement is small, its area can be muchsmaller than that of the TVS diode (DZ) in order to providelow capacitance. The top diode normally is not intended tobe used in avalanche. Consequently if a negative voltageexceeding the reverse rating of the stack could occur, the lowcapacitance diode must be protected by another diode (DP)shown connected by dotted lines on Figure 21. Thearrangement of Figure 21 is satisfactory for situations wherethe signal on the line is always positive. When the signal isac, diode DP is replaced by another low capacitance stack,connected in anti-parallel.

http://onsemi.com75

DS

DZ

DP

Figure 21. A Series Stack to Achieve LowCapacitance with Zener TVS Diodes

Switching speed is a prime attribute of the zener TVS.Avalanche action occurs in picoseconds but performing teststo substantiate the theory is extremely difficult. As apractical matter, the device may be regarded as respondinginstantaneously. Voltage overshoots which may appear onprotected lines are the result of poor layout and packagingor faulty measurement techniques.

The p-n junction diode is a unidirectional device. For useon ac signal lines, bidirectional devices are available whichare based upon stacking two diodes back to back. Mostmanufacturers use monolithic NPN and PNP structures. Thecenter region is made relatively wide compared to atransistor base to minimize transistor action which can causeincreased leakage current.

No wearout mechanism exists for properly manufacturedZener diode chips. They are normally in one of two states;good, or shorted out from over-stress. Long-term life studiesshow no evidence of degradation of any electricalparameters prior to failure. Failures result from stress whichcauses separation of the metal heat sink from the silicon chipwith subsequent overheating and then failure. Like MOVs,silicon chips quickly fail short under steady state or longduration pulses which exceed their capabilities.

The strengths and weaknesses of Zener TVS devices arelisted below.

Advantages:• High repetitive pulse power ratings• Low clamping factor• Sub-nanosecond turn-on• No wearout• Broad voltage spectrum• Short circuit failure modeDisadvantages:• Low non-repetitive pulse current• High capacitance for low voltage types

Because of their fast response and low clamping factor,silicon devices are used extensively for protectingmicroprocessor based equipment from voltage surges on dcpower buses and I/O ports.

THYRISTOR DIODES

The most recent addition to the TVS arsenal is thethyristor surge suppressor (TSS). The device has the lowclamping factor and virtually instantaneous responsecharacteristic of a silicon avalanche (Zener) diode but, inaddition, it switches to a low voltage on-state whensufficient avalanche current flows. Because the on-statevoltage is only a few volts, the TSS can handle much highercurrents than a silicon diode TVS having the same chip areaand breakdown voltage. Furthermore, the TSS does notexhibit the large overshoot voltage of the gas tube.

Thyristor TVS diodes are available with unidirectional orbidirectional characteristics. The unidirectional typebehaves somewhat like an SCR with a Zener diodeconnected from anode to gate. The bidirectional typebehaves similarly to a triac having a bidirectional diode(Diac) from main terminal to gate.

Packaged TSS chips are shown in Figure 18. Figure 22ashows a typical positive switching resistance bidirectionalTSS chip. Construction of the device starts with an nmaterial wafer into which the p-bases and n-emitters arediffused. There are four layers from top to bottom on eachside of the chip, forming an equivalent SCR. Only half thedevice conducts for a particular transient polarity.

http://onsemi.com76

Figure 22a. Chip Construction

ÈÈÈÈÈÈÈÈ

ÈÈÈÈÈÈÈÈ

A

K

Figure 22c. Circuit Modelof Left Side

A

K

P EMITTER

N BASE

N E

MIT

TE

R

P B

AS

E

Figure 22b. Cross Sectionof Left Side

MT1 MT2

Figure 22d. Circuit Symbol

The “gate” does not trigger the SCR, instead, operation inthe Zener mode begins when the collector junctionavalanches. Note that the p-bases pass through then-emitters in a dot pattern and connect to the contact metalcovering both halves of the chip. This constructiontechnique provides a low resistance path for current flowand prevents it from turning on the NPN transistor.Therefore, at relatively low currents, the device acts like alow gain PNP transistor in breakdown. The Zener diode isthe collector base junction of the PNP transistor. Negativeresistance TSS devices are similarly constructed but startwith p-type material wafer, allowing the fabrication of ahigh-gain NPN transistor. The switchback in voltage withincreasing current is caused by the gain of the NPN.

Both device types switch on completely when the currentflow through the base emitter shunt resistance causesenough voltage drop to turn on the emitter and begin fourlayer action. Now the device acts like an SCR. The collectorcurrent of the PNP transistor (Figure 22c) provides the basedrive for the NPN transistor. Likewise, the collector current

of the NPN transistor drives the base of the PNP causing thetwo devices to hold one another on. Both the p and n emittersflood the chip with carriers resulting in high electricalconductivity and surge current capability.

When driven with high voltage ac, which occurs during apower cross, positive resistance TSS devices act like a Zenerdiode until the ac voltage drives the load line through thepoint where regeneration occurs. Then it abruptly switchesto a low voltage. When the peak ac current is just below thecurrent required for breakover, the device operates mainlyas a Zener and power dissipation is high although the currentis low. When the ac current peak is well above breakover,(>10 A), the device operates mainly as an SCR, and the lowon-state voltage causes power dissipation to be relativelylow.

Negative resistance devices operate in a similar fashion.However, their behavior is dependent upon the “load line,”that is, the equivalent resistance which the device “sees.”When the load resistance is high (>1000 W) behavior issimilar to that of a positive resistance TSS in that high

http://onsemi.com77

instantaneous power dissipation occurs as the load line isdriven along the high voltage region of the TSS prior toswitching.

When a TSS with a negative resistance characteristic isdriven with a low “load” resistance, switching occurs whenthe load line is tangent to the peak of the negative resistancecurve. Thus, complete turn-on can occur at a very lowcurrent if the load resistance is low and the device has a“sharp” switchback characteristic.

Leakage and the Zener knee voltage increase withtemperature at eight percent and 0.11% per °C respectivelyfor 200 V positive resistance types. But the current requiredto cause regeneration falls with temperature, causing lessZener impedance contribution to the breakover voltage,resulting in a large reduction in the breakover voltagetemperature coefficient to as little as 0.05%/°C. Negativeresistance types can show positive or negative breakovervoltage coefficients depending on temperature and thesharpness of the negative switchback.

The response of both positive and negative switchingresistance units to fast transients involves a race betweentheir Zener and regenerative attributes. At first the deviceconducts only in the small chip area where breakdown isoccurring. Time is required for conduction to spread acrossthe chip and to establish the currents and temperaturesleading to complete turn-on. The net result is that both typesexhibit increasing breakover voltage with fast transients.However, this effect is very small compared to gas dischargetubes, being less than 25% of the breakover voltage.

Negative resistance types are more sensitive to unwantedturn-on by voltage rates (dv/dt) at peak voltages below theavalanche value. The transient current that flows to chargethe self-capacitance of the device sets up an operating pointon the negative resistance slope leading to turn-on.Reduction of dv/dt capability becomes significant when thesignal voltage exceeds 80% of the avalanche value.

Complete turn-off following a transient requires the loadline to intersect the device leakage characteristic at a pointbelow the avalanche knee. During turn-off the load line mustnot meet an intermediate conducting state which can occurwith a negative resistance device. Positive resistance typesare free of states causing turn-off “sticking.” Both typeshave temperature sensitive holding currents that lie between1 and 4 mA/°C.

Recent product developments and published studies havegenerated much interest. Based on a study sponsored by BellSouth,9 the authors concluded that these new devices offeredthe highest level of surge protection available.

Key electrical parameters for the Thyristor TVS includeoperating voltage, clamping voltage, pulse current, on-statevoltage, capacitance, and holding current.

Operating voltage is defined as the maximum normalvoltage which the device should experience. Operatingvoltages from 60 V to 200 V are available.

Clamping voltage is the maximum voltage level attainedbefore thyristor turn-on and subsequent transition to the

on-state conduction mode. The transition stage toconduction may have any of the slopes shown in Figure 9.

The important voltages which define the thyristoroperating characteristics are also shown in Figure 9. VD isoperating voltage, VC is the clamping voltage and VT ison-state voltage.

On-state voltage for most devices is approximately 3 V.Consequently, transient power dissipation is much lower forthe thyristor TVS than for other TVS devices because of itslow on-state voltage. For example, under power crossconditions Bell South Services reported their tests showedthat the thyristor TVS devices handled short bursts ofcommercial power with far less heating than arc type surgearrestors.9

Capacitance is also a key parameter since in many casesthe TSS is a replacement for gas surge arrestors which havelow capacitance. Values for the TSS range from 100 pF to200 pF at zero volts, but drop to about half of these valuesat a 50 Vdc bias.

Holding current (IH) is defined as the current required tomaintain the on-state condition. Device thru-current mustdrop below IH before it will restore to the non-conductingstate. Turn-off time is usually not specified but it can beexpected to be several milliseconds in a telecom applicationwhere the dc follow-on current is just slightly below theholding current.

The major advantages and limitations of the thyristor are:

Advantages:• Fast response• No wearout• Produces no noise• Short circuit failure modeDisadvantages:• Narrow voltage spectrum• Non-restoring in dc circuits unless current is below IH• Turn-off delay time

The thyristor TVS is finding wide acceptance in telecomapplications because its characteristics uniquely matchtelecom requirements. It handles the difficult “power cross”requirements with less stress than other TVS devices whileproviding the total protection needed.

SURGE PROTECTOR MODULESTYPES OF SURGE PROTECTOR MODULES

Several component technologies have been implementedeither singly or in combination in surge protector modulesand devices. The simplest surge protectors contain nothingmore than a single transient voltage suppression (TVS)component in a larger package. Others contain two or morein a series, parallel or series-parallel arrangement. Stillothers contain two or more varieties of TVS elements incombination, providing multiple levels of protection.

Many surge protectors contain non-semiconductorelements such as carbon blocks and varistors. If required,

http://onsemi.com78

other modes of protection components may be incorporated,such as circuit breakers or EMI noise filters.

Surge protector modules are one solution to theovervoltage problem. Alternatives include:

• Uninterruptible power systems (UPS), whose mainduty is to provide power during a blackout, butsecondarily provide protection from surges, sags andspikes.

• Power line conditioners, which are designed to isolateequipment from raw utility power and regulatedvoltages within narrow limits.

Both UPS and power line conditioners are far moreexpensive than surge protector modules.

THE 6 MAJOR CATEGORIES OFSURGE PROTECTION MODULES

Plug-inHardwiredUtilityDatacomTelecomRF and microwave

PLUG-IN MODULES

Plug-in modules come in a variety of sizes and shapes, andare intended for general purpose use. They permit theprotection of vulnerable electronic equipment, such as homecomputers, from overvoltage transients on the 115 vac line.These products are sold in retail outlets, computer stores andvia mail order. Most models incorporate a circuit breaker orfuse, and an on/off switch with a neon indicator. The modulemay have any number of receptacles, with common modelshaving from two to six. These products comply with UL1449, and are generally rated to withstand the application ofmultiple transients, as specified in IEEE 587. Plug-inmodules generally provide their protection through the useof these devices which are typically connected between lineand neutral, and between neutral and ground.

HARDWIRED

Hardwired modules take on a wide variety of styles,depending upon their designed application. They provideprotection for instrumentation, computers, automatic testequipment, industrial controls, motor controls, and forcertain telecom situations.

Many of these modules provide snubbing networksemploying resistors and capacitors to produce an RC timeconstant. Snubbers provide common mode and differentialmode low-pass filtering to reduce interference from line toequipment, and are effective in reducing equipmentgenerated noise from being propogated onto the line.Snubbers leak current however, and many of these modulesare designed with heat sinks and require mounting to achassis. The surge protection is performed in a similarmanner to the plug-in modules mentioned earlier.Hardwired products, therefore, present a prime opportunityfor avalanche TVS components.

UTILITY

The power transmission and distribution equipmentindustry has an obvious need for heavy duty protectionagainst overvoltage transients. Many utility situationsrequire a combination of techniques to provide the necessarysolution to their particular problems. This industry utilizesmany forms of transient suppression outside the realm ofsemiconductors.

DATACOM

Local area networks and other computer links requireprotection against high energy transients originating on theirdata lines. In addition, transients on adjacent power linesproduce electromagnetic fields that can be coupled ontounprotected signal lines. Datacom protectors have a groundterminal or pigtail which must be tied to the local equipmentground with as short a lead as possible. Datacom protectorsshould be installed on both ends of a data link, or at all nodesin a network. This protection is in addition to the ac linetransient protection, which is served by the plug-in orhardwired protection modules. Some datacom protectormodules contain multi-stage hybrid circuits, speciallytailored for specific applications, such as 4−20 mA analogcurrent loops.

TELECOM

Included here are devices used to protect central office andstation telecommunications (telecom) equipment againstvoltage surges. None of these devices are grounded throughan ac power receptacle. Those that are grounded through anac power receptacle are categorized as plug-in modules. Notonly can overvoltages cause disruptions of telecom service,but they can destroy the sophisticated equipment connectedto the network. Also, users or technicians working on theequipment can be injured should lightning strike nearby. Itis estimated that 10 to 15 people are killed in the U.S. eachyear while talking on the telephone during lightning storms.For these reasons, surge protectors are used both in centraloffices and in customer premises.

There are three types of telecom surge protectors now inservice: air-gap carbon block, gas tube, and solid state. Thedesire of the telecom market is to convert as many of thenon-solid state implementations into solid state as cost willpermit.

SELECTING TVS COMPONENTSFrom the foregoing discussion it should be clear that the

silicon junction avalanche diode offers more desirablecharacteristics than any other TVS component. Its ability toclamp fast rising transients without overshoot, low clampingfactor, non-latching behavior, and lack of a wearoutmechanism are the overriding considerations. Its one-shotsurge capability is lower than most other TVS devices but isnormally adequate for the application. Should an unusuallysevere event occur, it will short yet still protect theequipment.

http://onsemi.com79

For example, an RS-232 data line is specified to operatewith a maximum signal level of ± 25 V. Failure analysisstudies11 have shown that the transmitters and receiversused on RS-232 links tolerate 40 V transients. A1.5KE27CA diode will handle the maximum signal levelwhile holding the peak transient voltage to less than 40 Vwith a 40 A 10/1000 pulse which is adequate for all indoorand most outdoor data line runs. As a practical matter, fewdata links use 25 V signals; 5 V is most common.Consequently, much lower voltage silicon diodes may beused which will allow a corresponding increase in surgecurrent capability. For example, a 10 V breakdown devicefrom the same 1500 W family will clamp to under 15 V(typically 12 V) when subjected to a 100 A pulse.

Telecommunications lines which must accommodate thering voltage have much more severe requirements. Forexample, one specification11 from Bellcore suggests thatleakage current be under 20 mA over the temperature rangefrom −40°C to +65°C with 265 V peak ac applied. To meetthis specification using Zener TVS parts, devices must bestacked. Devices which breakdown at 160 V are chosen toaccommodate tolerances and the temperature coefficient. Apart number with a 10% tolerance on breakdown couldsupply a unit which breaks down at 144 V. At −40°Cbreakdown could be 133 V. The breakdown of two devicesstacked just barely exceeds the worst case ring peak of265 V. A 1500 W unit has a surge capability of 6.5 A(10/1000) which is too low to be satisfactory while higherpower units are too expensive as a rule.

Another problem which telephone line service presents toa surge suppressor is survival during a power cross. Anavalanche diode is impractical to use because the energydelivered by a power cross will produce diode failure beforeany overcurrent protective element can react.

As indicated by the Bell South studies, the thyristor TVSis ideal for telephone line applications. Suppliers offerunidirectional and bidirectional units which meet the FCCimpulse wave requirements as shown in Table 1. In addition,the thyristor can handle several cycles of 50/60 Hz powerbefore failure. The ON Semiconductor MKT1V200 series,for example, can handle 10 A for 4 cycles, which is enoughtime for a low current fuse or other current activatedprotective device to react.

APPLICATION CONSIDERATIONSIn most cases, it is not advisable to place a zener TVS

directly across a data line because of its relatively highcapacitance. The arrangement previously discussed andshown in Figure 21 works well for an unbalanced line suchas RS-232. When using discrete steering diodes, they shouldhave low capacitance and low turn-on impedance to avoidcausing an overshoot on the clamped voltage level.

Most noise and transient surge voltages occur on lineswith respect to ground. A signal line such as RS-232 whichuses ground as a signal reference is thus very vulnerable tonoise and transients. It is, however, easy to protect using asingle TVS at each end of the line.

Telephone and RS-422 lines are called balanced linesbecause the signal is placed between two lines which are“floating” with respect to ground. A signal appears betweeneach signal line and ground but is rejected by the receiver;only the difference in potential between the two signal linesis recognized as the transmitted signal. This system has beenin use for decades as a means of providing improved noiseimmunity, but protection from transient surge voltages ismore complex.

A cost-effective means of protecting a balanced line isshown in Figure 23. The bridge diode arrangement allowsprotection against both positive and negative transients to beachieved, an essential requirement but the TVS devices Z1and Z2 need only be unidirectional. The diodes are chosento have low capacitance to reduce loading on the line andlow turn-on impedance to avoid causing an overshoot on theclamped voltage level. Although a zener TVS is shown, aTSS is more appropriate when a telephone line having ringvoltages is to be protected.

D2D1

D3 D4

Z2Z1

Figure 23. A Method of Reducing Capacitance andProtecting a Balanced Line

Since transients are usually common-mode, it is importantthat the TVS circuit operate in a balanced fashion;otherwise, common mode transients can cause differentialmode disturbances which can be devastating to the linereceiver. For example, suppose that an identical positivecommon mode surge voltage appears from eachline-to-ground. Diodes D2 and D4 will conduct the transientsto Z2. However, if one of these diodes has a slower turn-onor higher dynamic impedance than the other, the voltagedifference caused by the differing diode response appearsacross the signal lines. Consequently, the bridge diodes mustbe chosen to be as nearly identical as possible.

Should a differential mode transient appear on the signallines, it will be held to twice that of the line-to-groundclamping level. In many cases a lower clamping level isneeded which can be achieved by placing another TVSacross the signal lines. It must be a bidirectional low

http://onsemi.com80

capacitance device. With a line-to-line TVS in the circuitdiode matching is not required.

Other schemes appearing in the literature use twobidirectional TVS devices from each line-to-ground asshown in Figure 24 that of the line-to-ground voltage. Toavoid generating a differential mode signal, the TVS mustbe closely matched or a third TVS must be placedline-to-line. By using a third TVS differential modetransients can be held to a low level.

T3

TIP

RING

T1

T2

Figure 24. Protecting a Balanced Line withBidirectional TSS Devices

The arrangement of Figure 25 offers the advantage oflower capacitance when differential mode transientprotection is required. If all three TSS devices have the samecapacitance (C), the line-to-line and line-to-groundcapacitance of Figure 24 is 1.5C. However, the arrangementof Figure 25 exhibits a capacitance of only C/2. To design thecircuit to handle the same simultaneous common modeenergy as the circuit of Figure 24, T3 must be twice as largeas T1 and T2. In this case the capacitance of T3 is doubledwhich causes the line-to-ground capacitance to be 2C/3, stilla considerable improvement over the arrangement ofFigure 24.

