Industrial Signal Conditioning, A Tutorial OVERVIEW OF INDUSTRIAL MEASUREMENT ..................................................................................2 USES OF INDUSTRIAL MEASUREMENT ............................................................................................2 INDUSTRIAL MEASUREMENT ENVIRONMENT ..................................................................................3 Field Control Room Field Wiring SENSORS ..........................................................................................................................................3 Terminology Sensor Linearization Sensor Classification Temperature Sensors Motion Sensors Strain Gages LOOPS AND ANALOG SIGNALS ......................................................................................................11 Measurement Loop Configurations Analog Signals SIGNAL INTEGRITY ........................................................................................................................14 Sources of Error Ways to Preserve Signal Integrity DESIGN EXAMPLES ........................................................................................................................18 Servo Control Aluminum Smelting Grounded Thermocouples PRODUCT SELECTION GUIDE ........................................................................................................21
Transcript
Industrial Signal Conditioning, A TutorialOVERVIEW OF INDUSTRIAL MEASUREMENT ..................................................................................2
USES OF INDUSTRIAL MEASUREMENT............................................................................................2
LOOPS AND ANALOG SIGNALS ......................................................................................................11Measurement Loop ConfigurationsAnalog Signals
SIGNAL INTEGRITY ........................................................................................................................14Sources of ErrorWays to Preserve Signal Integrity
The need to measure and control the opera-tion of machinery or process equipment is asold as the Industrial Revolution. Plant instru-mentation has now become the nerves andbrain of the modern manufacturing plant. Itregulates and supervises the operation of theequipment within the plant. It also providesthe means to make plants economicallyviable. Instrumentation allows the use ofprocesses which would be difficult or impos-sible to operate without automation.
Instruments have grown from purely analogsystems to the ‘smart’ systems in use today,ranging from simple potentiometers to com-plex analyzers such as infra-red spectropho-tometers. Yet, for all the advances in systemsdevelopment, analog field measurements andthe electronic signals that carry them are stillnecessary ingredients in all systems.
Analog measurements take many forms, butcan be roughly classified into two types—physical measurements and compositionalmeasurements. The first type includes pres-sure, temperature, flow, force, vibration,mass and density. The second includes suchmeasurements as conductivity, pH and chem-ical analysis.
Obtaining, maintaining and improving thequality of these measurements is the goal ofproper signal conditioning. Good signal con-
ditioning preserves the quality of the meas-urements available and allows the plant sys-tems to make best use of the control and dataacquisition systems installed. HelpingDataforth customers achieve good signalconditioning is the goal of this handbook.
USES OF INDUSTRIAL MEASUREMENT
There are several distinct uses of analogmeasurements.
INDICATE-ONLY measurements are used toindicate the condition of various elements ofa process. Estimates place the ratio of indi-cate-only to control inputs at somewherebetween 2-to-1 and 3-to-1. Regardless, thesemeasurements are useful to monitor the con-dition of intermediary events at every stageof manufacture or processing and may pro-vide necessary information to the plant oper-ator if a control measurement fails. An exam-ple of this kind of measurement is the com-plete temperature monitoring of the distilla-tion trays in a distillation tower. Each meas-urement is not essential to the control of theside-draw products, but does provide valu-able insight about the operating conditionsand material and energy balances within thetower. They also allow the operator to inter-vene manually if a control measurementfails.
CONTROL measurements are essential to theeconomic viability, safety or functioning of a
manufacturing process. They provide controlover a physical or compositional characteris-tic of the process. For example, the tempera-ture of a heat exchanger is an essentialparameter for both process and safety rea-sons. Flow measurements and control such asthose illustrated by Figure 1 appear in almostevery plant.
CUSTODY TRANSFER measurements needhighly accurate and stable characteristics.These measurements provide information forplant inventory, quantify the amount of mate-rial bought or sold between parties or trackinternal transfers of material from one oper-ating unit to another within the plant.Frequently the calibration of the instrumentsis regulated by municipal, state or Federalagencies. The gasoline pump in your neigh-borhood is an example of these measure-ments.
ENVIRONMENTAL measurements have grownenormously in recent years to provide trace-able records of plant effluents, and wasteproducts in compliance with governmentregulations. An entire technology hasevolved to detect and control hazardousmaterials of all kinds.
SAFETY MEASUREMENTS– Finally, there isan entirely separate and autonomous type ofmeasurement system whose sole function isto monitor and limit dangerous conditions.Measurements include critical processparameters that indicate unsafe operation andpotential danger. These systems override the
2
Figure 1. Typical Measurement/Control Loop
3
regulatory controls and cause a plant shut-down to a safe status should emergency con-ditions dictate. Known as EMERGENCYSHUTDOWN systems, they are frequentlyequipped with sophisticated events-monitor-ing recorders so that later analysis of theshutdown events can be made and futuremalfunctions avoided or controlled.
INDUSTRIAL MEASUREMENT ENVIRONMENT
Figure 2 shows a simplified view of a meas-urement and control system. It shows onlythe essential elements, but demonstrates thedivision between field and control roomfunctions.
FIELDThe term ‘field’ refers to the area where theequipment making a product or running aprocess resides. It is most often the factoryfloor or the outside areas of an industrialcomplex such as a chemical plant. What setsit apart from other areas is its harsh electricaland physical environment. The equipmentlocated there is exposed to a much greaterrange of electrical noise, power surges, tem-perature, humidity, and corrosive or damag-ing environments.
The field is where process variables must bemeasured and where measuring and somesignal conditioning equipment must be locat-ed. The measuring equipment and wiring
may be near heavy electrical equipment,motor contactors and even lightning. Oftenthe wiring runs several hundreds or thou-sands of feet, increasing the likelihood ofoutside interference from this environment.
CONTROL ROOMThe control room is usually a more benignplace than the field, with a cleaner atmos-phere, air conditioning, and fewer hazardousconditions. However, it also contains electri-cal equipment and the potential for degradingthe quality of measurements. The controlroom contains signal conditioning and com-puting equipment that is sensitive to electri-cal interference.
The control room is usually the locationwhere people interact with the measurementand control systems in a plant. There areexceptions, but the control room is wheremost decisions about the plant or process aremade.
FIELD WIRINGInstrumentation wiring connecting fielddevices to the control room typically consistsof heavy-duty (16-18 AWG) pairs. They areoften twisted together to aid in reducingmagnetically coupled interference and runwith other signal wires in a separate wiringtray away from power distribution wiring.Large numbers of sensor or transmitter sig-nals may be gathered in terminal cabinetslocated either in the control room area or in
an intermediate site for ease of connection tothe signal conditioning and display equip-ment.
In most instances, the cost of wiring is a largepercentage of the installed cost of the instru-ment system. This is especially true when thewiring is in or passes through plant areascontaining flammable gases or vapors. Thehazards represented by these atmospheresforce the use of very expensive techniques toprevent fires or explosions caused by an elec-tric spark.
Data concentrators may be used to reducewiring costs. These devices collect largenumbers of signals close to their origins inthe field, perform signal conditioning anddigital data conversion locally and send thedigitized information by communicationlinks to a local area network or to the controlroom equipment directly.
SENSORS
TERMINOLOGYThe terms ‘sensor’ and ‘transmitter’ areoften used interchangeably. However, thereis an important difference between sensorsand transmitters. A sensor is a device thatconverts a physical quantity into a formwhich can be further used to indicate or con-trol the measured variable. This form may bemechanical, like a pressure dial gauge, ormay produce an electrical signal. A transmit-ter takes this idea one step further and pro-
Figure 2. Control and Field Conditions — Industrial Measurement Environment
4
vides some manipulation of the sensor signalat the sensor location through amplification,filtering, isolation or other electronic means.For the purposes of this handbook the maindifference between sensors and transmittersis that transmitters manipulate the signal atthe measurement point. Usually, a dataacquisition or control system contains a mixof sensors and transmitters.