T3

TIP

RING

T1

T2

Figure 25. Preferred Method of Protecting aBalanced Line Using Bidirectional TSS Devices

Protectors are usually designed to be “fail safe” if theirenergy ratings are exceeded, but the definition of “fail safe”is often dependent upon the application. The most commonrequirement is that the surge voltage protective elementshould fail short and remain shorted regardless of the

resulting current flow. To insure that this occurs,semiconductor devices use heavy gauge clips or bondingwires between the chip and terminals. In addition, parts areavailable in plastic packages having a spring type shortingbar which shorts the terminals when the package softens atthe very high temperatures generated during a severeoverload.

The shorted TVS protects the equipment, but the linefeeding it could be destroyed if the source of energy whichshorted the TVS is from a power cross. Therefore, it is wiseand necessary for a UL listing to provide a series elementsuch as a fuse or PTC device to either open the circuit orrestrict its value to a safe level.

The design of circuit boards is critical and layout must bedone to minimize any lead or wiring inductance in serieswith the TVS. Significant voltage is developed in any loopsubject to transients because of their high current amplitudesand fast risetimes.

REFERENCES1. D. W. Bodle and P. A. Gresh, “Lightning Surges in

Paired Telephone Cable Facilities,” The Bell SystemTechnical Journal, Vol. 4, March 1961, pp. 547−576.

2. D. G. Stroh, “Static Electricity Can Kill Transistors,”Electronics, Vol. 35, 1962, pp. 90−91.

3. J. D. Norgard and C. L. Chen, “Lightning-InducedTransients on Buried Shielded Transmission Lines,”IEEE Transactions on EMC, Vol. EMC28, No. 3,August 1986, pp. 168−171.

4. O. M. Clark, “Transient Voltage Suppression (TVS),”1989, pp. 6−7.

5. Clark, p. 7.6. World Information Technologies, “U. S. Electrical and

Electronic Surge Protection Markets,” 1989, p. 5.7. T. J. Tucker, “Spark Initiation Requirements of a

Secondary Explosive,” Annals of the New YorkAcademy of Sciences, Vol. 152, Article I, 1968, pp.643−653.

8. H. K. Florig, “The Future Battlefield: A Blast ofGigawatts?,” IEEE Spectrum, Vol. 25, No. 3, March1988, pp. 50−54.

9. Mel Thrasher, “A Solid State Solution,” Telephony,June 1989, pp. 48−52.

10. A. Urbieta, “Sensitivity Study to EOS/SSD of BipolarIntegrated Circuits,” EOS-8, 1986.

11. M. Tetreault and F.D. Martzloff, “Characterization ofDisturbing Transient Waveforms on Computer DataCommunication Lines,” EMC Proceedings, Zurich,March 1985, pp. 423−428.

12. F. Martzloff, “Coupling, Propagation, and Side Effectsof Surges in an Industrial Building Wiring System,”

http://onsemi.com81

Conference Record of IEEE IAS Meeting, 1988, pp.1467−1475.

IMPORTANT REGULATORY REQUIREMENTSAND GUIDELINES

GENERAL

DOD-STD-1399, MIL-STD-704, MIL-STD-1275,MIL-STD-461C. These military specifications areimportant, if we intend to target devices for military orcommercial aviation markets.

IEEE 587. This specification describes multiple transientwaveforms.

UL1449. This is a compulsory test which demonstratesperformance to criteria establishing the maximum voltagethat can pass through a device after clamping has takenplace. It is important that we comply, and say so on our datasheets.

INDUSTRIAL

ANSI/IEEE C62.41. Established by the AmericanNational Standards Institute (ANSI) and the Institute ofElectrical and Electronic Engineers (IEEE), this guidelinetests the effectiveness of devices to typical powerdisturbances. To meet the most rigorous category of thisspec, a device or module must be capable of withstanding amaximum repetitive surge current pulse of 3000 amps witha 8/20 μs waveform.

IEC TC 102 D. Requirements for remote controlreceivers for industrial applications are detailed in thisInternational Electrotechnical Commission (IEC)document.

IEC 255-4 and IEC TC 41. These documents describetesting for static relays for industrial use.

IEC 801-1 thru -3. These are specifications for variousindustrial control applications.

IEC 801-4. The IEC specifies transient voltage impulseswhich occur from the switching of inductive loads. We mustbe aware of the importance of this specification, especiallyin Europe, and characterize our devices’ performance to it.

IEEE 472/ANSI C 37.90.1. Requirements for protectiverelays, including 10/1000 nS waveform testing is described.

UL943. This requirement defines a 0.5 μs/100 kHzwaveform for ground fault and other switching applications.

VDE 0420. Industrial remote control receivers aredetailed, and test procedures defined.

VDE 0860/Part 1/II. This includes a description of0.2/200 μs, 10kV test requirements.

TELECOM

CCITT IX K.17, K.20, K.15. These documents relate torepeaters.

EIA PM-1361. This document covers requirements fortelephone terminals and data processing equipment.

FTZ 4391 TV1. This is a general German specificationfor telecom equipment.

FCC Part 68. The Federal Communications Commission(FCC) requirements for communications equipment isdefined. Of special note is §68.302, dealing with telecompower lines.

NT/DAS/PRL/003. Telephone instrument, subscriberequipment and line requirements are documented.

PTT 692.01. This is a general Swiss specification fortelecom exchange equipment.

REA PE-60. The Rural Electrification Administration(REA) has documented the predominant waveform forinduced lightning transients. This test is now commonlyknown as the 10/1000 μs pulse test.

REA PE-80. The REA defines requirements for gas tubesand similar devices for telecom applications.

TA-TSY-000974. This technical advisory by Bellcoredefines double exponential waveforms, which are the basisfor many telecom applications norms.

UL 1459 and UL 4978. These document detail tests forstandard telephone equipment and data transmission.

TA-TSY-000974. This technical advisory by Bellcoredefines double exponential waveforms, which are the basisfor many telecom applications norms.

UL 1459 and UL 4978. These document detail tests forstandard telephone equipment and data transmission.

© Semiconductor Components Industries, LLC, 2001

March, 2001 − Rev.082 Publication Order Number:

AN784/D

AN784/D

Transient Power Capabilityof Zener Diodes

Prepared byApplications Engineering andJerry Wilhardt, Product Engineer — Industrial and Hi-Rel Zener Diodes

INTRODUCTIONBecause of the sensitivity of semiconductor components

to voltage transients in excess of their ratings, circuits areoften designed to inhibit voltage surges in order to protectequipment from catastrophic failure. External voltagetransients are imposed on power lines as a result of lightningstrikes, motors, solenoids, relays or SCR switching circuits,which share the same ac source with other equipment.Internal transients can be generated within a piece ofequipment by rectifier reverse recovery transients,switching of loads or transformer primaries, fuse blowing,solenoids, etc. The basic relation, v = L di/dt, describes mostequipment developed transients.

ZENER DIODE CHARACTERISTICSZener diodes, being nearly ideal clippers (that is, they

exhibit close to an infinite impedance below the clippinglevel and close to a short circuit above the clipping level), areoften used to suppress transients. In this type of application,it is important to know the power capability of the zener forshort pulse durations, since they are intolerant of excessivestress.

Some ON Semiconductor data sheets such as the ones fordevices shown in Table 1 contain short pulse surgecapability. However, there are many data sheets that do notcontain this data and Figure 1 is presented here tosupplement this information.

Table 1. Transient Suppressor Diodes

SeriesNumbers

SteadyState Power Package Description

1N4728 1 W DO-41 Double Slug Glass

1N6267 5 W Case 41A Axial Lead Plastic

1N5333A 5 W Case 17 Surmetic 40

1N746/957A/4371

400 mW DO-35 Double Slug Glass

1N5221A 500 mW DO-35 Double Slug Glass

Some data sheets have surge information which differsslightly from the data shown in Figure 1. A variety of reasonsexist for this:

1. The surge data may be presented in terms of actual surgepower instead of nominal power.

2. Product improvements have occurred since the datasheet was published.

Figure 1. Peak Power Ratings of Zener Diodes

Power is defined as VZ(NOM) x IZ(PK) where VZ(NOM) is the nominalzener voltage measured at the low test current used for voltageclassification.

1N6267 SERIES

GLASS DO-35 & GLASS DO-41250 mW TO 1 W TYPES

5 WATT TYPES

PULSE WIDTH (ms)

0.1

100

0.01 0.02

PP

K(N

OM

), N

OM

INA

L P

EA

K P

OW

ER

(kW

) 50

20

10

5

2

1

0.5

0.2

0.1

0.05

0.020.01

0.05 0.2 0.5 1 2 5 10

1 TO 3 W TYPESPLASTIC DO-41

3. Larger dice are used, or special tests are imposed on theproduct to guarantee higher ratings than those shown onFigure 1.

4. The specifications may be based on a JEDECregistration or part number of another manufacturer.

The data of Figure 1 applies for non-repetitive conditionsand at a lead temperature of 25°C. If the duty cycle increases,the peak power must be reduced as indicated by the curvesof Figure 2. Average power must be derated as the lead orambient temperature rises above 25°C. The average powerderating curve normally given on data sheets may benormalized and used for this purpose.

At first glance the derating curves of Figure 2 appear to bein error as the 10 ms pulse has a higher derating factor thanthe 10 μs pulse. However, when the derating factor for agiven pulse of Figure 2 is multiplied by the peak power valueof Figure 1 for the same pulse, the results follow theexpected trend.

http://onsemi.com

APPLICATION NOTE

AN784/D

http://onsemi.com83

When it is necessary to use a zener close to surge ratings,and a standard part having guaranteed surge limits is notsuitable, a special part number may be created having a surgelimit as part of the specification. Contact your nearestON Semiconductor OEM sales office for capability, price,delivery, and minimum order criteria.

MATHEMATICAL MODEL

Figure 2. Typical Derating Factor for Duty Cycle

0.1

1

0.7

0.5

0.3

0.2

0.02

0.1

0.07

0.05

0.03

0.010.2 0.5 1 52 10 20 50 100

PULSE WIDTH10 ms

1 ms

100 μs

10 μs

D, DUTY CYCLE (%)

DE

RAT

ING

FA

CT

OR

Since the power shown on the curves is not the actualtransient power measured, but is the product of the peakcurrent measured and the nominal zener voltage measuredat the current used for voltage classification, the peak currentcan be calculated from:

IZ(PK) =P(PK)

VZ(NOM)(1)

The peak voltage at peak current can be calculated from:

(2)VZ(PK) = FC × VZ(NOM)

where FC is the clamping factor. The clamping factor isapproximately 1.20 for all zener diodes when operated attheir pulse power limits. For example, a 5 watt, 20 volt zenercan be expected to show a peak voltage of 24 volts regardlessof whether it is handling 450 watts for 0.1 ms or 50 watts for10 ms. This occurs because the voltage is a function ofjunction temperature and IR drop. Heating of the junction ismore severe at the longer pulse width, causing a highervoltage component due to temperature which is roughlyoffset by the smaller IR voltage component.

For modeling purposes, an approximation of the zenerresistance is needed. It is obtained from:

RZ(NOM) =VZ(NOM)(FC-1)

PPK(NOM)/VZ(NOM)(3)

The value is approximate because both the clampingfactor and the actual resistance are a function of temperature.

CIRCUIT CONSIDERATIONSIt is important that as much impedance as circuit

constraints allow be placed in series with the zener diode andthe components to be protected. The result will be a lowerclipping voltage and less zener stress. A capacitor in parallelwith the zener is also effective in reducing the stress imposedby very short duration transients.

To illustrate use of the data, a common application will beanalyzed. The transistor in Figure 3 drives a 50 mH solenoidwhich requires 5 amperes of current. Without some meansof clamping the voltage from the inductor when thetransistor turns off, it could be destroyed.

Figure 3. Circuit Example

Used to select a zener diode having the proper voltageand power capability to protect the transistor.

10 ms

2 s

26 Vdc50 mH, 5 Ω

The means most often used to solve the problem is toconnect an ordinary rectifier diode across the coil; however,this technique may keep the current circulating through thecoil for too long a time. Faster switching is achieved byallowing the voltage to rise to a level above the supply beforebeing clamped. The voltage rating of the transistor is 60 V,indicating that approximately a 50 volt zener will berequired.

The peak current will equal the on-state transistor current(5 amperes) and will decay exponentially as determined bythe coil L/R time constant (neglecting the zener impedance).A rectangular pulse of width L/R (0.01 sec) and amplitudeof IPK (5 A) contains the same energy and may be used toselect a zener diode. The nominal zener power ratingtherefore must exceed (5 A × 50) = 250 watts at 10 ms anda duty cycle of 0.01/2 = 0.5%. From Figure 2, the duty cyclefactor is 0.62 making the single pulse power rating requiredequal to 250/0.62 = 403 watts. From Figure 1, one of the1N6267 series zeners has the required capability. The1N6287 is specified nominally at 47 volts and should provesatisfactory.

Although this series has specified maximum voltagelimits, equation 3 will be used to determine the maximumzener voltage in order to demonstrate its use.

RZ =47(1.20 − 1)

500/47

9.4

10.64= = 0.9Ω

At 5 amperes, the peak voltage will be 4.5 volts abovenominal or 51.5 volts total which is safely below the 60 volttransistor rating.


March, 2001 − Rev.084 Publication Order Number:

AN843/D

AN843/D

A Review of Transients andTheir Means of Suppression

Prepared bySteve CherniakApplications Engineering

INTRODUCTIONOne problem that most, if not all electronic equipment

designers must deal with, is transient overvoltages.Transients in electrical circuits result from the suddenrelease of previously stored energy. Some transients may bevoluntary and created in the circuit due to inductiveswitching, commutation voltage spikes, etc. and may beeasily suppressed since their energy content is known andpredictable. Other transients may be created outside thecircuit and then coupled into it. These can be caused bylightning, substation problems, or other such phenomena.These transients, unlike switching transients, are beyond thecontrol of the circuit designer and are more difficult toidentify, measure and suppress.

Effective transient suppression requires that the impulseenergy is dissipated in the added suppressor at a low enoughvoltage so the capabilities of the circuit or device will not beexceeded.

REOCCURRING TRANSIENTSTransients may be formed from energy stored in circuit

inductance and capacitance when electrical conditions in thecircuit are abruptly changed.

Switching induced transients are a good example of this;

the change in current �didt� in an inductor (L) will generate

a voltage equal to L didt. The energy (J) in the transient is equal

to 1/2Li2 and usually exists as a high power impulse for arelatively short time (J = Pt).

If load 2 is shorted (Figure 1), devices parallel to it maybe destroyed. When the fuse opens and interrupts the faultcurrent, the slightly inductive power supply produces a

transient voltage spike of V � L didt with an energy content of

J = 1/2Li2. This transient might be beyond the voltagelimitations of the rectifiers and/or load 1. Switching out ahigh current load will have a similar effect.

TRANSFORMER PRIMARYBEING ENERGIZED

If a transformer is energized at the peak of the line voltage(Figure 2), this voltage step function can couple to the straycapacitance and inductance of the secondary winding andgenerate an oscillating transient voltage whose oscillationsdepend on circuit inductance and capacitance. Thistransient’s peak voltage can be up to twice the peakamplitude of the normal secondary voltage.

In addition to the above phenomena the capacitivelycoupled (CS) voltage spike has no direct relationship withthe turns ratio, so it is possible for the secondary circuit to seethe peak applied primary voltage.

Figure 1. Load Dump with Inductive Power Supply

POWER

SUPPLYA

BLOAD1

LOAD2

+

−

SHORTACROSSLOAD 2

FUSE

VAB

+

0

−

http://onsemi.com

APPLICATION NOTE

AN843/D

http://onsemi.com85

Figure 2. Situation Where Transformer Capacitance Causes a Transient

+

−

+

−

VLine

PEAK SHOULDBE 30% LIGHTER

CS

SWITCHCLOSEDVAB

VAB

LOADMAY HAVE STRAYINDUCTANCE ORCAPACITANCE

A

B

SWITCH

TRANSFORMER PRIMARYBEING DE-ENERGIZED

If the transformer is driving a high impedance load,transients of more than ten times normal voltage can becreated at the secondary when the primary circuit of thetransformer is opened during zero-voltage crossing of the acline. This is due to the interruption of the transformermagnetizing current which causes a rapid collapse of themagnetic flux in the core. This, in turn, causes a high voltagetransient to be coupled into the transformer’s secondarywinding (Figure 3).

Transients produced by interrupting transformersmagnetizing current can be severe. These transients candestroy a rectifier diode or filter capacitor if a lowimpedance discharge path is not provided.

SWITCH “ARCING”When a contact type switch opens and tries to interrupt

current in an inductive circuit, the inductance tries to keepcurrent flowing by charging stray capacitances. (SeeFigure 4.)

Figure 3. Typical Situation Showing Possible Transient When InterruptingTransformer Magnetizing Current

Figure 4. Transients Caused by Switch Opening

SWITCH

LOAD

AC

LINE

Im

SWITCH

OPENED

LINE

VOLTAGE

MAGNETIZING

CURRENT AND

FLUX

SECONDARY

VOLTAGE

VCap

VLine

TRANSIENT

SENSITIVE

LOAD

LINE

VOLTAGE

AN843/D

http://onsemi.com86

This can also happen when the switch contacts bounceopen after its initial closing. When the switch is opened (orbounces open momentarily) the current that the inductancewants to keep flowing will oscillate between the straycapacitance and the inductance. When the voltage due to theoscillation rises at the contacts, breakdown of the contactgap is possible, since the switch opens (or bounces open)relatively slowly compared to the oscillation frequency, andthe distance may be small enough to permit “arcing.” Thearc will discontinue at the zero current point of theoscillation, but as the oscillatory voltage builds up again andthe contacts move further apart, each arc will occur at ahigher voltage until the contacts are far enough apart tointerrupt the current.

WAVESHAPES OF SURGE VOLTAGESIndoor Waveshapes

Measurements in the field, laboratory, and theoreticalcalculations indicate that the majority of surge voltages inindoor low-voltage power systems have an oscillatorywaveshape. This is because the voltage surge excites thenatural resonant frequency of the indoor wiring system. Inaddition to being typically oscillatory, the surges can alsohave different amplitudes and waveshapes in the variousplaces of the wiring system. The resonant frequency canrange from about 5 kHz to over 500 kHz. A 100 kHzfrequency is a realistic value for a typical surge voltage formost residential and light industrial ac wire systems.

The waveshape shown in Figure 5 is known as an “0.5 μs− 100 kHz ring wave.” This waveshape is reasonablyrepresentative of indoor low-voltage (120 V − 240 V) wiringsystem transients based on measurements conducted byseveral independent organizations. The waveshape isdefined as rising from 10% to 90% of its final amplitude in0.5 μs, then decays while oscillating at 100 kHz, each peakbeing 60% of the preceding one.

The fast rise portion of the waveform can induce theeffects associated with non-linear voltage distribution inwindings or cause dv/dt problems in semiconductors.Shorter rise times can be found in transients but they arelengthened as they propagate into the wiring system orreflected from wiring discontinuities.

Figure 5. 0.5 μs 100 kHz Ring Wave

0.9 Vpk

Vpk

t = 10 μs (f = 100 kHz)

0.1 Vpk

0.5 μs

60% OF Vpk

The oscillating portion of the waveform produces voltagepolarity reversal effects. Some semiconductors are sensitiveto polarity changes or can be damaged when forced into orout of conduction (i.e. reverse recovery of rectifier devices).The sensitivity of some semiconductors to the timing andpolarity of a surge is one of the reasons for selecting thisoscillatory waveform to represent actual conditions.

Outdoor Locations

Both oscillating and unidirectional transients have beenrecorded in outdoor environments (service entrances andother places nearby). In these locations substantial energy isstill available in the transient, so the waveform used tomodel transient conditions outside buildings must containgreater energy than one used to model indoor transientsurges.

Properly selected surge suppressors have a goodreputation of successful performance when chosen inconjunction with the waveforms described in Figure 6. Therecommended waveshape of 1.2 × 50 μs (1.2 μs is associatedwith the transients rise time and the 50 μs is the time it takesfor the voltage to drop to 1/2Vpk) for the open circuit voltageand 8 × 20 μs for the short circuit current are as defined inIEEE standard 28-ANSI Standard C62.1 and can beconsidered a realistic representation of an outdoor transientswaveshape.

Figure 6. Unidirectional Wave Shapes

V

0.9 Vpk

0.3 Vpk

0.1 Vpk

Vpk

0.5 Vpk

50 μs

t1 t1 × 1.67 = 1.2 μs

IIpk

0.9 Ipk

0.1 Ipk

0.5 Ipk

t220 μs

t2 × 1.25 = 8 μs

(a) Open-Circuit Voltage Waveform (b) Discharge Current Waveform

AN843/D

http://onsemi.com87

The type of device under test determines whichwaveshape in Figure 6 is more appropriate. The voltagewaveform is normally used for insulation voltage withstandtests and the current waveform is usually used for dischargecurrent tests.

RANDOM TRANSIENTSThe source powering the circuit or system is frequently

the cause of transient induced problems or failures. Thesetransients are difficult to deal with due to their nature; theyare totally random and it is difficult to define theiramplitude, duration and energy content. These transients aregenerally caused by switching parallel loads on the same

branch of a power distribution system and can also be causedby lightning.

AC POWER LINE TRANSIENTSTransients on the ac power line range from just above

normal voltage to several kV. The rate of occurrence oftransients varies widely from one branch of a powerdistribution system to the next, although low-level transientsoccur more often than high-level surges.

Data from surge counters and other sources is the basis forthe curves shown in Figure 7. This data was taken fromunprotected (no voltage limiting devices) circuits meaningthat the transient voltage is limited only by the sparkoverdistance of the wires in the distribution system.

Figure 7. Peak Surge Voltage versus Surges per Year*

*EIA paper, P587.1/F, May, 1979, Page 10

20

10

5

4

0.7

0.2

0.1

0.5

0.3

9

8

7

6

1

3

2

0.90.8

0.4

0.6

0.01 10.1 10 100 1000

SURGES PER YEAR

PEAK

SURGE

VOLTAGE

(kV)

HIGH EXPOSURE

MEDIUM EXPOSURE

LOW EXPOSURE

AN843/D

http://onsemi.com88

The low exposure portion of the set of curves is datacollected from systems with little load-switching activitythat are located in areas of light lightning activity.

Medium exposure systems are in areas of frequentlightning activity with a severe switching transient problem.

High exposure systems are rare systems supplied by longoverhead lines which supply installations that have highsparkover clearances and may be subject to reflections atpower line ends.

When using Figure 7 it is helpful to remember that peaktransient voltages will be limited to approximately 6 kV inindoor locations due to the spacing between conductorsusing standard wiring practices.

TRANSIENT ENERGY LEVELS ANDSOURCE IMPEDANCE

The energy contained in a transient will be divided betweenthe transient suppressor and the source impedance of thetransient in a way that is determined by the two impedances.With a spark-gap type suppressor, the low impedance of theArc after breakdown forces most of the transient’s energy tobe dissipated elsewhere, e.g. in a current limiting resistor inseries with the spark-gap and/or in the transient’s sourceimpedance. Voltage clamping suppressors (e.g. zeners,mov’s, rectifiers operating in the breakdown region) on theother hand absorb a large portion of the transient’s surgeenergy. So it is necessary that a realistic assumption of thetransient’s source impedance be made in order to be able toselect a device with an adequate surge capability.

The 100 kHz “Ring Wave” shown in Figure 5 is intendedto represent a transient’s waveshape across an open circuit.The waveshape will change when a load is connected and theamount of change will depend on the transient’s sourceimpedance. The surge suppressor must be able to withstandthe current passed through it from the surge source. Anassumption of too high a surge impedance (when testing thesuppressor) will not subject the device under test tosufficient stresses, while an assumption of too low a surgeimpedance may subject it to an unrealistically large stress;there is a trade-off between the size (cost) of the suppressorand the amount of protection obtained.

In a building, the transient’s source impedance increaseswith the distance from the electrical service entrance, but

open circuit voltages do not change very much throughoutthe structure since the wiring does not provide muchattenuation. There are three categories of service locationsthat can represent the majority of locations from theelectrical service entrance to the most remote wall outlet.These are listed below. Table 1 is intended as an aid for thepreliminary selection of surge suppression devices, since itis very difficult to select a specific value of sourceimpedance.

Category I: Outlets and circuits a “long distance” fromelectrical service entrance. Outlets more than 10 metersfrom Category II or more than 20 meters from CategoryIII (wire gauge #14 − #10)

Category II: Major bus lines and circuits a “short distance”from electrical service entrance. Bus system in industrialplants. Outlets for heavy duty appliances that are “close”to the service entrance.

Distribution panel devices.Commercial building lighting systems.

Category III. Electrical service entrance and outdoorlocations.

Power line between pole and electrical service entrance.Power line between distribution panel and meter.Power line connection to additional near-by buildings.Underground power lines leading to pumps, filters, etc.

Categories I and II in Table 1 correspond to the extremerange of the “medium exposure” curve in Figure 7. Thesurge voltage is limited to approximately 6 kV due to thesparkover spacing of indoor wiring.

The discharge currents of Category II were obtained fromsimulated lightning tests and field experience withsuppressor performance.

The surge currents in Category I are less than in CategoryII because of the increase in surge impedance due to the factthat Category I is further away from the service entrance.

Category III can be compared to the “High Exposure”situation in Figure 7. The limiting effect of sparkover is notavailable here so the transient voltage can be quite large.

Table 1. Surge Voltages and Currents Deemed to Represent the Indoor Environment Depending Upon LocationEnergy (Joules) Dissipated in a

Suppressor with a Clamping Voltage of (3)

Category Waveform Surge Voltage(1) Surge Current(2) 250 V 500 V 1000 V

I 0.5 μs 100 kHzRing Wave

6 kV 200 A 0.4 0.8 1.6

II 0.5 μs 100 kHzRing Wave

6 kV 500 A 1 2 4

1.2 × 50 μs8 × 20 μs

6 kV3 kA 20 40 80

III 1.2 × 50 μs8 × 20 μs

10 kV or more10 kA or more

Notes: 1. Open Circuit voltageNotes: 2. Discharge current of the surge (not the short circuit current of the power system)Notes: 3. The energy a suppressor will dissipate varies in proportion with the suppressor’s clamping voltage, which can be different with different system voltages (assuming the sameNotes: 3. discharge current).

AN843/D

http://onsemi.com89

LIGHTNING TRANSIENTSThere are several mechanisms in which lightning can

produce surge voltages on power distribution lines. One ofthem is a direct lightning strike to a primary (before thesubstation) circuit. When this high current, that is injectedinto the power line, flows through ground resistance and thesurge impedance of the conductors, very large transientvoltages will be produced. If the lightning misses theprimary power line but hits a nearby object the lightningdischarge may also induce large voltage transients on theline. When a primary circuit surge arrester operates andlimits the primary voltage the rapid dv/dt produced willeffectively couple transients to the secondary circuit throughthe capacitance of the transformer (substation) windings inaddition to those coupled into the secondary circuit bynormal transformer action. If lightning struck the secondarycircuit directly, very high currents may be involved whichwould exceed the capability of conventional surgesuppressors. Lightning ground current flow resulting fromnearby direct to ground discharges can couple onto thecommon ground impedance paths of the groundingnetworks also causing transients.

AUTOMOTIVE TRANSIENTSTransients in the automotive environment can range from

the noise generated by the ignition system and the variousaccessories (radio, power window, etc.) to the potentiallydestructive high energy transients caused by the charging(alternator/regulator) system. The automotive “Load

Dump” can cause the most destructive transients; it is whenthe battery becomes disconnected from the charging systemduring high charging rates. This is not unlikely when oneconsiders bad battery connections due to corrosion or otherwiring problems. Other problems can exist such as steadystate overvoltages caused by regulator failure or 24 Vbattery jump starts. There is even the possibility of incorrectbattery connection (reverse polarity).

Capacitive and/or inductive coupling in wire harnesses aswell as conductive coupling (common ground) can transmitthese transients to the inputs and outputs of automotiveelectronics.

The Society of Automotive Engineers (SAE) documenteda table describing automotive transients (see Table 2) whichis useful when trying to provide transient protection.

Considerable variation has been observed while gatheringdata on automobile transients. All automobiles have theirelectrical systems set up differently and it is not the intent ofthis paper to suggest a protection level that is required. Therewill always be a trade-off between cost of the suppressor andthe level of protection obtained. The concept of one mastersuppressor placed on the main power lines is the mostcost-effective scheme possible since individual suppressorsat the various electronic devices will each have to suppressthe largest transient that is likely to appear (Load Dump),hence each individual suppressor would have to be the samesize as the one master suppressor since it is unlikely forseveral suppressors to share the transient discharge.

Table 2. Typical Transients Encountered in the Automotive Environment

Length ofTransient Cause

Energy Capability

Voltage AmplitudePossible Frequency

of Application

Steady State Failed Voltage Regulator

Booster starts with 24 V battery

Load Dump — i.e., disconnection of battery duringhigh charging rates

Inductive Load Switching Transient

Alternator Field Decay

Ignition PulseDisconnected Battery

Mutual Coupling in Harness

Ignition Pulse Normal

Accessory Noise

Transceiver Feedback

5 Minutes

4.5−100 ms

≤ 0.32 s

≤ 0.2 s

90 ms

1 ms

15 μs

∞

+18 VInfrequent

∞

±24 V

≥ 10 J

≤ 125 V

Infrequent

Infrequent

< 1 J

−300 V to +80 VOften

< 1 J

−100 V to −40 VEach Turn-Off

≤ 500 HzSeveral Times invehicle life

< 0.5 J

≤ 75 V

< 1 J

≤ 200 V

< 0.001 J

3 V

≤ 1.5 V

≈ 20 mV

Often

3 500 HzContinuous

50 Hz to 10 kHz

R.F.

AN843/D

http://onsemi.com90

There will, of course, be instances where a need forindividual suppressors at the individual accessories will berequired, depending on the particular wiring system orsituation.

TRANSIENT SUPPRESSOR TYPESCarbon Block Spark Gap

This is the oldest and most commonly used transientsuppressor in power distribution and telecommunicationsystems. The device consists of two carbon block electrodesseparated by an air gap, usually 3 to 4 mils apart. Oneelectrode is connected to the system ground and the other tothe signal cable conductor. When a transient over-voltageappears, its energy is dissipated in the arc that forms betweenthe two electrodes, a resistor in series with the gap, and alsoin the transient’s source impedance, which depends onconductor length, material and other parameters.

The carbon block gap is a fairly inexpensive suppressorbut it has some serious problems. One is that it has arelatively short service life and the other is that there arelarge variations in its arcing voltage. This is the majorproblem since a nominal 3 mil gap will arc anywhere from300 to 1000 volts. This arcing voltage variation limits its usemainly to primary transient suppression with more accuratesuppressors to keep transient voltages below an acceptablelevel.

Gas Tubes

The gas tube is another common transient suppressor,especially in telecommunication systems. It is made of twometallic conductors usually separated by 10 to 15 milsencapsulated in a glass envelope which is filled with severalgases at low pressure. Gas tubes have a higher currentcarrying capability and longer life than carbon block gaps.The possibility of seal leakage and the resultant of lossprotection has limited the use of these devices.

Selenium Rectifiers

Selenium transient suppressors are selenium rectifiersused in the reverse breakdown mode to clamp voltagetransients. Some of these devices have self-healingproperties which allows the device to survive energydischarges greater than their maximum capability for alimited number of surges. Selenium rectifiers do not havethe voltage clamping capability of zener diodes. This iscausing their usage to become more and more limited.

METAL OXIDE VARISTORS (MOV’S)An MOV is a non-linear resistor which is voltage

dependent and has electrical characteristics similar toback-to-back zener diodes. As its name implies it is made upof metal oxides, mostly zinc oxide with other oxides addedto control electrical characteristics. MOV characteristics arecompared to back-to-back zeners in Photos 2 through 7.

When constructing MOV’s the metal oxides are sinteredat high temperatures to produce a polycrystalline structureof conductive zinc oxide separated by highly resistive

intergranular boundaries. These boundaries are the source ofthe MOV’s non-linear electrical behavior.

MOV electrical characteristics are mainly controlled bythe physical dimensions of the polycrystalline structuresince conduction occurs between the zinc oxide grains andthe intergranular boundaries which are distributedthroughout the bulk of the device.

The MOV polycrystalline body is usually formed into theshape of a disc. The energy rating is determined by thedevice’s volume, voltage rating by its thickness, and currenthandling capability by its area. Since the energy dissipatedin the device is spread throughout its entire metal oxidevolume, MOV’s are well suited for single shot high powertransient suppression applications where good clampingcapability is not required.

The major disadvantages with using MOV’s are that theycan only dissipate relatively small amounts of averagepower and are not suitable for many repetitive applications.Another drawback with MOV’s is that their voltageclamping capability is not as good as zeners, and isinsufficient in many applications.

Perhaps the major difficulty with MOV’s is that they havea limited life time even when used below their maximumratings. For example, a particular MOV with a peak currenthandling capability of 1000 A has a lifetime of about 1 surgeat 1000 Apk, 100 surges at 100 Apk and approximately 1000surges at 65 Apk.

TRANSIENT SUPPRESSIONUSING ZENERS

Zener diodes exhibit a very high impedance below thezener voltage (VZ), and a very low impedance above VZ.Because of these excellent clipping characteristics, the zenerdiode is often used to suppress transients. Zeners areintolerant of excessive stress so it is important to know thepower handling capability for short pulse durations.

Most zeners handle less than their rated power duringnormal applications and are designed to operate mosteffectively at this low level. Zener transient suppressorssuch as the ON Semiconductor 1N6267 Mosorb series aredesigned to take large, short duration power pulses.

This is accomplished by enlarging the chip and theeffective junction area to withstand the high energy surges.The package size is usually kept as small as possible toprovide space efficiency in the circuit layout, and since thepackage does not differ greatly from other standard zenerpackages, the steady state power dissipation does not differgreatly.

Some data sheets contain information on short pulse surgecapability. When this information is not available for ONSemiconductor devices, Figure 8 can be used. This dataapplies for non-repetitive conditions with a lead temperatureof 25°C.

It is necessary to determine the pulse width and peakpower of the transient being suppressed when usingFigure 8. This can be done by taking whatever waveform

AN843/D

http://onsemi.com91

the transient is and approximating it with a rectangular pulsewith the same peak power. For example, an exponentialdischarge with a 1 ms time constant can be approximated bya rectangular pulse 1 ms wide that has the same peak poweras the transient. This would be a better approximation thana rectangular pulse 10 ms wide with a correspondingly loweramplitude. This is because the heating effects of differentpulse width lengths affect the power handling capability, ascan be seen by Figure 8. This also represents a conservativeapproach because the exponential discharge will contain ≈1/2 the energy of a rectangular pulse with the same pulsewidth and amplitude.


1N6267 SERIES


5 WATT TYPES

PULSE WIDTH (ms)

0.1

100

0.01 0.02

PP

K(N

OM

), N

OM

INA

L P

EA

K P

OW

ER

(kW

)

50

20

10

5

2

1

0.5

0.2

0.1

0.05

0.020.01

0.05 0.2 0.5 1 2 5 10


When used in repetitive applications, the peak power mustbe reduced as indicated by the curves of Figure 9. Averagepower must be derated as the lead or ambient temperatureexceeds 25°C. The power derating curve normally given ondata sheets can be normalized and used for this purpose.


0.1

1

0.7

0.5

0.3

0.2

0.02

0.1

0.07

0.05

0.03

0.010.2 0.5 1 52 10 20 50

PULSE WIDTH10 ms

1 ms

100 μs

10 μs

D, DUTY CYCLE (%)

DE

RAT

ING

FA

CT

OR

The peak zener voltage during the peak current of thetransient being suppressed can be related to the nominalzener voltage (Eqtn 1) by the clamping factor (FC).

Eqtn 1: VZ(pK) = FC (VZ(nom))

Unless otherwise specified FC is approximately 1.20 forzener diodes when operated at their pulse power limits.

For example, a 5 watt, 20 volt zener can be expected toshow a peak voltage of 24 volts regardless of whether it ishandling 450 watts for 0.1 ms or 50 watts for 10 ms. (SeeFigure 8.)

This occurs because the zener voltage is a function of bothjunction temperature and IR drop. Longer pulse widthscause a greater junction temperature rise than short ones; theincrease in junction temperature slightly increases the zenervoltage. This increase in zener voltage due to heating isroughly offset by the fact that longer pulse widths ofidentical energy content have lower peak currents. Thisresults in a lower IR drop (zener voltage drop) keeping theclamping factor relatively constant with various pulsewidths of identical energy content.

An approximation of zener impedance is also helpful inthe design of transient protection circuits. The value ofRZ(nom) (Eqtn 2) is approximate because both the clampingfactor and the actual resistance is a function of temperature.

Eqtn 2: RZ(nom) =V2

Z(nom) (FC −1)

PpK(nom)

VZ(nom) = Nominal Zener VoltagePpK(nom) = Found from Figure 8 when device type andpulse width are known. For example, from Figure 8 a1N6267 zener suppressor has a PpK(nom) of 1.5 kW at apulse width of 1 ms.As seen from equation 2, zeners with a larger PpK(nom)

capability will have a lower RZ(nom).

ZENER versus MOV TRADEOFFSThe clamping characteristics of Zeners and MOV’s are

best compared by measuring their voltages under transientconditions. Photos 1 through 9 are the result of anexperiment that was done to compare the clampingcharacteristics of a Zener (ON Semiconductor 1N6281,approximately 1.5J capability) with those of an MOV (G.E.V27ZA4, 4J capability); both are 27 V devices.

Photo 1 shows the pulse generator output voltage. Thisgenerator synthesizes a transient pulse that is characteristicof those that may appear in the real world.

Photos 2 and 3 are clamping voltages of the MOV andZener, respectively with a surge source impedance of500 Ω .

Photos 4 and 5 are the clamping voltages with a surgesource impedance of 50 Ω� .

Photos 6 and 7 simulate a condition where the surgesource impedance is 5 Ω� .

Photos 8 and 9 show a surge source impedance of 0.55 Ω,which is at the limits of the Zener suppressor’s capability.