Ideally, each sensor would have signal-con-ditioning at the point of measurement andtransmit a high-level signal back to the dataacquisition system or control system. Theshorter interconnection from sensor to signalconditioner is less likely to pick up noise, andthe high-level output signal offers betterimmunity against induced pickup from natu-ral or man-made sources. However, this idealconflicts with the economic reality that sig-nal-conditioning at the measurement point isa costlier approach than shared signal condi-tioning at the data collection/control system.Thus a compromise must be made betweensignal integrity and system cost.
SENSOR LINEARIZATIONMany sensors exhibit a deviation from anideal (linear) relationship between input anoutput. For example, a given change in tem-perature does not give rise to the samechange in EMF for most thermocoupleswhen measured over different temperatureranges. Sensors or signals which exhibit thisbehavior are said to be non-linear. A hypo-thetical non-linear transfer function is shownin Figure 3. This figure illustrates the conceptof ‘terminal-based linearity’ which is thedeviation of the actual characteristic from astraight line coinciding with the actual char-acteristic endpoint (terminal) values.
Several families within the Dataforth productoffering – such as the DSCA and DSCT DINrail products, SCM5B and SCM7B plug-inpanel products, and SCT instrument headmount products – can be used in the follow-ing examples. For simplicity and uniformity,we have referred to the SCM5B familythroughout the tutorial.
Several of the Dataforth SCM5B series mod-ules have the ability to create a non-lineartransfer function through the module itself.This non-linear transfer function is config-ured at the factory and is designed to be equaland opposite to the sensor or signal non-lin-earity. The net result is that the module out-put signal is linear with respect to a giveninput parameter such as temperature. An out-put signal which has been linearized with
hardware internal to the SCM5B moduleseliminates the need for tedious software rou-tines which determine a linearized signalthrough the use of high-order polynomials orlook-up tables.
A hardware piece-wise linear technique isused in the SCM5B modules to correct thenon-linearity of the signal. The differencebetween the sensor non-linearity and the lin-earization provided by the SCM5B module iscalled the ‘conformity error’. This is adescription of how well the linearizationtechnique ‘conforms’ to the non-linear curve.Breakpoints are placed along the curve so asto equalize the positive and negative con-formity errors. SCM5B modules use 9 break-points (10 segments) to correct non-linearity,achieving typical conformity of ±0.015% of
span. A normalized plot of sensor non-linear-ity and hardware linearization is shown inFigure 4. Only three breakpoints are shownin the diagram for simplicity. It should beclear that an increase in the number of break-points used will result in a decrease in con-formity error.
Linearization of a given input is based uponthe input minimum and maximum values.For any input within these limits, the outputof the module will be a linear representationof the input. If the input exceeds the mini-mum or maximum values, the output of themodule is no longer a linear representation ofthe signal. This is also shown in Figure 4.Operation of an SCM5B module beyond thespecified input span is not recommendedbecause the output is difficult to calculate
Figure 3. Sensor Terminal-Based Linearity
Figure 4. Hardware Piecewise Linearization Using Three Breakpoints
5
and will have increased conformity error. If astandard module input span does not meetcustomer requirements a custom module canbe easily designed for optimum performancein a given system. Dataforth supplies stan-dard models for most thermocouple and RTDtypes. Consult Dataforth for details on cus-tom modules or non-standard ranges.
SENSOR CLASSIFICATIONSensors can be classified in many ways. Onemethod is to separate them into sensorswhich supply a voltage or current output bythemselves and sensors that must have anexternal voltage or current applied to pro-duce a useful signal. The first type of sensoris called self-excited. The second kindrequires ‘external excitation’. These seemlike small distinctions, but many sensorsrequire external excitation and the quality ofthat excitation directly bears on the quality oftheir output signals. Therefore, the quality ofexcitation is part of the overall design andapplication of signal conditioning modulessuch as the SCM5B series.