AN843/D

http://onsemi.com92

PHOTO 1Open Circuit Transient PulseVert: 20 V/divHoriz: 0.5 ms/divVpeak = 90 V

PHOTO 2MOV (27 V)Vert: 10 V/divHoriz: 0.5 ms/divTransient Source Impedance: 500 ΩVpeak = 39.9 V

0%

10090

10

0%

10090

10

AN843/D

http://onsemi.com93

PHOTO 3Zener (27 V)Vert: 10 V/divHoriz: 0.5 ms/divTransient Source Impedance: 500 ΩVpeak = 27 V

PHOTO 4MOV (27 V)Vert: 10 V/divHoriz: 0.5 ms/divTransient Source Impedance: 50 ΩVpeak = 44.7 V


0%

10090

10

0%

10090

10

0%

10090

10

AN843/D

http://onsemi.com94

PHOTO 6MOV (27 V)Vert: 10 V/divHoriz: 0.5 ms/divTransient Source Impedance: 5 ΩVpeak = 52 V


PHOTO 8MOV (27 V)Vert: 10 V/divHoriz: 0.5 ms/divTransient Source Impedance: 0.55 ΩVpeak = 62.5 V

0%

10090

10

0%

10090

10

AN843/D

http://onsemi.com95

PHOTO 9Zener (27 V)Vert: 10 V/divHoriz: 0.5 ms/divTransient Source Impedance: 0.55 ΩVpeak: 30.2 VPeak Power: Approx 2000 Wpeak(The limit of this device’s capability)

As can be seen by the photographs, the Zener suppressorhas significantly better voltage clamping characteristicsthan the MOV even though that particular Zener has lessthan one-fourth the energy capability of the MOV it wascompared with. However, the energy rating can bemisleading because it is based on the clamp voltage times thesurge current, and when using an MOV, the high impedanceresults in a fairly high clamp voltage. The major tradeoffwith using a zener type suppressor is its cost versus powerhandling capability, but since it would take an “oversized”MOV to clamp voltages (suppress transients) as well as thezener, the MOV begins to lose its cost advantage.

If a transient should come along that exceeds thecapabilities of the particular Zener, or MOV, suppressor thatwas chosen, the load will still be protected, since they bothfail short.

The theoretical reaction time for Zeners is in thepicosecond range, but this is slowed down somewhat withlead and package inductance. The 1N6267 Mosorb series oftransient suppressors have a typical response time of lessthan one nanosecond. For very fast rising transients it isimportant to minimize external inductances (due to wiring,etc.) which will minimize overshoot.

Connecting Zeners in a back-to-back arrangement willenable bidirectional voltage clamping characteristics. (SeeFigure 10.)

If Zeners A and B are the same voltage, a transient ofeither polarity will be clamped at approximately that voltagesince one Zener will be in reverse bias mode while the otherwill be in the forward bias mode. When clamping lowvoltage it may be necessary to consider the forward drop ofthe forward biased Zener.

The typical protection circuit is shown in Figure 11a. Inalmost every application, the transient suppression device isplaced in parallel with the load, or component to beprotected. Since the main purpose of the circuit is to clampthe voltage appearing across the load, the suppressor should

be placed as close to the load as possible to minimizeovershoot due to wiring (or any inductive) effect. (SeeFigure 11b.)

Figure 10. Zener Arrangement forBidirectional Clamping

Figure 11a. Using Zener to Protect LoadAgainst Transients

ÎÎÎÎÎÎ

OR

B

A

Vin

Zin

B

A

LOAD VL

Figure 11b. Overshoot Due to Inductive Effect

ZENER

VOLTAGE

PEAK VOLTAGE

DUE TO OVERSHOOT

TRANSIENT

INPUT

AN843/D

http://onsemi.com96

Zener capacitance prior to breakdown is quite small (forexample, the 1N6281 27 Volt Mosorb has a typicalcapacitance of 800 pF). Capacitance this small is desirablein the off-state since it will not attenuate wide-band signals.

When the Zener is in the breakdown mode of operation(e.g. when suppressing a transient) its effective capacitanceincreases drastically from what it was in the off-state. Thismakes the Zener ideal for parallel protection schemes since,during transient suppression, its large effective capacitancewill tend to hold the voltage across the protected elementconstant; while in the off-state (normal conditions, notransient present), its low off-state capacitance will notattenuate high frequency signals.

Input impedance (Zin) always exists due to wiring andtransient source impedance, but Zin should be increased asmuch as possible with an external resistor, if circuitconstraints allow. This will minimize Zener stress.

CONCLUSIONThe reliable use of semiconductor devices requires that

the circuit designer consider the possibility of transientovervoltages destroying these transient-sensitivecomponents.

These transients may be generated by normal circuitoperations such as inductive switching circuits, energizingand deenergizing transformer primaries, etc. They do notpresent much of a problem since their energy content,duration and effect may easily be obtained and dealt with.

Random transients found on power lines, or lightningtransients, present a greater threat to electronic componentssince there is no way to be sure when or how severe they willbe. General guidelines were discussed to aid the circuitdesigner in deciding what size (capability and cost)suppressor to choose for a certain level of protection. Therewill always be a tradeoff between suppressor price andprotection obtained.

Several different suppression devices were discussed withemphasis on Zeners and MOV’s, since these are the mostpopular devices to use in most applications.

REFERENCES1. GE Transient Voltage Suppression Manual, 2nd

edition.

2. ON Semiconductor Zener Diode Manual.

http://onsemi.com97

DESIGN CONSIDERATIONS AND PERFORMANCEOF ON SEMICONDUCTOR TEMPERATURE-COMPENSATED

ZENER (REFERENCE) DIODES

Prepared byZener Diode EngineeringandRonald N. RacinoReliability and Quality Assurance

INTRODUCTION

This application note defines ON Semiconductortemperature-compensated zener (reference) diodes,explains the device characteristics, describes electricaltesting, and discusses the advanced concepts of devicereliability and quality assurance. It is a valuable aid to thosewho contemplate designing circuits requiring the use ofthese devices.

Zener diodes fall into three general classifications:Regulator diodes, reference diodes and transient voltagesuppressors. Regulator diodes are normally employed inpower supplies where a nearly constant dc output voltage isrequired despite relatively large changes in input voltage orload resistance. Such devices are available with a wide rangeof voltage and power ratings, making them suitable for awide variety of electronic equipments.

Regulator diodes, however, have one limitation: They aretemperature-sensitive. Therefore, in applications in whichthe output voltage must remain within narrow limits duringinput-voltage, load-current, and temperature changes, atemperature-compensated regulator diode, called areference diode, is required.

The reference diode is made possible by taking advantageof the differing thermal characteristics of forward- andreverse-biased silicon p-n junctions. A forward-biasedjunction has a negative temperature coefficient ofapproximately 2 mV/°C, while reverse-biased junctionshave positive temperature coefficients ranging from about2 mV/°C at 5.5 V to 6 mV/°C at 10 V. Therefore it ispossible, by judicious combination of forward- andreverse-biased junctions, to fabricate a device with a verylow overall temperature coefficient (Figure 1).

The principle of temperature compensation is furtherillustrated in Figure 2, which shows the voltage-currentcharacteristics at two temperature points (25 and 100°C) forboth a forward- and a reverse-biased junction. The diagramshows that, at the specified test current (IZT), the absolutevalue of voltage change (ΔV) for the temperature changebetween 25 and 100°C is the same for both junctions.Therefore, the total voltage across the combination of thesetwo junctions is also the same at these temperature points,since one ΔV is negative and the other is positive. However,the rate of voltage change with temperature over the

Figure 1. Temperature Compensationof a 6.2 Volt Reference Diode (1N821 Series)

DIO

DE

VO

LTA

GE

DR

OP

(V

)(R

EF

ER

EN

CE

D T

O I Z

T=

7.5

mA

)

6.4

6.2

6

5.8

5.6

5.4

0.8

0.6

0.4

0.2

0−75 −50 −25 0 +25 +50 +75 +100

TEMPERATURE (°C)

6.2 − VOLT REFERENCE DIODE(COMBINATION OF ZENER

AND FORWARD DICE)

ZENER DIE

FORWARD-BIASED COMPENSATING DIE

temperature range defined by these points is not necessarilythe same for both junctions, thus the temperaturecompensation may not be linear over the entire range.

Figure 2 also indicates that the voltage changes of the twojunctions are equal and opposite only at the specified testcurrent. For any other value of current, the temperaturecompensation may not be complete.

http://onsemi.com98

Figure 2. Temperature Compensation ofP-N Junctions

DIRECTION OF CURRENT FLOW

PACKAGE

OUTLINE25°C

100°C

VF

+

I

IZT

−ΔV

+ΔV

25°CVR I

100°C

IZT−

FORWARD-BIASED

P-N JUNCTION

REVERSE-BIASED

ZENER JUNCTION

IMPORTANT ELECTRICAL CHARACTERISTICSOF REFERENCE DIODES

The three most important characteristics of referencediodes are 1) reference voltage, 2) voltage-temperaturestability, and 3) voltage-time stability.

1. Reference Voltage. This characteristic is defined as thevoltage drop measured across the diode when the specifiedtest current passes through it in the zener direction. It is alsocalled the zener voltage (VZ, Figure 3). On the data sheets,the reference voltage is given as a nominal voltage for eachfamily of reference diodes.

The nominal voltages are normally specified to atolerance of ±5%, but devices with tighter tolerances, suchas ±2% and ±1%, are available on special order.

2. Voltage-Temperature Stability. The temperaturestability of zener voltage is sometimes expressed by meansof the temperature coefficient. This parameter is usuallydefined as the percent voltage change across the device perdegree centigrade. This method of indicating voltagestability accurately reflects the voltage deviation at the testtemperature extremes but not necessarily at other pointswithin the specified temperature range. This fact is due tovariations in the rate of voltage change with temperature forthe forward- and reverse-biased dice of the reference diode.Therefore, the temperature coefficient is given inON Semiconductor data sheets only as a quick reference,for designers who are accustomed to this method ofspecification.

A more meaningful way of defining temperature stabilityis the “box method.” This method, used byON Semiconductor, guarantees that the zener voltage willnot vary by more than a specified amount over a specifiedtemperature range at the indicated test current, as verified bytests at several temperatures within this range.

Some devices are accurately compensated over a widetemperature range (−55°C to 100°C), others over a narrowerrange (0 to 75°C). The wide-range devices are, as a rule,more expensive. Therefore, it would be economicallywasteful for the designer to specify devices with atemperature range much wider than actually required for thespecific device application.

During actual production of reference diodes, it is difficultto predict the compensation accuracy. In the interest ofmaximum economy, it is common practice to test all devicescoming off the production line, and to divide the productionlot into groups, each with a specified maximum ΔVZ. Eachgroup, then, is given a different device type number.

Figure 3. Typical Voltage − Current Characteristic of Reference Diodes

I F(m

A)

0.3

0.2

0.1

−6 −4 −2

−0.1

−0.2

−0.3

2 4 6 8 10 12 14 16

VF (V)

I R(m

A)

VR (V)

IZVZ

http://onsemi.com99

On the data sheet, the voltage-temperature characteristicsof the most widely used device types are illustrated in agraph similar to the one shown in Figure 4. The particularproduction line represented in this figure produces 6.2 voltdevices, but the line yields five different device typenumbers (1N821 through 1N829), each with a differenttemperature coefficient. The 1N829, for example, has amaximum voltage change of less than 5 mV over atemperature range of −55 to +100°C, while the 1N821 mayhave a voltage change of up to 96 mV over the sametemperature range.

Figure 4. Temperature Dependenceof Zener Voltage (1N821 Series)

ΔVZ = +31 mV

ΔVZ = −31 mV

IZT = 7.5 mA

1N821,A

1N823,A

1N825,A

1N829,A

1N827,A

1N827,A

1N825,A

1N823,A

1N821,A

100

75

50

25

0

−25

−50

−75

−100−55 0 50 100

ΔV

Z, M

AX

IMU

M V

OLT

AG

E C

HA

NG

E (

mV

)(R

efer

ence

d to

−55

°C)

In the past, design data and characteristic curves on data sheets forreference diodes have been somewhat limited: The devices havebeen characterized principally at the recommended operating point.ON Semiconductor has introduced a data sheet, providing devicedata previously not available, and showing limit curves that permitworst-case circuit design without the need for associated tests re-quired in conjunction with the conventional data sheets.

Graphs such as these permit the selection of thelowest-cost device that meets a particular requirement. Theyalso permit the designer to determine the maximum voltagechange of a particular reference diode for a relatively smallchange in temperature. This is done by drawing vertical linesfrom the desired temperature points at the abscissa of thegraph to intersect with each the positive- and negative-goingcurves of the particular device of interest. Horizontal linesare then drawn from these intersects to the ordinate of thegraph. The difference between the intersections of thesehorizontal lines with the ordinate yields the maximumvoltage change over the temperature increment. Forexample, for the 1N821, a change in ambient temperaturefrom 0 to 50°C results in a voltage change of no more thanabout ±31 mV.

The reason that the device reference voltage may changein either the negative or positive direction is that afterassembly, some of the devices within a lot may beovercompensated while others may be undercompensated.In any design, the “worst-case” condition must beconsidered. Therefore, in the above example, it can beassumed that the maximum voltage change will not exceed31 mV.

It should be understood, however, that the abovecalculations give the maximum possible voltage change forthe device type, and by no means the actual voltage changefor the individual unit.

3. Voltage-Time Stability. The voltage-time stability ofa reference diode is defined by the voltage change duringoperating time at the standard test current (IZT) and testtemperature (TA). In general, the voltage stability of areference diode is better than 100 ppm per 1000 hours ofoperation.

Figure 5. Current Dependence of Zener Voltageat Various Temperatures

(1N821 Series)

I Z, Z

EN

ER

CU

RR

EN

T (m

A)

10

9

8

7.5

7

6

5

4−75 −50 −25 0 25 50

+100°C

IZT

+25°C

−55°C

+25°C

+100°C−55°C

ΔVZ, MAXIMUM VOLTAGE CHANGE (mV)

(Referenced to IZT = 7.5 mA)

THE EFFECT OF CURRENT VARIATION ONZENER VOLTAGE

The nominal zener voltage of a reference diode isspecified at a particular value of current, called the zener testcurrent (IZT). All measurements of voltage change withtemperature are referenced to this test current. If theoperating current is varied, all these specifications willchange.

The effect of current variation on zener voltage, at varioustemperatures, is graphically illustrated on the 1N821 datasheet as “Zener Current versus Maximum Voltage Change.”A typical example of such a graph is shown for the 1N821series in Figure 5. The voltage change shown is due entirelyto the impedance of the device at the fixed temperature. Itdoes not reflect the change in reference voltage due to thechange in temperature since each curve is referenced to IZT= 7.5 mA at the indicated temperature. As shown, the

http://onsemi.com100

greatest voltage change occurs at the highest temperaturerepresented in the diagram. (See “Dynamic Impedance”under the next section).

Figure 5 shows that, at 25°C, a change in zener currentfrom 4 to 10 mA causes a voltage shift of about 90 mV.Comparing this value with the voltage-change example inFigure 4 (31 mV), it is apparent that, in general, a greatervoltage variation may be due to current fluctuations than totemperature change. Therefore, good current regulation ofthe source should be a major consideration when usingreference diodes in critical applications.

It is not essential, however, that a reference diode beoperated at the specified test current. The newvoltage-temperature characteristics for a change in currentcan be obtained by superimposing the data of Figure 5 onthat of Figure 4. A new set of characteristics, at a test currentof 4 mA, is shown for the 1N823 in Figure 6, together withthe original characteristics at 7.5 mA.

Figure 6. Voltage Change with Temperaturefor 1N823 at Two Different Current Levels

+100

+50

0

−50

−100

−150

−200−50 0 50 100

7.5 mA

4 mA

TEMPERATURE (°C)

V Z(m

V)

(RE

FE

RE

NC

ED

TO

−55

C)

°Δ

From these characteristics, it is evident that the voltagechange with temperature for the new curves is different fromthat for the original ones. It is also apparent that if the testcurrent varies between 7.5 and 4 mA, the voltage changeswould lie along the dashed lines belonging to the giventemperature points. This clearly shows the need for awell-regulated current source.

It should be noted, however, that even when awell-regulated current supply is available, other factorsmight influence the current flowing through a referencediode. For example, to minimize the effects oftemperature-sensitive passive elements in the load circuit oncurrent regulation, it is desirable that the load in parallel withthe reference diode have an impedance much higher than thedynamic impedance of the reference diode.

OTHER CHARACTERISTICS

In addition to the three major characteristics discussedearlier, the following parameters and ratings of referencediodes may be considered in some applications.

Power Dissipation

The maximum dc power dissipation indicates the powerlevel which, if exceeded, may result in the destruction of thedevice. Normally a device will be operated near thespecified test current for which the data-sheet specificationsare applicable. This test current is usually much below thecurrent level associated with the maximum powerdissipation.

Dynamic Impedance

Zener impedance may be construed as composed of acurrent-dependent resistance shunted by avoltage-dependent capacitance. Figure 7 indicates thetypical variations of dynamic zener impedance (ZZ) withcurrent and temperature for the 1N821 reference diodeseries. These diagrams are given in the 1N821 data sheet. Asshown, the zener impedance decreases with current butincreases with ambient temperature.

Figure 7. Variation of Zener ImpedanceWith Current and Temperature (1N821 Series)

1000800600400

200

100806040

20

10864

2

11 2 4 6 8 10 20 40 60 80 100

IZ, ZENER CURRENT (mA)

ZZ

, MA

XIM

UM

ZE

NE

R IM

PE

DA

NC

E (

OH

MS

)

−55°C

25°C

100°C

The impedance of a reference diode is normally specifiedat the test current (IZT). It is determined by measuring the acvoltage drop across the device when a 60 Hz ac current withan rms value equal to 10% of the dc zener current issuperimposed on the zener current (IZT). Figure 8 shows theblock diagram of a circuit used for testing zener impedance.


Figure 8. Block Diagram of Test Circuit for Measuring Dynamic Zener Impedance

DC POWER

SUPPLY

�88i8.88888

�HP712A

A

�88i8.88888

�HP650A

SIGNAL

GENERATOR

�88i8.88888

�HP400H

AC

VTVM

�88i8.88888

�HP412A

DC

VTVM

METER

DUMMY

LOAD

600

0.1�μF1�k1�k

E1

E2

Rx�=�Zz

READ

SET

READ

SET (B)

S1

(A) 10�M

100�pF

R2

1

Rx�=E1�−�E2

E2

ELECTRICAL TESTING

All devices are tested electrically as a last step in themanufacturing process.

The subsequent final test procedures represent anautomated and accurate method of electrically classifyingreference diodes. First, an electrical test is performed on alldevices to insure the correct voltage-breakdown andstability characteristics. Next, the breakdown voltage anddynamic impedance are measured. Finally, the devices areplaced in an automatic data acquisition system thatautomatically cycles them through the completetemperature range specified. The actual voltagemeasurements at the various temperature points are retainedin the system computer memory until completion of the fulltemperature excursion. The computer then calculates thechanges in voltage for each device at each test temperatureand classifies all units on test into the proper category. Thesystem provides a printed readout for every device,including the voltage changes to five digits duringtemperature cycling, and the corresponding EIA typenumber, as well as the data referring to test conditions suchas device position, lot number, and date.

DEVICE RELIABILITY ANDQUALITY ASSURANCE

Insuring a very low failure rate requires maximumperformance in all areas effecting device reliability: Devicedesign, manufacturing processes, quality control, andreliability testing. ON Semiconductor’s basic reliabilityconcept is based on the belief that reference diode reliabilityis a complex yet controllable function of all these variables.

Under this “total reliability” concept, ON Semiconductorcan mass-produce high-reliability reference diodes.

The reliability of a reference diode fundamentallydepends upon the device design, regardless of the degree ofeffort put into device screening and circuit designing.Therefore, reliability measures must be incorporated at the

device design and process development stages to establisha firm foundation for a comprehensive reliability program.The design is then evaluated by thorough reliability testing,and the results are supplied to the Design Engineeringdepartment. This closed-loop feedback procedure providesvaluable information necessary to improve important designfeatures such as electrical instability due to surface effects,mechanical strength, and uniformly low thermal resistancebetween the die and ambient environment.

Process Control

There are more than 2000 variables that must be keptunder control to fabricate a reliable reference diode. Thein-process quality control group controls most of thesevariables. It places a strict controls on all aspects ofmanufacturing from materials procurement to the finishedproduct. Included in this broad spectrum of controls are:

• Materials Control. All materials purchased orfabricated in-plant are checked against rigid specifications.A quality check on vendors’ products is kept up to date toinsure that only materials of a proven quality level will bepurchased.

• In-Process Inspection and Control. Numerous on-lineinspection stations maintain a statistical process controlprogram on specific manufacturing processes. If any ofthese processes are found to be out of control, the discrepantmaterial is diverted from the normal production flow and thecognizant design engineer notified. Corrective action isinitiated to remedy the cause of the discrepancy.

Reliability Testing

The Reliability Engineering group evaluates all newproducts and gives final conclusions and recommendationsto the device design engineer. The Reliability Engineeringgroup also performs independent testing of all products andincludes, as part of this testing program,step-stress-to-failure testing to determine the maximumcapabilities of the product.


SOME STRAIGHT TALK ABOUT MOSORBSTRANSIENT VOLTAGE SUPPRESSORS

INTRODUCTION

Distinction is sometimes made between devicestrademarked Mosorb (by ON Semiconductor Inc.), andstandard zener/avalanche diodes used for reference,low-level regulation and low-level protection purposes. Itmust be emphasized from the beginning that Mosorbdevices are, in fact, zener diodes. The basic semiconductortechnology and processing are identical. The primarydifference is in the applications for which they are designed.Mosorb devices are intended specifically for transientprotection purposes and are designed, therefore, with a largeeffective junction area that provides high pulse powercapability while minimizing the total silicon use. Thus,Mosorb pulse power ratings begin at 500 watts − well inexcess of low power conventional zener diodes which inmany cases do not even include pulse power ratings amongtheir specifications.

MOVs, like Mosorbs, do have the pulse powercapabilities for transient suppression. They are metal oxidevaristors (not semiconductors) that exhibit bidirectionalavalanche characteristics, similar to those of back-to-backconnected zeners. The main attributes of such devices arelow manufacturing cost, the ability to absorb high energysurges (up to 600 joules) and symmetrical bidirectional“breakdown” characteristics. Major disadvantages are: highclamping factor, an internal wear-out mechanism and anabsence of low-end voltage capability. These limitationsrestrict the use of MOVs primarily to the protection ofinsensitive electronic components against high energytransients in application above 20 volts, whereas, Mosorbsare best suited for precise protection of sensitive equipmenteven in the low voltage range − the same range covered byconventional zener diodes. The relative features of the twodevice types are covered in Table 1.

IMPORTANT SPECIFICATIONS FORMOSORB PROTECTIVE DEVICES

Typically, a Mosorb suppressor is used in parallel with thecomponents or circuits being protected (Figure 1), in orderto shunt the destructive energy spike, or surge, around themore sensitive components. It does this by avalanching at its“breakdown” level, ideally representing an infiniteimpedance at voltages below its rated breakdown voltage,and essentially zero impedance at voltages above this level.

In the more practical case, there are three voltagespecifications of significance, as shown in Figure 1a.

a) VRWM is the maximum reverse stand-off voltage atwhich the Mosorb is cut off and its impedance is at itshighest value − that is, the current through the device isessentially the leakage current of a back-biased diode.

b) V(BR) is the breakdown voltage − a voltage at which thedevice is entering the avalanche region, as indicated bya slight (specified) rise in current beyond the leakagecurrent.

c) VRSM is the maximum reverse voltage (clampingvoltage) which is defined and specified in conjunctionwith the maximum reverse surge current so as not toexceed the maximum peak power rating at a pulsewidth (tp) of 1 ms (industry std time for measuringsurge capability).

RELATIVE FEATURES OF MOVs and MOSORBS

Table 1.

MOV Mosorb/Zener Transient Suppressor

• High clamping factor.• Symmetrically bidirectional.

• Energy capability per dollar usually higher than a silicon device.However, if good clamping is required the energy capability wouldhave to be grossly overspecified resulting in higher cost.

• Inherent wear out mechanism clamp voltage degrades after everypulse, even when pulsed below rated value.

• Ideally suited for crude ac line protection.• High single-pulse current capability.• Degrades with overstress.• Good high voltage capability.• Limited low voltage capability.

• Very good clamping close to the operating voltage.• Standard parts perform like standard zeners. Symmetrical bidirec-

tional devices available for many voltages.• Good clamping characteristic could reduce overall system cost.

• No inherent wear out mechanism.

• Ideally suited for precise DC protection.• Medium multiple-pulse current capability.• Fails short with overstress.• Limited high voltage capability unless series devices are used.• Good low voltage capability.


In practice, the Mosorb is selected so that its VRWM isequal to or somewhat higher than the highest operatingvoltage required by the load (the circuits or components tobe protected). Under normal conditions, the Mosorb isinoperative and dissipates very little power. When atransient occurs, the Mosorb converts to a very low dynamicimpedance and the voltage across the Mosorb becomes theclamping voltage at some level above V(BR). The actualclamping level will depend on the surge current through theMosorb. The maximum reverse surge current (IRSM) isspecified on the Mosorb data sheets at 1 ms and for alogrithmically decaying pulse waveform. The data sheetalso contains curves to determine the maximum surgecurrent rating at other time intervals.

Typically, Mosorb devices have a built-in safety margin atthe maximum rated surge current because the clamp voltage,VRSM, is itself, guardbanded. Thus, the parts will beoperating below their maximum pulse-power (Ppk) ratingeven when operated at maximum reverse surge current).

If the transients are random in nature (and in many casesthey are), determining the surge-current level can be aproblem. The circuit designer must make a reasonableestimate of the proper device to be used, based on hisknowledge of the system and the possible transients to beencountered. (e.g., transient voltage, source impedance andtime, or transient energy and time are some characteristicsthat must be estimated). Because of the very low dynamicimpedance of Mosorb devices in the region between V(BR)and VRSM, the maximum surge current is dependent on, andlimited by the external circuitry.

In cases where this surge current is relatively low, aconventional zener diode could be used in place of a Mosorbor other dedicated protective device with some possiblesavings in cost. The surge capabilities most ofON Semiconductor’s zener diode lines are discussed inON Semiconductor’s Application Note AN784.

In the data sheets of some protective devices, theparameter for response time is emphasized. Response timeon these data sheets is defined as the time required for thevoltage across the protective device to rise from 0 to V(BR),and relates primarily to the effective series impedanceassociated with the device. This effective impedance issomewhat complex and changes drastically from theblocking mode to the avalanche mode. In most applications(where the protective device shunts the load) this responsetime parameter becomes virtually meaningless as indicatedby the waveforms in Figures 1b and 1c. If the response timeas defined is very long, it still would not affect theperformance of the surge suppressor.

However, if the series inductance becomes appreciable, itcould result in “overshoot” as shown in Figure 1d that wouldbe detrimental to circuit protection. In Mosorb devices,series inductance is negligible compared to the inductiveeffects of the external circuitry (primarily lead lengths).Hence, Mosorbs contribute little or nothing to overshootand, in essence, the parameter of response time has very littlesignificance. However, care must be exercised in the designof the external circuitry to minimize overshoot.

SUMMARY

In selecting a protective device, it is important to know asmuch as possible about the transient conditions to beencountered. The most important device parameters arereverse working voltage (VRWM), surge current (IRSM), andclamp voltage (VRSM). the product of VRSM and IRSM yieldsthe peak power dissipation, which is one of the primecategories for device selection.

The selector guide, in this book, gives a broad overviewof the Mosorb transient suppressors now available from ONSemiconductor. For more detailed information, pleasecontact your ON Semiconductor sales representative ordistributor.


Figure 1a. Figure 1b.

Figure 1.

Figure 1c. Figure 1d.

MOSORB

PROTECTED

LOAD

Z

OVERSHOOT

VOLTAGE

VRSM

TIME

V(BR)VRWM

tp

Vin

Vout

tclamping, VERY SHORT

VOLTAGE

TIME

VRSM

V(BR)VRWM

Vout

Vin

VOLTAGE

TIME

VOLTAGE

TIME

VRSM

V(BR)VRWM

VRSM

V(BR)VRWM

Vout

Vin

tclamping, VERY LONG tclamping, WITH OVERSHOOT

Vout

Vin

tp = PULSE WIDTH OF INCOMING TRANSIENT

Vin Vout


TYPICAL MOSORB APPLICATIONS

+

−

+

−

+

−

MOSORBMOSORBS

AC

INPUT

DC

POWER

SUPPLY

DC Power Supplies Input/Output Regulator Protection

MOSORB MOSORB

IC

VOLTAGE

REGULATOR

MOSORB

MOSORB

MOSORB

MOSORB

IC

OP AMP

+B

−B

Op Amp Protection

+

−

MOSORB

Inductive Switching Transistor Protection

DC

MOTOR

DC Motors − Reduces EMI Memory Protection

Microprocessor Protection

Computer Interface Protection

MOS

MEMORY

+5 V

MOSORBS

I/O

ADDRESS BUS

RAM ROM

DATA BUS

CONTROL BUS

MOSORB MOSORB

VDD

VGG

Gnd

CPU

CLOCK

I/O

KEYBOARD

TERMINAL

PRINTER

ETC.

Gnd

FUNCTIONAL

DECODER

A

B

C

D

MOSORB

−8 V

MOSORB

−10 V

MOSORBS

Vout


AR450 − CHARACTERIZING OVERVOLTAGE TRANSIENT SUPPRESSORS

Prepared byAl PshaenichON Semiconductor Power Products Division

The use of overvoltage transient suppressors forprotecting electronic equipment is prudent andeconomically justified. For relatively low cost, expensivecircuits can be safely protected by one or even several of thetransient suppressors on the market today. Dictated by thetype and energy of the transient, these suppressors can takeon several forms.

For example, in the telecommunication field, wherelightning induced transients are a problem, such primarysuppressors as gas tubes are often used followed bysecondary, lower energy suppressors. In an industrial orautomotive environment, where transients aresystematically generated by inductive switching, thetransient energy is more well-defined and thus adequatelysuppressed by relatively low energy suppressors. Theselower energy suppressors can be zener diodes, rectifiers withdefined reverse voltage ratings, metal oxide varistors(MOVs), thyristors, and trigger devices, among others. Eachdevice has its own niche: some offer better clamping factorsthan others, some have tighter voltage tolerances, some arehigher voltage devices, others can sustain more energy andstill others, like the thyristor family, have low on-voltages.The designer’s problem is selecting the best device for theapplication.

Thus, the intent of this article is twofold:

1. To describe the operation of the surge current testcircuits used in characterizing lower energy transientsuppressors.

2. To define the attributes of the various suppressors,allowing the circuit designer to make thecost/performance tradeoffs.

Surge suppressors are generally specified withexponentially decaying and/or rectangular current pulses.The exponential surge more nearly simulates actual surgecurrent conditions − capacitor discharges, line and switchingtransients, lightning induced transients, etc., whereasrectangular surge currents are usually easier to implementand control.

To generate an exponential rating, a charged capacitor issimply dumped into the device under test (DUT) and theenergy of each successive pulse increased until the deviceultimately fails. The simplified circuit of Figure 1a describesthe circuit. By varying the size of the capacitor C, thelimiting resistor R2, and the voltage to which C is chargedto, various peak currents and pulse widths (defined to the10% discharge point in this paper) can be obtained. To

Figure 1a. Simplified Exponential Tester

Figure 1b. Simplified Rectangular TesterUsing PNP Switch

Figure 1. Basic Surge Current Testers

C

S1

VIN

R1 R2

S2

IZ

DUT10%

tW

IZ

IZ

DUTIZ

tWRL

VEE

VZ

automate this circuit, the series switches S1 and S2 can bereplaced with appropriately controlled transistors or SCRs.

One method of easily implementing a rectangular surgecurrent pulse is shown in the simplified schematic ofFigure 1b. A PNP transistor switch connected to thepositive supply VEE applies power to the DUT. The currentis obviously set by varying either VEE and/or RL. If however,the transistor switch were replaced with a variable, constantcurrent source, measurement procedures are simplified ashow the limiting resistor need not be selected for variouscurrent conditions.

As in most surge current evaluations, the DUT isultimately subjected to destructive energy (current, voltage,pulse width), the failure points noted, and the derated pointsplotted to produce the energy limitation curve. Of particularinterest is the junction temperature at which the DUTs areoperated, be it near failure or at the specified derated point.This measurement relates to the overall reliability of thesuppressor, i.e., can the suppressor sustain one surge currentpulse or a thousand, and will it be degraded when operatedabove the specified maximum operating temperature?

The Rectangular Current Surge Suppressor Test Circuit tobe described addresses these questions by implementing andmeasuring the rectangular current capability of the


suppressor and determining the device junction temperatureTJ shortly after the end of the surge current pulse. KnowingTJ, the energy to the DUT can be limited just short of failureand thus a complete surge curve generated with only one, ora few DUTs (Figure 6). Second, with the junctiontemperature known, a reliability factor can be determinedfor a practical application.

CIRCUIT OPERATION FOR THERECTANGULAR CURRENT TESTER

The Surge Suppressor Test Circuit block diagram isshown in Figure 2 with the main blocks being the ConstantCurrent Amplifier supplying IZ to the DUT (a zener diodein this instance) during the power pulse and the DiodeForward Current Switch supplying IF during thetemperature sensing time. These two pulses are appliedsequentially, first the much larger IZ, and then the very smallsense current IF. During the IF time, the forward voltage VFof the diode is measured from which the junctiontemperature of the zener diode can be determined. This issimply done by calibrating the forward biased DUT with aspecified low value of IF in a temperature chamber, one pointat 25°C and a second point at some elevated temperature.The result is the familiar diode forward voltage versustemperature linear plot with a slope of about −2 mV/°C fortypical diodes (Figure 7a). Comparing the plot with the testcircuit measured VF yields the DUT junction temperaturefor that particular pulse width and IZ (Figure 7b).

Figure 2. Surge Suppressor TestCircuit Block Diagram

25 μS

BLANKING

MVGATE

SAMPLE

PULSE

300 μS

SENSE

MV

SHORT

DETECTOR

OPEN

DETECTOR

VFS/H

IZDUT

IF

PULSE

GEN

CLOCK

IZ

IF

The System Clock, Pulse Generator, the severalmonostable multivibrators (25 μs Blanking, Sample Pulseand 300 μs Sense MVs) and Gate are fashioned from three

CMOS gate ICs. The remaining blocks are the Sample andHold (S/H) circuit and two detectors for determining thestatus of failed DUTs, either shorted devices or open.

Shown in Figures 3 and 4 respectively, are the completecircuit and significant waveforms. Clocking for the systemis derived from a CMOS, two inverters, astable MV (gates1A and 1B) whose output triggers the two input NOR gateconfigured monostable MV (gates 1C and 1D) to producethe Pulse Generator output pulse (Figure 4b). Alternatively,a single pulse can be obtained by setting switch S2 to the OneShot position and depressing the pushbutton Start switch S1.Contact bounce is suppressed by the 100 ms MV (gates 4Cand D). Frequency of the astable MV, set by potentiometerR1, can vary from about 200 Hz to 0.9 Hz and the pulsewidth, controlled by R2 and the capacitor timing selectorswitch S3, from about 300 μs to 1.3 s.

The positive going Pulse Generator output feeds theConstant Current Amplifier IZ and turns on, in order, NPNtransistor Q1, PNP transistor Q3, NPN Darlington Q4, PNPPower Darlington Q6 and parallel connected PNP PowerTransistors Q8 and Q9. Transistor Q4 is configured as aconstant current source whose current is set by the variablebase voltage potentiometer R3. Thus, the voltage to thebases of Q6, Q8 and Q9 are also accordingly varied.Transistors Q8 and Q9 (MJ14003, IC continuous of 60 A),also connected as constant current sources with their 0.1 Ωemitter ballasting resistors, consequently can produce arectangular current pulse from a minimum of about 0.5 Aand still have adequate gain for 1 ms pulses of 150A peak.Due to propagation delays of this amplifier, the IZ currentwaveform is as shown in Figure 4f. Since Q8 and Q9 mustbe in the linear region for constant current operation, thesetransistors are power dissipation limited at high currents tothe externally connected power supply V+ of 60 V. Thus themaximum DUT voltage, taking into account the clampingfactor of the device, should be limited to about 50 V. Atwider pulse widths and consequently lower currents beforethe DUT fails, the V+ supply should be proportionallyreduced to minimize Q8, Q9 dissipation. As an example, a28 V surge suppressor operating at 100 ms pulse widths canbe tested to destructive limits with V+ of about 40 V.Although a zener diode is shown as the DUT in theschematic, the test devices can be any rectifier with definedreverse voltage, e.g., surge suppressors.

Immediately after the power pulse is applied to the DUT,the negative going sense pulse from the 300 μs MV (Gate2A, Figure 4e) turns on series connected PNP transistor Q10and NPN transistor Q11 of the Diode Forward CurrentSwitch IF. Sense current, set by current limiting resistorselector switch S4, thus flows up from ground through theforward biased DUT, the limiting resistor, and Q11 to the−15 V supply. The result, by monitoring the cathode of theDUT, is a 300 μs wide, approximately −0.6 V pulse.


P.W

. M.V

.

SW

S4

5 m

A

RE

D

4D12 13

114C

S1

STA

RT

SW

+15

V0.

001

μF0.

1 μF

47 k

100

k

8 9

100

mS

MV

CO

NTA

CT

BO

UN

CE

1.8M

+15

V

10

2

1A1

1M

U4C

,D(1

/2)

MC

1400

1

47 k

100

k

1 S

HO

TS

WS

24

22 k

1B3

SW S6

270

pF

FR

EE

RU

N

1C6 7

5

FR

EQ

CO

NTR

1

5M

0.01

μF

3D12 13

0.1

CLO

CK

U1

MC

1457

2

5 m

s <

T1

< 1.