Sensors may also be grouped according tothe basic measurement they make. Some arestraight-forward in their classification, butothers can be used for a number of differentmeasurements. For example, the strain gageis really just a variable resistor, but it can beused to measure stress, strain, weight, pres-sure and acceleration. The most commonindustrial measurement is temperature.
TEMPERATURE SENSORSThree types of sensors are most commonlyused to measure temperature in industrialenvironments: thermocouples, resistancetemperature detectors (RTDs) and thermis-tors. Each has its unique advantages, disad-vantages and signal conditioning require-ments.
Thermocouples (TCs)Thermocouples are inexpensive, proven sen-sors and provide the widest range of temper-ature measurement. TC’s generate their ownsignal and do not need excitation. They areamong the most numerous sensors used.
TC’s do have some drawbacks. Substitutingone thermocouple for another one of thesame type can produce a slightly differentoutput voltage, forcing recalibration of thesignal conditioner for best accuracy.
TC’s may also be contaminated by the enviri-onment. This contamination is frequentenough that virtually all thermocouple signalconditioners provide open-circuit or ‘burn-out’ detection. Dataforth’s SCM5B37 andSCM5B47 thermocouple signal conditioningmodules can be configured for either upscaleor downscale burnout detection.
TC operation is based on two physical prop-erties. When a metal rod is heated at one end,a small voltage (Thomson EMF) develops
between the hot and cool ends. If two dis-similar metals are joined and heated at theirjunction but not connected at the unheatedends, a similar EMF occurs. This is called thePeltier EMF. The magnitude and polarity ofthe Peltier EMF are dependent on the tem-peratures of the junctions and the combina-tion of the two metals involved.
The algebraic sum of the Thomson EMFsand the Peltier EMF appear at the unjoinedends of the metal pair. This EMF is the basisfor all thermocouple operation and is calledthe Seebeck EMF. If both the joined and openends of the metallic pair are at the same tem-perature, this EMF is zero. If the tempera-tures at the open ends are equal and kept con-stant, the Seebeck EMF is a direct function ofthe temperature at the measurement junctionand can be used to measure that temperature.The Seebeck EMF depends on the TC’s com-position and ranges from 10 to 80mV forfull-scale output. See Table 1 for commonthermocouples and their measurementranges.
Figure 5. Type 1 Non-Linearity
Table 1. Thermocouple Types
TYPE COMPOSITION MEASUREMENT LIMITS (°C)
J Fe vs Cu-Ni -210 to 760
K Ni-Cr vs Ni-Al -270 to 1372
T Cu vs Cu-Ni -270 to 400
E Ni-Cr vs Cu-Ni -270 to 1000
R Pt-13% Rh vs Pt 0 to 1768
S Pt-10% Rh vs Pt 0 to 1768
B Pt-30% Rh vs Pt-6%Rh 0 to 1820
C W-5% Re vs W-26% Re 0 to 2320
N Ni-14.2% Cr-1.4% Si vs Ni-4.4% Si-0.1%Mg -270 to 1300
6
TC outputs have non-linear relationships(Figure 5) to the measured temperature. Inaddition, each thermocouple type has its ownnon-linear characteristic. This makes it diffi-cult to provide universal linearization for themany types of TCs available. However,Dataforth’s SCM5B47 series provides hard-ware linearization and has a complete selec-tion to match all popular TC types.
Thermocouple accuracy depends on the com-position and purity of their metals and theirfabrication. Usually a thermocouple will notbe more accurate than 0.5% to 1% of its totalmeasurement range. This translates to ameasurement error as large as 2°C for someTC types. TCs almost always require ampli-fication and cold-junction compensation.
It is important to note that thermocouplesalways indicate the difference between twotemperatures at two junctions. The measure-ment junction is the one whose temperatureis of interest. The other junction is eithermaintained at a reference temperature (0°C)by physical means or this condition is simu-lated electronically. It is called the referencejunction.