2S

47 k

1 k

W

SY

NC

R2

SW S3

0.00

33 μ

F

N

0.4

μF

39 k

5M PW

CO

NT

+15

V

I 2 S

WIT

CH

3.9

k

5.6

k

22 k

10 k

47 k

270

pF

100

k

8 μs

SA

MP

LEP

ULS

E

270

pF

1.2M

390

pF

TR

IAD

F90

X

1D9

1F11

2A1 2

3

2C8 9

10

2D12 13

14

3B5 6

4

3C8 9

103A

1 23

4B6 5

4

4A2 1

3

− +

2 3

+15

V

+15

V

60 μ

s <

T2N

< 8

.5 m

S

14 151E

+15

V47

k

1316

8

330

pF

100

k

2B5 6

4

12

U2

MC

1400

1

25 μ

s M

.V.

300

μsS

EN

SE

M.V

.20

0 μF

25 V

+−

120

V

7 m

s <

T2W

< 1

.1 S

MD

A10

1A

0.1

μF

PW

R O

NLE

D 2

MC

7815

100

k10

k

Q18

2N39

04

390

k

100

k

270

pF

U3

MC

1400

1

μC 79152

RE

D

1.2

k1/

2 W

+

200

μF25

V(2

) 1

10 k

2 W

0.1

Ω25

W

(2)

+15

V

1N91

4

220

pF

+15

V

390

pF

200

μF25

V

DU

T

1N91

4

+15

V

−15

V

0.1

μF +15

V+1

5 V 7

100

μS S

AM

PLE

DE

LAY

M.V

.

P.W

.

4,7

k1

W

25 μ

s

+15

V

Q14

MP

S80

99

2N39

06Q

15

10 k

1 k

SH

OR

T L

ED

#1

RE

D

RE

D

OP

EN

LE

D

100

k

470

1 W

470

1 W

SH

OR

TD

ET

EC

TO

R #

1Q

162N

5060

0.1

μF

0.1

μF43

k

10 k

4.7

k

−15

V

+15

V

0.1

μF

4.7

k

−15

V1

k

0.1

μF1

k0.

1 μF

4.7

k

10 k

0.1

μF

−15

V

Q13

2N48

58

25 k

100

k

(1/2

) M

C14

001

U4A

,B

10 k

100

k

150

k

+15

V

47 k

1 k

5 pF

+15

V

0.1

μF

1N91

4(2

)

MR

821

1.3

k

4.7

k

−15

V

I F S

WIT

CH

0.1

μF

10 k18

030

0

1/2

W82

01.

2 k

10 k

390 22

1/2

W

1 k

3.3

k

12 k

10 m

Ω

100

k

47 k

1 k

1N91

4

10 k

1 k

1/2

W

10 k

1N91

4

0.1

μF

0.1

μF

U8

GA

TE

300

μS

100

μS

8 μS

SE

NS

E

DE

LAY

SA

MP

LE

I 2 V2

DU

T −0.7

+3 V

+15

V

2N50

60Q

17

OP

EN

DE

TE

CT

OR

RE

SE

T S

W S

5

0.1

μF

0.00

1 μF

1

U5A

(1/2

) M

C14

58

0.1

μF

VF

+15

V+ −

5 67

+ −

3 26

8

4

15

7

4

S/H U

7 LF

355

SD

−15

V

+ −

3 26

7

4

U6

LF35

5

CI R

4

+15

V

1 k

0.1

μF

0.00

1 μF

0.00

1Q

122N

3906

SH

OR

TLE

D #

2

47 k

22 k

10 k

Q8

Q9

Q7

2N66

68

V +

≤ 60

V10

.000

μF

75 V

Q5

8599

MP

S

15 k

1/2

W+1

5 V 0.

1 μF

680

1/2

W R3

1 k

2.7

k

10 m

A

560

25 m

A

V2

10

2N62

87

Q6

MP

SA

29Q

4

2N39

06Q

3Q

12N

3904

Q2

2N39

06

+15

V

Q10

2N39

06

Q11

MP

S80

99

U9

SH

OR

TD

ET

#2

+15

V

233

U5B

(1/2

) M

C14

58

I 2

Figure 3. Surge Suppressors Surge Current Fixture


For accurate measurements of this pulse amplitude,sample and hold circuitry is employed. This consists of unitygain buffer amp U6, series FET switch Q13 and capacitorhold buffer amp U7. The sample pulse (Figure 4H) to thegate of the FET is delayed about 100 μs (by monostable MVG-2C and G-2D) to allow for switching and thermaltransients to settle down. This pulse is derived from thenegative going, trailing edge output pulse of Gate 2D cuttingoff transistor Q18 for the RC time constant in its base circuit.The result is an approximate 8 μs wide sample pulse.Consequently, the DC output voltage from hold amplifierU7 is a measure of the DUT junction temperature.

Invariably, most DUTs will fail short. When the surgesuppressor tester is in the Free-run Mode, the power pulsesubsequent to the DUT shorting could excessively stress theconstant current drivers Q8 and Q9. To prevent thisoccurrence, the Short Detector circuit was implemented.This circuit consists of comparator U5A, 2 input NOR gateconfigured 25 μs monostable MV (G1E and G1F), GateCircuit G3A, 3B and 3C, and SCR Q16. The 25 μs MV

(Figure 4D) is required to blank out turn-on switchingtransients to produce the waveform shown in Figure 4I.During the power pulse, U5A is normally high for a goodDUT (Figure 4J). This waveform is NOR’d with gate 3B(inverted waveform of Figure 4I) to produce a low level(0 V) gate 3C output (Figure 4K).

If, however, the DUT is shorted, U5A output switches lowresulting in a positive pulse output from G3C. This pulsetriggers the SCR on, lighting the LED in its anode circuit andturning on the PNP transistor Q2 across the emitter-base ofQ3, thus clamping off the IZ power pulse. The circuit (Q16)can be reset by opening switch S5.

By and large, this Short Detector circuit was foundadequate to protect transistors Q8 and Q9. However, forsome wide pulse widths, relatively high current conditions,the propagation delay through the Short Detector was toogreat, resulting in excessive FBSOA (Forward Bias SafeOperating Area) stress on Q8 and Q9. Consequently, a fasterShort Detector #2 was implemented.

Figure 4. Surge Suppressor Test Circuit Waveforms

+15 V−14 V

CLOCK

G1B

PULSE GEN

G1D

PULSE GEN

G1C

25 μS MV

G1E

300 μS

SENSE MV

G2A

IZ

100 μS SAMPLE

DELAY MV G2D

8 μS

SAMPLE PULSE Q8

GATE

G3C

SHORT

COMPARATOR U6A

OPEN SCR

TRIGGER U5B

SHORT SCR

TRIGGER G3C

VF

U6−0.6 V

−14 V

GOOD

SHORT

GOOD

SHORT

GOOD

GOOD

OPEN

OPEN

(4A)

(4B)

(4C)

(4D)

(4E)

(4F)

(4G)

(4H)

(4I)

(4J)

(4K)

(4L)

(4M)


This circuit, connected to the collectors of Q8 and Q9, usesa differentiating network (R4, C1) to discriminate betweenthe normally relatively slow fall time of the voltage pulse onthe DUT, and the exceedingly fast fall time when the devicefails. Thus, the R4-C1 time constant (5 ns) will only generatea negative going trigger to PNP transistor Q12 when the DUTvoltage collapses during device failure. The positive goingoutput from Q12 resets the flip-flop (gates 4A and 4B), whichturns on the NPN transistor Q14. This transistor suppliesdrive to the two PNP clamp transistors (Q5 and Q7) placedrespectively across the emitter-bases of the high, constantcurrent stages Q6 and Q8 and Q9. Propagation delay is thusminimized, providing greater protection to the power stagesof the tester. As an added safety feature, the positive goingoutput from Gate 3C when Short Detector #1 is activated isalso used to trigger the flip-flop.

On the few occasions when the DUT fails open, then theOpen Detector consisting of comparator U5B and SCR Q17comes into play. This circuit measures the DUT integrityduring the sense time. For a good DUT (VF < −1 V), U5Boutput remains low (see Figures 4L and 4M). However foran open DUT, VF switches to the negative rail and U5B goeshigh, turning on Q17. As in the Short Detector, Q2 clampsoff the IZ power amplifier.

All of the circuitry including the +15 V and −15 Vregulated power supplies are self-contained, with theexception of the V+ supply. For high current, narrow pulsewidth testing, this external supply should have 10 to 15 Acapability. If not, additional energy storing capacitors acrossthe supply output may be required.

CIRCUIT OPERATION FOR THEEXPONENTIAL SURGE CURRENT TESTER

To generate the surge current curve of peak current versusexponential discharge pulse width, the test circuit ofFigure 5 was designed. This tester is an implementation ofthe simplified capacitor discharge circuit shown inFigure 1A, with the PNP high voltage transistor Q2allowing the capacitor C to charge through limiting resistorR1 and a triggered SCR discharging the capacitor. As shownin Figure 5, the DUTs can be of any technology, although thedevice connected to the capacitor and discharge limitingresistor RS is shown as an MOS SCR. It could just as wellhave been an SCR as the DUT or as the switch for the zenerdiode, rectifier, SIDAC, etc., DUTs.

System timing for this Exponential Surge Current Testeris derived from a CMOS quad 2 input NOR gate with gates1A and 1B comprising a non-symmetrical astable MV ofabout 13 seconds on and about one second off (switch S3open). The positive On pulse from gate 1B turns on the 500V power MOSFET Q1 and the following PNP transistor Q2.The extremely high current gain FET allows for the largebase current variation of Q2 with varying supply voltage(V+). This capacitor charging circuit has a 400 V blockingcapability (limited by the VCEO of Q2) and thus the capacitorC1 used should be comparably rated. When operating with

high voltage (V+ = 200 to 350 V) and large capacitors(>3000 μF), the power dissipated in the current limitingresistor R1 can be substantial, thus necessitating theillustrated 20 W rating. For longer charging times, switch S3is closed, doubling the timing capacitor and the astable MVon time.

To discharge capacitor C1 and thereby generate theexponential surge current, the SCR must be fired. Thistrigger is generated by the positive going one second pulsefrom gate 1A being integrated by the R2C2 network, andthen shaped by gates 1C and 1D. The net result of about 100μs time delay from gate 1D ensures noncoincident timingconditions. This pulse output is then differentiated by C3-R3with the positive going leading edge turning on Q3, Q4 andfinally the SCR with about a 4 ms wide, 15 mA gate pulse.Consequently, the DUT is subjected to a surge current pulsewhose magnitude is dictated by the voltage on the capacitorC1 and value of resistor RS, and also whose pulse width tothe 10% point is 2.3 RSC1. For a fixed pulse width, the DUTis then stressed with increasing charge (by increasing V+)until failure occurs, usually a shorted device.

If the DUT is the SCR (or MOS SCR), the failed conditionis obvious as the capacitor C1 will not be allowed to chargefor subsequent timing cycles. However, when the DUT is thezener, rectifier, SIDAC or even an MOV, and the SCR is anadequately rated switch, the circuit will still dischargethrough the shorted DUT, but now the SCR alone will bestressed by the surge current. A shorted DUT can be detectedby noting the voltage across the device during testing.

One problem encountered when stressing SCRs with highvoltage is when the DUT fails short. The limiting resistorR1, which is only rated for 20 W, would now experiencecontinuous power dissipation for the full On time − as muchas 123 W ([350 V]2/1K). To prevent this occurrence, the PR1Short Protection Circuit is incorporated. Since this is only aproblem when high V+’s (>100 V) are used, the circuit canbe switched in or out by means of switch S2. Whenactivated, this circuit monitors the voltage on capacitor C1some time after the charging cycle begins. If the capacitor ischarging, normal operation occurs. However, if the SCRDUT is shorted, the absence of voltage on the capacitor isdetected and the system is disabled.

The circuit consists of one CMOS IC with NAND gates2A and 2B comprising a one second monostable time delayMV and gates 2C and 2D forming a comparator and NANDgate, respectively.

The negative going, trailing edge of gate 2A isdifferentiated by R4-C4, and amplified by Q5 to form apositive, 10 ms wide pulse (delayed by 1 sec) to gate 2Dinput. If the capacitor C1 is shorted, gate 2C output is high,allowing the now negative pulse from gate 2D to turn onPNP transistor Q6 and SCR Q7. This latches the input to theastable MV gate 1A low, disabling the timing andconsequently removing the power from R1. Resetting thetester for a new device is accomplished by depressing thepushbutton switch S1.


Figure 5. Exponential Surge Current Tester

12

13 2D11

Q2MJE5852

GRNLED

Q42N3906

Q32N3904

CAPDISCHARGE

SIDAC DUT

RECTIFIER DUT

12

13 1B11

14

7

+15 V

1N914

1K20 W

47K1N914

R447K

5

6 2B4

1

2 1C3

10K

+15 V

1.5K1/2 W

1K

15M

10K

ZENER DUT

1N91422K

10K

0.1 μFC3

R310K

1K15K20 W

+15 V

0.001 μFC2

8

9 1A10

+15 V

MC14011R2

100 k

+15 V

1M

1.8M18M

1N91422M 0.47 μF

102 W

C1

RS

+15 V

22K

8

9 2C10

1M

+15 V

Q62N3906 2N5060

Q7

SW S1

1K1/2 W

+15 V

REDLED

RESET

0.1 μF

6.8K

Q52N3906

22K

+15 V

0.01 μF1

2 2A3

1 SEC DELAY MVMC14011

SW

S20.1 μFC4

+15 V

0.1 μF

100K+15 V

1K

10K

1N4005

SCRDUT/SW

MOS SCRDUT

DUT SHORTINDICATOR

R1

V+ ≤350 V

Q1MTP2N50

5

6 1D4

47K

22K

1K

+15 V

150K2 W

0.47 μF

13S

25SSWS3

LV

HV

SWS4

2N3904

ON TIME

Exponential surge current curves, as well as rectangular,are generated by destructive testing of at least several DUTsat various pulse widths and derating the final curve byperhaps 20−30%. These tests were conducted at low dutycycles (<2%). To ensure multicycle operation, the DUTs arethen tested for about 1000 surges at a derated point on thecurve.

TEST RESULTSIn trying to make a comparison of the several different

technologies of transient suppressors, some commondenominator has to be chosen, otherwise, the amount oftesting and data reduction becomes unwieldy. For thisexercise, voltage was used, generally in the 20 V to 30 Vrange, although some of the more unique suppressors(SIDACs, MOS SCRs, SCRs) were tested at their operatingvoltage. As an example, the SIDAC trigger families ofdevices were tested with voltages greater than theirbreakover voltage (104 V to 280 V) and the SCRs weresubjected to exponential surge currents derived from

voltages generally greater than 30 V. Also, since energycapability is related to die size, this parameter is also listed.

For several devices, both rectangular and exponential surgecurrent pulses are listed. Other devices were tested with onlyrectangular pulses (where the junction temperature can bedetermined) and still others, whose applications includecrowbars, with exponential current only.

AVALANCHE RECTIFIERThe Rectangular Surge Current Tester was originally

designed for characterizing rectifier surge suppressors usedin automotive applications. For this operation, wheretemperatures under the hood can reach well over 125°C, itis important to know the device junction temperature atelevated ambient temperature. Figures 6 and 7 describe theresults of such testing on a typical suppressor, the 24 V−32 VMR2520L. It should be noted that these axial leadsuppressors, as well as all other axial lead devices tested,were mounted between two spring loaded clips spaced 1inch apart.


As shown in Figure 6 of the actual current failure pointsof the DUTs, at least four devices were tested at the variouspulse widths, tw (in this example from 0.5 ms to 100 ms).

Also shown in Figure 6 is the curve derived with a singleDUT at an energy level just short of failure. Thismeasurement was obtained by maintaining a constantrectifier forward voltage drop, VF (0.25 V) for all pulsewidths (junction temperature, TJ of 230°C) by varying theavalanche current. Thus, one device can be used,non-destructively, to generate the complete rectangularsurge current curve.

It should also be pointed out that the definition for theexponential tw in this article is the current discharge point tothe 10% value of the peak test current IZM. Expressed in timeconstant τ, this would be 2.3 RC. Some data sheets describetw to the 50% point of IZM (0.69 τ) and others to 5 τ. Tonormalized these time scales (abscissa of curves) simplychange the scales accordingly; i.e., IZM/2 pulse widths wouldbe multiplied by 2.3/0.69 = 3.33 for tw at 10% current pulses.

Figure 7a describes the actual temperature calibrationcurve (measured in a temperature chamber) of theMR2520L and Figure 7b, the junction temperature of theDUT at various 10 ms rectangular pulse current amplitudes.These temperatures are taken from the calibration curve (inactuality, an extremely linear curve), knowing the rectifierforward voltage drop immediately (within 100 μs) aftercessation of power. Note that the junction temperature justprior to device failure is about 290°C.

Figure 6. Experimental Rectangular SurgeCurrent Capability Of The MR2520L Rectifier

Surge Suppressor

I ,

PE

AK

SU

RG

E C

UR

RE

NT

(AM

PS

)Z

ACTUAL DUTS FAILURE POINTS

ONE DUT WITH VF = 0.25 V

MR2520L RECTIFIER

SURGE SUPPRESSOR,

RECTANGULAR PULSE

VZ = 28 V TYP

TA = 25°C

100

50

20

10

200

10 20 50 1000.5 1 52

tW, RECTANGULAR PULSE WIDTH (ms)

TJ = 230°C

ZENER OVERVOLTAGE TRANSIENTSUPPRESSOR

Illustrated in Figure 8 are the actual rectangular andexponential surge current curves of the P6KE30 overvoltagetransient suppressor, an axial lead, Case 17, 30 V zenerdiode characterized and specified for surge currents. Thisdevice is specified for 600 W peak for a 1 ms exponentialpulse measured at IZM/2. From the exponential curve, it is

Figure 7a. Temperature Calibration CurveOf The MR2520L

V

, FO

RW

AR

D V

OLT

AG

E (

VO

LTS

)F

MR2520L AVALANCHE RECTIFIER

SURGE SUPPRESSOR

IF = 10 mA

0.2

0.6

1

0

0.8

0.4

50 1000 150 200 250 300 350TJ, JUNCTION TEMPERATURE (°C)

RECTANGULAR

PULSE

tW = 10 ms

IF = 10 mA

IZ(A)

VF

(V)

TJ

(°C)

11

10

20

30

40

50

0.64

0.57

0.48

0.36

0.25

0.15

225

75

120

180

230

290

55 0.10DUT

FAILED

Figure 7b. MeasuredForward Voltage

Figure 7. Calculated Junction TemperatureOf The MR2520L Surge Suppressor

At Various Avalanche Currents

apparent that the device is very conservatively specified.Also, the relative magnitudes of the two curves reflect thedifferences in the rms values of the two respective pulses.

SIDAC

SIDACs are increasingly being used as overvoltagetransient suppressors, particularly in telephone applications.Being a high voltage bilateral trigger device with relativelyhigh current capabilities, they serve as a costeffectiveovervoltage protection device. As in other trigger devices,when the SIDACs breakover voltage is exceeded, the deviceswitches to a low voltage conduction state, allowing aninordinate amount of surge current to be passed. This is wellillustrated by the surge current curves of Figure 9 whichdescribe the small die size ([37]2mil) axial lead, Case 59-04,MKP9V240 SIDAC. The curves show that this 240 V devicewas able to handle, to failure, as much as 31 A and 15 A,respectively, for 1 ms exponential and rectangular currentpulses. Under the same pulse conditions, the large die([78]2mil) MK1V270 SIDAC handled 170 A and 60 A,respectively, as shown in Table 2.


Figure 8. Surge Current Capability Of The P6KE30Overvoltage Transient Suppressor As A Function Of

Exponential & Rectangular Pulse Widths

I ,

PE

AK

SU

RG

E C

UR

RE

NT

(AM

PS

)Z

IZ

10

%tW

RECTANGULAR

EXPONENTIAL

IZ2

@

tW, PULSE WIDTH (ms)

100

50

10

10 50 1000.5 1 50.1 500 1000

P6KE30 OVERVOLTAGE

TRANSIENT SUPPRESSORVZ = 30 V

PSPEC = 600 W pk

5

1

Figure 9. Measured Surge Current To FailureOf A SIDAC MKP9V240

I ,

PE

AK

SU

RG

E C

UR

RE

NT

(AM

PS

)Z

100

50

20

10

10 20 50 1000.2 0.5 1 52

5

1

2

SIDAC MKP9V240240 VCASE 59-04372 MILS

EXPONENTIAL

RECTANGULAR

IZ

10

%tW


OVERALL RATINGS

The compilation of all of the testing to date on the varioustransient suppressors is shown in Tables 1 and 2. Table 1describes the zener suppressors, avalanche rectifiers andMOVs, comparing the die size and normalized costs(referenced to the MOV V39MA2A). From this data, thedesigner can make a cost/performance judgment.

Of interest is that the small pellet MOV is not the leastexpensive device. The P6KE30 overvoltage transientsuppressor costs about 85% of the MOV, yet it can handleabout three times the current (2.5 A to 0.7 A) for a 100 msrectangular pulse. Under these conditions, the resultantclamping voltages for the zener and MOV were 32 V and60 V respectively.

Also shown in the table is a 1.5 W zener diode specifiedfor zener applications. This low surge current device coststhree times the MOV, illustrating that tight tolerance zenerdiodes are not cost effective and that the user should usedevices designed and priced specifically for the suppressorapplication.

Thyristor type surge suppressors are shown in Table 2.They include four SIDAC series, two SCRs designed andcharacterized specifically for crowbar applications and alsothe MOS SCR MCR1000. The MOS SCR, a processvariation of the vertical structure power MOSFET,combines the input characteristics of the FET with thelatching action of an SCR.

All devices were surge current tested with the resultantpeak currents being impressively high. The TO-220 (150)2

mil SCR MCR69 for example, reached peak current levelsapproaching 700 A for a 1 ms exponential pulse. Theguaranteed, derated, time base translated curves for thecrowbar SCR family of devices are shown in Figure 10, asis the MK1V SIDAC in Figure 11.

Figures 12A−C describe the guaranteed, reverse surgedesign limits for the avalanche rectifier devices. These threefigures illustrate, respectively, the peak current, power andenergy capabilities of these overvoltage transientsuppressors derived from exponential testing. The peakpower, Ppk, ordinate of the curve is simply the product of thederated IZ and VZ and the energy curve, the product of Ppkand tw.


Figure 10. SCR Crowbar Derating Curves

Figure 11. Exponential Surge CurrentCapability Of The MK1V SIDAC, Pulse Width

versus Peak Current

Figure 12b. Peak Power

Ipk

5 TCtW

I ,

PE

AK

CU

RR

EN

T (A

MP

S)

pkC = 8400 μF TA = 25°C

ESR ≈ 25 mΩ N = 2000 PULSES

VC ≤ 60 V f = 3 PULSES/MIN.

a. Peak Surge Current versus Pulse Width

MCR70

MCR69MCR68

MCR67

MCR71

100

30

3000

1000

300

0.5 1 50.1 5010 100

tW, BASE PULSE WIDTH (ms)

SIDAC MK1V115

V(BO) = 115 V MAX

TA = 25°C


I ,

PE

AK

SU

RG

E C

UR

RE

NT

(AM

PS

)Z

100

50

30

10

1

3

5

10 30 50 1000.3 0.5 1 3 5 300

IZ

10%

tW

TC = 25°C, DUTY CYCLE ≤ 1%

SEE NOTE FOR TIME

CONSTANT DEFINITION

MR2530L

MR2525L

MR2520L

, PE

AK

RE

VE

RS

E P

OW

ER

(W

AT

TS

)R

SM

P 700

500

7000

5000

3000

2000

1000

10 20 50 1001 52 200 500 1000τ, TIME CONSTANT (ms)

10000

b. Peak Surge Current versusAmbient Temperature

25 7550 1000 125

0.6

0.8

1

0.4

NO

RM

ALI

ZE

D P

EA

K S

UR

GE

CU

RR

EN

T

N = 2000 PULSES

Figure 12a. Peak Current


SEE NOTE FOR TIME

CONSTANT DEFINITION

MR2530L

MR2525L

MR2520L

200

70

50

20

300

100

30, PE

AK

RE

VE

RS

E C

UR

RE

NT

(AM

PS

)R

SM

I

10 20 50 1001 52 200 500 1000

τ, TIME CONSTANT (ms)


SEE NOTE FOR TIME

CONSTANT DEFINITION

MR2530L

MR2525L

MR2520L

τ, TIME CONSTANT (ms)

100

50

20

30

200

300

, PE

AK

RE

VE

RS

E E

NE

RG

Y (

JOU

LES

)R

SM

W

10 20 50 1001 52 200 500 1000

Figure 12c. Energy

Figure 12. Guaranteed Reverse Surge Design Limits for theMR2525L & MR2530L Overload Transient Suppressors

TA, AMBIENT TEMPERATURE (°C)

NOTE: τ = RC


Table 1. Measured Surge Current Capability of Transient Suppressors

Spec. Peak Current at Pulse Widths, Ipk (Amps)Clamping

Dev PoweDie1 ms 10 ms 20 ms 100 ms

ClampingFactorV1ms

Nor

CosDev

Type TitlePart No. Cas Volt

Powe

(Ene

Die

Siz Exp.Rect.Exp.Rect.Exp.Rect.Exp.Rect.

V1�msV100�ms

Cos

*

Avalanche

Surge Supp.,

OvervoltageMR2520L

194 05

24−32 V2.5�KW

Peak

1502

mil85 A 40 30 18 27�V

22�V� 1.2

Avalanche

Rectifier

Overvoltage

Transient

Suppressor MR2525L

194−05

24−32 V10�KW

Peak

1962

mil150 A 70 54 37 31

23� 1.3 4.0

1.5 W Zener1N5936A

DO 41

30 V1.5 W 372

12 A 5 6 2.5 5 2 3 1.3 4130

� 1.4

3 21.5 W Zener

Diode1N5932A

DO−41

20 V

1.5 W

Cont.

37

mil23 A 6 10 2.8 7 2.3 5 1.4 28

23� 1.2

3.2

Zener

Overvoltage

Transient

P6KE30

17

30 V600 W 602

43 A 14 14 5 10 4.5 5 2.5 4132

� 1.3

0 85Zener Transient

Suppressor P6KE10

17

10 V

600 W

Peak

60

mil24 A 12 9 5.5 16

13� 1.2

0.85

MOSORB

1.5KE30

41A 02

30 V1500 W 1042

35 A 10 4 3533

� 1.1

1 8MOSORB

1.5KE24

41A−02

24 V

1500 W

Peak

104

mil45 A 14 6 30�V

28�V� 1.1

1.8

MOV**

Metal

Oxide

V39MA2AAxial

Lead28 V � 0.16

Joules� 3 mm 9 A 5 0.780�V

60�V

6�A

0.7�A1.0

MOV** Oxide

Varistor V33ZA1Radial

Lead26 V � 1.0

Joules� 7 mm 35 4 A105�V

80�V

35�A

4�A1.4

**G.E.

Table 2. Measured Surge Current Of Thyristor Type Devices

Ipk @ tW

Voltage Die1 ms 10 ms Norm

CostTechnology Device

VoltageRatings Case

DieSize Exponent. Rectang. Exponent. Rectang.

Cost*

MKP9V130 Series 104 V−135 V59 04 372 mil

40 A 13 A 16 A 8 A0 87

SIDACMKP9V240 Series 220 V−280 V

59−04 372 mil31 A 15 A 20 A 8 A

0.87

SIDACMK1V135 Series 120 V−135 V

267 01 782 mil140 A 80 A 55 A 30 A

1 1MK1V270 Series 220 V−280 V

267−01 782 mil170 A 60 A 90 A 28 A

1.1

SCRMCR68 Series

25 V 400 V922 mil 300 A 170 A 1.2

SCRMCR69 Series

25 V−400 V

O1502 mil 700 A 400 A 1.9

MOS SCR MCR1000 Series 200 V−600 V

TO−220127 mil

x183 mil

250 A 170 A 9.3

*Normalized to G.E. MOV V39MA2A, Qty 1-99, 1984 Price

Additionally, the published non-repetitive peak powerratings of the various zener diode packages are illustrated inFigure 13. Figure 14 describes the typical derating factor forrepetitive conditions of duty cycles up to 20%. Using thesetwo empirically derived curves, the designer can thendetermine the proper zener for the repetitive peak currentconditions.

At first glance the derating of curves of Figure 14 appearto be in error as the 10 ms pulse has a higher derating factorthan the 10 μs pulse. However, when the mathematics ofmultiplying the derating factor of Figure 14 by the peakpower value of Figure 13 is performed, the resultantrespective power and current capability of the devicefollows the expected trend. For example, for a 5 W, 20 V

zener operating at a 1.0% duty cycle, the respective deratingfactors for 10 μs and 10 ms pulses are 0.08 and 0.47. Thenon-repetitive peak power capabilities for these two pulses(10 μs and 10 ms) are about 1300 W and 50 W respectively,resulting in repetitive power and current capabilities ofabout 104 W and 24 W and consequently 5.2 A and 1.2 A.

MOV

All of the surge suppressors tested with the exception ofthe MOV are semiconductors. The MOV is fabricated froma ceramic (Zn0), non-linear resistor. This device has wideacceptance for a number of reasons, but for manyapplications, particularly those requiring good clamping



Power is defined as VZ(NOM) x IZ(PK) where VZ(NOM) is thenominal zener voltage measured at the low test current used forvoltage classification.

1N6267 SERIES


5 WATT TYPES

PULSE WIDTH (ms)

0.1

100

0.01 0.02

PP

K(N

OM

), N

OM

INA

L P

EA

K P

OW

ER

(kW

)50

20

10

5

2

1

0.5

0.2

0.1

0.05

0.02

0.010.05 0.2 0.5 1 2 5 10



0.1 0.2 0.5 1 52 10 20 50 100

PULSE WIDTH10 ms

1 ms

100 μs

10 μs

D, DUTY CYCLE (%)

DE

RAT

ING

FA

CT

OR

1

0.70.5

0.3

0.2

0.02

0.1

0.070.05

0.03

0.01

factors, the MOV is found lacking; (clamping factor isdefined as the ratio of VZ at the test current to that at 1.0 mA).This is photographically illustrated in Figure 15 whichcompares a 27 V zener (1N6281) with a 27 V MOV(V27ZA4). The input waveform, through a sourceimpedance resistance to the DUTs, was an exponentiallydecaying voltage waveform of 90 V peak. Figures 15A andB compare the output waveforms (across the DUTs) whenthe source impedance was 500 Ω and Figures 15C and D fora 50 Ω condition. The zener clamped at about 27 V for bothimpedances whereas the MOV was about 40 V and 45 Vrespectively.

Surge current capabilities of a comparably powered MOVwere also determined, as shown in the curve of Figure 16.Although the MOV, a V39MA2A, is specified as a 28 V

Figure 15a.

Figure 15b.

SOURCE IMPEDANCE RS = 500 Ω

27 V MOV

G.E. V27ZA4, 4 JOULES CAPABILITY

27 V ZENER DIODE

ON Semiconductor 1N6281, APPROX. 1.5 JOULES


continuous device (39 V ±10% at 1 mA) at the pulse widthsand currents tested, the resultant voltage VZ across the MOV− 80 V at about 6 A − necessitated a high voltage fixture. Thiswas accomplished with a circuit similar to that of Figure 1B.

But MOVs do have their own niche in the marketplace, asdescribed in Table 3, the Relative Features of MOVs andMOSORBs.


Figure 15c.

Figure 15d.

27 V MOV

27 V ZENER DIODE

Figure 15. Clamping Characteristics of a27 V Zener Diode and 27 V MOV



G.E. V39MA2A MOV

VDCM = 28 V

VNOM = 39 V @ 1 mA

TA = 25°C

10

5

0.3

0.5

3

0.1

1

10 30 50 1001 3 5

I ,

PE

AK

SU

RG

E C

UR

RE

NT

(AM

PS

)Z


Figure 16. Rectangular Surge Current CapabilityOf The V39MA2A MOV

Table 3. Relative Features of MOVsand MOSORBs

MOVMOSORB/Zener Transient

Suppressor

High Clamping Factor Very good clamping close tothe operating voltage.

Symmetrically bidirectional Standard parts perform likestandard zeners. Symmetricalbidirectional devices availablefor many voltages.

Energy capability per dollarusually much greater than asilicon device. However, ifgood clamping is required ahigher energy device would beneeded, resulting in highercost.

Good clampingcharacteristics could reduceoverall cost.

Inherent wear out mechanism,clamp voltage degrades afterevery pulse, even whenpulsed below rated value.

No inherent wear outmechanism.

Ideally suited for crude AC lineprotection.

Ideally suited for precise DCprotection.

High single-pulse currentcapability.

Medium multiple-pulsecurrent capability.

Degrades with overstress. Fails short with overstress.

Good high voltage capability. Limited high voltage capabilityunless series devices areused.

Limited low voltage capability. Good low voltage capability.


SUMMARY

The surge current capabilities of low energy overvoltagetransient suppressors have been demonstrated, includingcost/performance comparison of rectifiers, zeners, thyristortype suppressors, and MOVs. Both rectangular andexponential testing have been performed with the describedtesters. Additionally, the Rectangular Current Surge Testerhas the capability of measuring the diode junctiontemperature of zeners and rectifiers at various power levels,thus establishing safe operating limits.

REFERENCES

1. Cherniak, S., A Review of Transients and Their Meansof Suppression, ON Semiconductor Application NoteAN843.

2. Wilhardt, J., Transient Power Capability of ZenerDiodes, ON Semiconductor Application Note AN784.

3. Pshanenich, A., Characterizing the SCR for CrowbarApplications, ON Semiconductor Application NoteAN879.

4. Pshaenich, A., The SIDAC, A New High VoltageTrigger that Replaces Circuit Complexity and Cost,ON Semiconductor Engineering Bulletin EB-106.

5. General Electric, Transient Voltage SuppressionManual, Second Edition.


MEASUREMENT OF ZENER VOLTAGE TO THERMAL EQUILIBRIUMWITH PULSED TEST CURRENT

Prepared byHerb SaladinDiscrete Power Application Engineering

INTRODUCTION

This paper discusses the zener voltage correlationproblem which sometimes exists between the manufacturerand the customer’s incoming inspection. A method is shownto aid in the correlation of zener voltage between thermalequilibrium and pulse testing. A unique double pulsedsample and hold test circuit is presented which improves theaccuracy of correlation.1

Several zener voltages versus zener pulsed test currentcurves are shown for four package styles. An appendix isattached for incoming inspection groups giving detailedinformation on tolerances involved in correlation.

For many years the major difficulty with zener diodetesting seemed to be correlation of tight tolerance voltagespecifications where accuracy between different test setupswas the main problem. The industry standard and the EIARegistration system adopted thermal equilibrium testing ofzener diodes as the basic test condition unless otherwisespecified. Thermal equilibrium was chosen because it wasthe most common condition in the final circuit design and itwas the condition that the design engineers needed for theircircuit design and device selection. Thermal equilibriumtesting was also fairly simple to set-up for sample testing atincoming inspection of standard tolerance zeners.

In recent years with the advent of economicalcomputerized test systems many incoming inspection areashave implemented computer testing of zener diodes whichhas been generating a new wave of correlation problemsbetween customers and suppliers of zener diodes.

The computerized test system uses short duration pulsetest techniques for testing zener diodes which does notdirectly match the industry standard thermal equilibriumtest specifications.

This paper was prepared in an attempt to clarify thedifferences between thermal equilibrium and short durationpulse testing of zener diodes, to provide a test circuit thatallows evaluation at various pulse widths and a suggestedprocedure for incoming inspection areas that will allowmeaningful correlation between thermal equilibrium andpulse testing.

In the measurement of zener voltage (VZ), thetemperature coefficient effect combined with test currentheating can present a problem if one is attempting tocorrelate VZ measurements made by another party (FinalTest, Quality Assurance or Incoming Inspection).2 Thispaper is intended as an aid in determining VZ at some test

current (IZT) pulse width other than the pulse width used bythe manufacturer.

Thermal equilibrium (TE) is reached when diode junctiontemperature has stabilized and no further change will occurin VZ if the IZT time is increased.2 This absolute value canvary depending on the mounting method and amount ofheatsinking. Therefore, thermal equilibrium conditionshave to be defined before meaningful correlation can exist.

Normalized VZ curves are shown for four package stylesand for three to five voltage ratings per package. Pulsewidths from 1 ms up to 100 seconds were used to arrive ator near thermal equilibrium for all packages with a givenmethod of mounting.

Mounting

There are five conditions that can affect the correlation ofVZ measurements and are: 1) instrumentation, 2) TA, 3) IZTtime, 4) PD and 5) mounting. The importance of the firstfour conditions is obvious but the last one, mounting, canmake the difference between good and poor correlation. Themounting can have a very important part in VZ correlationas it controls the amount of heat and rate of heat removalfrom the diode by the mass and material in contact with thediode package.

Two glass axial lead packages (DO-35 and DO-41),curves (Figures 5 and 6) were measured with standardGrayhill clips and a modified version of the Grayhill clips topermit lead length adjustment.

Test Circuit

The test circuit (Figure 8) consists of standard CMOSlogic for pulse generation, inverting and delaying. The logicdrives three bipolar transistors for generation of the powerpulse for IZT. VZ is fed into an unique sample and hold (S/H)circuit consisting of two high input impedance operationalamplifiers and a field effect transistor switch.

For greater accuracy in VZ measurements using a singlepulse test current, the FET switch is double pulsed. Doublepulsing the FET switch for charging the S/H capacitorincreases accuracy of the charge on the capacitor as thesecond pulse permits charging the capacitor closer to thefinal value of VZ.

The timing required for the two pulse system is shown inwaveform G-3C whereby the initial sample pulse is delayedfrom time zero by a fixed 100 μs to allow settling time andthe second pulse is variable in time to measure the analoginput at that particular point. The power pulse (waveformG-2D) must also encompass the second sample pulse.