Physically maintaining one or more refer-ence junctions at a constant temperature isnot an easy or desirable solution for industri-al measurements. Instead, the reference junc-tion is created by bringing the measurementthermocouple wires to the amplifier and con-necting them to a terminal block. This termi-nal block is often called an iso-block (short-hand for isothermal terminal block). Its highthermal conductivity assures that the termi-nals for the thermocouple wires are at thesame temperature. SCM5B modules use thevery predictable voltage drop of a silicondiode to measure the terminal block temper-ature and imitate a thermocouple in an icebath at 0°C. This entire process is calledcold-junction compensation and the circuit iscalled the cold-junction compensator (CJC).
While the technique sounds complicated, it iseasy to implement electronically. Figure 6shows a block diagram. The electronic com-pensation is usually 2-3 times more accuratethan the thermocouple itself. It allows preci-sion instrumentation to provide accurate tem-perature readings even when the ambient(cold-junction) temperature moves throughlarge swings occurring in industrial applica-tions.
Resistance Temperature Detectors (RTDs)RTD’s are among the most well-behaved andprecise temperature measuring devices avail-able for industry. An RTD is a precisionresistor with a well-defined resistance vs.temperature curve. RTDs are classifiedaccording to their material composition andchange in resistance versus temperature(Alpha Value). See Table 2.
Copper, nickel, and platinum RTDs enjoywide-spread use, although the platinum RTDis now almost universally specified for new
industrial installations.The platinum RTDoffers many outstanding characteristicsincluding high accuracy, wide measurementrange and chemical resistance to many of thenastier atmospheres in industrial applica-tions.
The PT1OO is the most commonly usedRTD curve and displays an ice-point resist-ance of 100 ohms. It is based upon severalEuropean standards and is supported by theIEC as well. PT200, PT500 and PT1000RTDs have appeared on the market recently,but they are simply multiples of the basicPT100 curve in their behavior. That is, aPT500 sensor will have five times the resist-ance of the PT100 at the same temperature.The higher resistance allows use of lessmaterial and provides a cost savings. It alsocan provide a smaller sensor. If, however, thesmaller sensor is used, careful attention must
be paid to the level of excitation current used.Too much current will cause the sensor’ssmall mass to self-heat and degrade its accu-racy. Dataforth’s SCM5B34 and SCM5B35signal conditioning modules specifically use0.25mA excitation current to eliminate thisproblem.
The D1OO curve is a commonly used USRTD curve and is also supported by aJapanese standard. For many years, theSAMA (Scientific Apparatus Manufacturer’sAssociation) curve was popular used in theUS, but it has since fallen into disuse. ThePT1OO curve now dominates, finding use in85-95% of the industrial applications.
RTDs can be used in several configurationsthat reflect the cost factors and the degree ofaccuracy desired. Figures 7,8 and 9 showSCM5B34 and SCM5B35 modules in 2-wire, 4-wire, and 3-wire connections respec-tively. The 2-wire configuration is used whensignal lines are short, highest accuracy is notrequired, and lowest cost of installation is ofparamount importance. Because RTD’s mustbe excited by a current, signal-line resistancewill appear in the apparent resistance of theRTD. If the resistance is high, or if the tem-perature coeffecient of the line resistance ishigh, errors will occur, giving a false temper-ature reading.
The most accurate connection method is toexcite the RTD by using two power leads tocarry the excitation current, and using twoadditional wires to sense the voltage at theRTD directly. If high impedance circuits areused to measure the voltage on the secondpair of wires, no appreciable current willflow through them and the voltage measuredwill only be that at the RTD. This setup,shown in Figure 8, is used in almost all labo-ratory or other high-accuracy situations. TheSCM5B35 is specifically designed for thisapplication.