To generate these waveforms, four time delay monostablemultivibrators (MV) are required. Also, an astable MV, isrequired for free-running operation; single pulsing is simplyinitiated by a push-button switch S1. All of the pulsegenerators are fashioned from two input, CMOS NOR gates;thus three quad gate packages (MC14001) are required.Gates 1A and 1B form a classical CMOS astable MV clockand the other gates (with the exception of Gate 2D) comprisethe two input NOR gate configured monostable MV’s. ThePulse Width variable delay output (Gate 1D) positions thesecond sample pulse and also triggers the 100 μs Delay MVand the 200 μs Extended Power Pulse MV, The respectivepositive going outputs from gates 3A and 2C are diodeNOR’ed to trigger the Sample Gate MV whose output willconsequently be the two sample pulses. These pulses thenturn on the PNP transistor Q1 level translator and thefollowing S/H N-channel FET series switch Q2. Op ampsU4 and U5, configured as voltage followers, respectivelyprovide the buffered low output impedance drive for theinput and output of the S/H. Finally, the pulse extendedPower Gate is derived by NORing (Gate 2D) the PulseWidth Output (Gate 1D) with the 200 μs MV output (Gate2C). This negative aging gate then drives the PowerAmplifier, which, in turn, powers the D.U.T. The poweramplifier configuration consists of cascaded transistorsQ3−Q5, scaled for test currents up to 2 A.

Push button switch (S4) is used to discharge the S/Hcapacitor. To adjust the zero control potentiometer, groundthe non-inverting input (Pin 3) of U4 and discharge the S/Hcapacitor.

Testing

The voltage VCC, should be about 50 volts higher than theD.U.T. and with RC selected to limit the IZT pulse to a valuemaking VZT IZT = 1/4 PD (max), thus insuring a good currentsource. All testing was performed at a normal roomtemperature of 25°C. A single pulse (manual) was used andat a low enough rate that very little heat remained from theprevious pulse.

The pulse width MV (1C and 1D) controls the width of thetest pulse with a selector switch S3 (see Table 1 for capacitorvalues). Fixed widths in steps of 1, 3 and 5 from 1 ms to 10seconds in either a repetitive mode or single pulse isavailable. For pulse widths greater than 10 seconds, a stopwatch was used with push button switch (S1) and with themode switch (S2) in the > 10 seconds position.

For all diodes with VZ greater than about 6 volts a resistorvoltage divider is used to maintain an input of about 6 V tothe first op amp (U4) so as not to overload or saturate thisdevice. The divider consists of R5 and R6 with R6 being10 kΩ and R5 is selected for about a 6 V input to U4.Precision resistors or accurate known values are required foraccurate voltage readout.

Table 1. S3 — Pulse Width

SwitchPosition *C(μF) t(ms)

12345678910111213

0.0010.0040.0060.010.040.060.10.40.61.01.26.010

1351030501003005001K3K5K10K

*Approximate Values

Using Curves

Normalized VZ versus IZT pulse width curves are shownin Figure 1 through 6. The type of heatsink used is shown orspecified for each device package type. Obviously, it isbeyond the scope of this paper to show curves for everyvoltage rating available for each package type. The objectwas to have a representative showing of voltages includingwhen available, one diode with a negative temperaturecoefficient (TC).

These curves are actually a plot of thermal responseversus time at one quarter of the rated power dissipation.With a given heatsink mounting, VZ can be calculated atsome pulse width other than the pulse width used to specifyVZ.

For example, refer to Figure 5 which shows normalizedVZ curves for the axial lead DO-35 glass package. Threemounting methods are shown to show how the mountingeffects device heating and thus VZ. Curves are shown for a3.9 V diode (1N5228B) which has a negative TC and a 12 Vdiode (1N5242B) having a positive TC.

In Figure 5, the two curves generated using the Grayhillmountings are normalized to VZ at TE using theON Semiconductor fixture. There is very little difference inVZ at pulse widths up to about 10 seconds and mounting onlycauses a very small error in VZ. The maximum error occursat TE between mountings and can be excessive if VZ isspecified at TE and a customer measures VZ at some narrowpulse width and does not use a correction factor.

Using the curves of Figure 5, VZ can be calculated at anypulse width based upon the value of VZ at TE which isrepresented by 1 on the normalized VZ scale. If the 1N5242Bdiode is specified at 12 V ± 1.0% at 90 seconds which is atTE, VZ at 100 ms using either of the Grayhill clips curveswould be 0.984 of the VZ value at TE or 1 using theON Semiconductor fixture curve. If the negative TC diodeis specified at 3.9 V ± 1.0% at TE (90 seconds), VZ at 100ms would be 1.011 of VZ at TE (using ON Semiconductorfixture curve) when using the Grayhill Clips curves.


In using the curves of Figure 5 and 6, it should be kept inmind that VZ can be different at TE for the three mountingsbecause diode junction temperature can be different for eachmounting at TE which is represented by 1 on the VZnormalized scale. Therefore, when the correlation of VZbetween parties is attempted, they must use the same type ofmounting or know what the delta VZ is between the twomountings involved.

The Grayhill clips curves in Figure 6 are normalized to theON Semiconductor fixture at TE as in Figure 5. Figures 1through 4 are normalized to VZ at TE for each diode andwould be used as Figures 5 and 6.

Measurement accuracy can be affected by test equipment,power dissipation of the D.U.T., ambient temperature andaccuracy of the voltage divider if used on the input of the firstop-amp (U4). The curves of Figures 1 through 6 are for anambient temperature of 25°C, at other ambients, θVZ has tobe considered and is shown on the data sheet for the1N5221B series of diodes. θVZ is expressed in mV/°C andfor the 1N5228B diode is about −2 mV/°C and for the1N5242B, about 1.6 mV/°C. These values are multiplied bythe difference in TA from the 25°C value and either

subtracted or added to the calculated VZ depending uponwhether the diode has a negative or positive TC.

General Discussion

The TC of zener diodes can be either negative or positive,depending upon die processing. Generally, devices with abreakdown voltage greater than about 5 V have a positiveTC and diodes under about 5 V have a negative TC.

Conclusion

Curves showing VZ versus IZT pulse width can be used tocalculate VZ at a pulse width other than the one used tospecify VZ. A test circuit and method is presented to obtainVZ with a single pulse of test current to generate VZ curvesof interest.

References

1. Al Pshaenich, “Double Pulsing S/H Increases SystemAccuracy”; Electronics, June 16, 1983.

2. ON Semiconductor Zener Diode Manual, Series A,1980.


FIGURES 1 thru 8 — Conditions: Single Pulse, TA = 25°C, VZ IZT = 1/4 PD (Max) Each device normalized to VZ at TE.

AXIAL LEAD PACKAGES: MOUNTING STANDARD GRAYHILL CLIPS

Figure 1. DO-35 (Glass) 500 mW Device Figure 2. DO-41 (Glass) 1 Watt Device

Figure 3. DO-41 (Plastic) 1.5 Watt Device Figure 4. Case 17 (Plastic) 5 Watt Device

V

, ZE

NE

R V

OLT

AG

E (

NO

RM

ALI

ZE

D)

Z

VZ = 3.9 V

6.2 V

12 V

1.06

1.04

1.02

1

0.98

0.96

0.94

0.9210 40 1001 4 400 1K 4K 10K 40K 100K


75 V

1.06

1.04

1.02

1

0.98

0.96

0.94

0.92

V

, ZE

NE

R V

OLT

AG

E (

NO

RM

ALI

ZE

D)

Z

10 40 1001 4 400 1K 4K 10K 40K100K


VZ = 3.9 V

6.2 V

12 V

V

, ZE

NE

R V

OLT

AG

E (

NO

RM

ALI

ZE

D)

Z

VZ = 3.3 V

6.2 V

12 V

1.06

1.04

1.02

1

0.98

0.96

0.94

0.9210 40 1001 4 400 1K 4K 10K 40K 100K


V

, ZE

NE

R V

OLT

AG

E (

NO

RM

ALI

ZE

D)

Z

VZ = 3.9 V

6 V

13 V

1.06

1.04

1.02

1

0.98

0.96

0.94

0.9210 40 1001 4 400 1K 4K 10K 40K100K


68 V 150 V


THREE MOUNTING METHODS: DO-35 AND DO-41

Figure 5. DO-35 (Glass) 500 mW Device

Figure 6. DO-41 (Glass) 1 Watt Device

Figure 7. Standard Grayhill Clips

MOUNTING FIXTURE

1N5242B (VZ = 12 V)

1N5228B (VZ = 3.9 V)

GRAYHILL CLIPSSTANDARD, L = 11/16″

ON SEMICONDUCTORFIXTURE L = 1/2″

MOUNTINGS:

MODIFIEDL = 3/8″

1.022

1.018

1.014

1.012

1.008

1.004

1

0.996

10 40 1001 4 400 1K 4K 10K 40K 100K

0.992

0.988

0.984

0.98

0.976

0.972

0.968

V

, ZE

NE

R V

OLT

AG

E (

NO

RM

ALI

ZE

D)

Z


1.004

1

10 40 1001 400 1K 4K 10K 40K 100K

0.992

0.988

0.984

0.98

0.9764

0.996

MOUNTINGS:GRAYHILL CLIPS

STANDARD, L = 11/16″MODIFIED, L = 3/8″

ON SEMICONDUCTORFIXTUREL = 1/2″

1N4742A

VZ = 12 V

V

, ZE

NE

R V

OLT

AG

E (

NO

RM

ALI

ZE

D)

Z


1.41

1.69

.78

.75

2.31

GRAYHILL CLIPSMODIFIED, L = 3/8″

STANDARD,L = 11/16″


Figure 8. Zener Voltage Double Pulsing S/H Test Circuit

3B

3D3C

2C

1D

0.1

+12

V

1A1B

1C

2A2B

3A

2D

100

100

k22

pF

1N91

4

1/2

MC

1400

1U

312

k

22 k

4.7

k

10 k

12 k

1 k

Q4

MP

SA

42

PO

WE

R A

MP

LIF

ER

RC

R5

68 k

2 W

DU

T

**

Q5A

*M

JE35

0

R6

10 k

1N91

4

U4

MC

1741

VIN

+ − −12

V0.

1 μF

2N48

56

SD

Q2

0.1

μF

S4

0.1

μF

−12

V

0.1

μF0.1

μF

VO

=V

INK

ZE

RO

CO

NT

RO

L+

12 V

25 k

VIN

=V

IN (

10 k

)

R5

+ 10

k

VIN K

=

+12

VV

CC

≤ 2

50 V

1N91

433

pF

−12

V10 k

22 k

47 k

Q1

2N39

06

+12

V

SA

MP

LE G

AT

E M

V51

0 pF

330

k

U3

1/2

MC

1400

1

+12

V

EX

TE

ND

ED

PO

WE

R-P

ULS

E M

V

1N91

4

47 k

47 k

0.00

1μF

47 k1N

914

0.00

1 μF

200

μs

+12

V 680

k

510

pF

27 k

+12

VU

33/

4 M

C14

001

0.00

1 μF

VD

D

510

pF

330

k

+12

V

VD

D

+12

V

0.00

1 μF

27 k

100

μsD

ELA

Y M

V

1/4

MC

1400

1U

2

S2B

≈2 R

1R

1

C1

VD

D

T1

≈ 2.

2R1C

1

MC

1400

1

1N91

4

MO

DE

SE

L S

WS

2A

10 k

ON

ES

HO

T

0.00

1 μF

>10

SE

C

STA

RT

SW

S1

+12

V

+12

V

100

k

FR

EE

RU

N

C2

C4

C5

C15C3

100

kR

2

5M R3

P.W

.C

ON

TR

OL

T2

≈ 0.

6C2

(R2

+ R

3)

1

S3

SE

ETA

BLE

1

PU

LSE

WID

TH

MV

2 3 13

27 k

TIM

ING

WA

VE

FO

RM

S

GA

TE

G-1

D

G-2

A

G-2

C

G-2

D

G-3

A G-3

C

100

μs

100

μs10

0 μs20

0 μs

2N 3906

Q3

+ −

+12

V+1

2 V

U5

LF15

5J

**Te

k C

urre

nt P

robe

**P

6302

/AM

503

1 k

12 k

68 k

2 W

RC

Q4

*FO

R D

UT

CU

RR

EN

TS

:20

0 m

A ≤

I ZT ≤

2 A

VC

C ≤

250

V

DU

T

Q5A

MJE

35

Q5B

MJE

5850

VC

C ≤

250

V


APPENDIX ARecommended Incoming Inspection Procedures

Zener Voltage TestingPulsed versus Thermal Equilibrium

This section is primarily for use of incoming inspectiongroups. The subject covered is the measurement of zenervoltage (VZ) and the inherent difficulty of establishingcorrelation between supplier and buyer when using pulsedtest techniques. This difficulty, in part, is due to theinterpretation of the data taken from the variety of availabletesters and in some cases even from the same model types.It is therefore, our intent to define and reestablish astandardized method of measurement to achieve correlationno matter what test techniques are being used. Thisstandardization will guarantee your acceptance of goodproduct while maintaining reliable correlation.

DEFINITION OF TERMS

Temperature Coefficient (TC):

The temperature stability of zener voltages is sometimesexpressed by means of the temperature coefficient (TC).This parameter is usually defined as the percent voltagechange across the device per degree centigrade, or as aspecific voltage change per degree centigrade. Temperaturechanges during test are due to the self heating effects causedby the dissipation of power in the zener junction. The VZ willchange due to this temperature change and will exhibit apositive or negative TC, depending on the zener voltage.Generally, devices with a zener voltage below five volts willhave a negative TC and devices above five volts will exhibita positive TC.

Thermal Equilibrium (TE)

Thermal equilibrium (TE) is reached when the diodejunction temperature has stabilized and no further changewill occur. In thermal equilibrium, the heat generated at thejunction is removed as rapidly as it is created, hence, nofurther temperature changes.

MEASURING ZENER VOLTAGE

The zener voltage, being a temperature dependentparameter, needs to be controlled for valid VZ correlation.Therefore, so that a common base of comparison can beestablished, a reliable measure of VZ can only occur whenall possible variables are held constant. This common baseis achieved when the device under test has had sufficienttime to reach thermal equilibrium (heatsinking is required tostabilize the lead or case temperature to a specified value for

stable junction temperatures). The device should also bepowered from a constant current source to limit changes ofpower dissipated and impedance.

All of the above leads us to an understanding of whyvarious pulse testers will give differing VZ readings; thesedifferences are, in part, due to the time duration of test (pulsewidth), duty cycle when data logging, contact resistance,tolerance, temperature, etc. To resolve all of this, one onlyneeds a reference standard to compare their pulsed resultsagainst and then adjust their limits to reflect thosedifferences. It should be noted that in a large percentage ofapplications the zener diode is used in thermal equilibrium.

ON Semiconductor guarantees all of it’s axial leadedzener products (unless otherwise specified) to be withinspecification ninety (90) seconds after the application ofpower while holding the lead temperatures at 30 ± 1°C, 3/8of an inch from the device body, any fixture that will meetthat criteria will correlate. 30°C was selected over thenormally specified 25°C because of its ease of maintenance(no environmental chambers required) in a normal roomambient. A few degrees variation should have negligibleeffect in most cases. Hence, a moderate to large heatsink inmost room ambients should suffice.

Also, it is advisable to limit extraneous air movementsacross the device under test as this could change thermalequilibrium enough to affect correlation.

SETTING PULSED TESTER LIMITS

Pulsed test techniques do not allow a sufficient time forzener junctions to reach TE. Hence, the limits need to be setat different values to reflect the VZ at lower junctiontemperatures. Since there are many varieties of test systemsand possible heatsinks, the way to establish these limits is toactually measure both TE and pulsed VZ on a serializedsample for correlation.

The following examples show typical delta changes inpulsed versus TE readings. The actual values you use forpulsed conditions will depend on your tester. Note, that thereare examples for both positive and negative temperaturecoefficients. When setting the computer limits for a positiveTC device, the largest difference is subtracted from theupper limit and the smallest difference is subtracted from thelower limit. In the negative coefficient example the largestchange is added to the lower limit and the smallest changeis added to the upper limit.


ON Semiconductor Zeners

• Thermal equilibrium specifications:VZ at 10 mA, 9 V minimum, 11 V maximum:(Positive TC)

TE Pulsed Difference

9.53 V9.35 V9.46 V9.56 V9.50 V

9.45 V9.38 V9.83 V9.49 V9.40 V

−0.08 V−0.07 V−0.08 V−0.07 V−0.10 V

Computer test limits:Set VZ max. limit at 11 V − 0.10 V = 10.9 VSet VZ min. limit at 9 V − 0.07 V = 8.93 V

• Thermal equilibrium specifications:VZ at 10 mA, 2.7 V minimum, 3.3 V maximum:(Negative TC)

TE Pulsed Difference

2.78 V2.84 V2.78 V2.86 V2.82 V

2.83 V2.91 V2.84 V2.93 V2.87 V

+0.05 V+0.07 V+0.05 V+0.07 V+0.05 V

Computer test limits:Set VZ min. limit at 2.7 V + 0.07 V = 2.77 VSet VZ max. limit at 3.3 V + 0.05 V = 3.35 V


April, 2005 − Rev. 1127 Publication Order Number:

NLAS3158/D

ON Semiconductor and are registered trademarks of Semiconductor Components Industries, LLC (SCILLC). SCILLC reserves the right to make changes without further noticeto any products herein. SCILLC makes no warranty, representation or guarantee regarding the suitability of its products for any particular purpose, nor does SCILLC assume anyliability arising out of the application or use of any product or circuit, and specifically disclaims any and all liability, including without limitation special, consequential or incidentaldamages. “Typical” parameters which may be provided in SCILLC data sheets and/or specifications can and do vary in different applications and actual performance may vary overtime. All operating parameters, including “Typicals” must be validated for each customer application by customer’s technical experts. SCILLC does not convey any license underits patent rights nor the rights of others. SCILLC products are not designed, intended, or authorized for use as components in systems intended for surgical implant into the body,or other applications intended to support or sustain life, or for any other application in which the failure of the SCILLC product could create a situation where personal injury or deathmay occur. Should Buyer purchase or use SCILLC products for any such unintended or unauthorized application, Buyer shall indemnify and hold SCILLC and its officers, employees,subsidiaries, affiliates, and distributors harmless against all claims, costs, damages, and expenses, and reasonable attorney fees arising out of, directly or indirectly, any claim ofpersonal injury or death associated with such unintended or unauthorized use, even if such claim alleges that SCILLC was negligent regarding the design or manufacture of the part.SCILLC is an Equal Opportunity/Affirmative Action Employer. This literature is subject to all applicable copyright laws and is not for resale in any manner.

PUBLICATION ORDERING INFORMATIONN. American Technical Support: 800−282−9855 Toll FreeUSA/Canada

Japan: ON Semiconductor, Japan Customer Focus Center2−9−1 Kamimeguro, Meguro−ku, Tokyo, Japan 153−0051Phone: 81−3−5773−3850

HBD854/D

LITERATURE FULFILLMENT:Literature Distribution Center for ON SemiconductorP.O. Box 61312, Phoenix, Arizona 85082−1312 USAPhone: 480−829−7710 or 800−344−3860 Toll Free USA/CanadaFax: 480−829−7709 or 800−344−3867 Toll Free USA/CanadaEmail: [email protected]

ON Semiconductor Website: http://onsemi.com

Order Literature: http://www.onsemi.com/litorder

For additional information, please contact yourlocal Sales Representative.

Date post:	19-Apr-2018
Category:	Documents
Upload:	ngotuyen
View:	246 times
Download:	5 times

TVS/Zener Theory and Design Considerations -...

Documents