Table 2. RTD Types
Figure 6. Cold Junction Compensation
DESIGNATION MATERIAL ICE POINT RESISTANCE ALPHA VALUE
A reasonable compromise between 2-wireand 4-wire connections is the three-wire con-nection. It offers high accuracy and a lowercost of wiring. This connection is the onemost frequently used in factory or plantinstrument systems. Figure 9 gives thedetails. By connecting the wires as shown,equal currents flow from balanced currentsources in the SCM5B34 through the groundwire, and back through the top and bottomRTD connections. If the line resistances areequal, the voltage drops from the amplifierinputs to the top and bottom of the RTD willbe equal and the error voltages added to eachline will also be equal. The two line voltagedrops cancel, and the differential input to theamplifier is the actual voltage at the RTD.The voltage drop in the common (ground)wire will be twice that in the other two linesand will add to both amplifier inputs equally.Thus it appears as a common mode voltage tothe amplifier inputs. One may question thebasic assumption that the line resistances areequal; using twisted three-wire cable helpsmake this assumption correct.
One of the few drawbacks to the use ofRTD’s is their nonlinear change in resistancewith temperature. Figure 10 shows the non-linearity of a PTIOO RTD measuring from0°C to 450°C: it is almost 2% of span.Dataforth SCM5B34 and SCM5B35 mod-ules provide hardware non-linearity correc-tion to 0.015% typical.
The RTD has become the preferred sensor fortemperature measurement when accuracy,repeatability and interchangability arerequired. The PT100 RTD has a uniformnon-linear resistance curve, is resistant tomost harsh environments, and is particularlyrobust for industrial measurements.
ThermistorsThermistors are relatively inexpensivedevices exhibiting very large changes inresistance for small changes in temperature(typically 4-6%/°C). For example, a typicalthermistor may have a nominal resistance of30kΩ at 25°C, but have a resistance of 2.5kΩ at 85°C. This large change in resistancemakes line resistance to the thermistor a verysmall source of error and use of the thermis-tor can avoid the three-and four-wire connec-tions common with the RTD.
In the past, thermistors have had very poorinterchangeability. That is, an instrument cal-ibrated for one thermistor would requiremajor recalibration when a new thermistorwas substituted. Also, the relationship
Figure 7. Two-Wire RTD Connection (SCM5B34)
Figure 8. Four-Wire RTD Connection (SCM5B35)
Figure 9. Three-Wire RTD Connection (SCM5B34)
Figure 10. PT100 Non-linearity
8
between temperature and resistance couldvary significantly from lot to lot and evenbetween sensors from the same manufactur-ing lot. Low-cost thermistors still retain thesecharacteristics.
Today, manufacturers can supply thermistorswith vastly improved interchangeability. Thisfact has given the thermistor a boost as a seri-ous process temperature sensor. This inter-changeability has its limitations, however.The temperature span over which inter-changeability exists lies between 50°C and100°C of the thermistor’s total measurementrange. As an example, some manufacturerscan achieve 0.1°C interchangeability from0°C to 70°C or from 120°C to 180°C.Thermistors are also somewhat limited intheir absolute temperature range. Realisticlimits lie between -100°C and 450°C. Thiscan eliminate them from consideration forsome industrial measurements, but they stillremain ideal for low-cost sensing within theirtemperature range.
Thermistors are very nonlinear devices.Figure 11 shows the behavior of a typicalthermistor. Thermistors can, however, beused in or purchased as part of a resistor net-work whose output is highly linear over mostof the useful temperature measurement rangeof the device. Figure 12 shows a resistor-thermistor network whose behavior is quitelinear over a modest range as shown inFigure 13. Note that the network resistancenow approaches that of some RTDs, and thethree- and four-wire connections used withRTDs may be necessary for best accuracy.Dataforth offers several signal conditioningmodules for thermistors. Because of the largevariety of thermistors available, consultDataforth for details.
One of the big advantages of thermistors istheir small size. This gives them perhaps thebest thermal response time of almost anytemperature sensor. Some can react in mil-liseconds to large temperature variations.The down side of this small mass is self-heat-ing of the sensor from the excitation current.This heating can pose a significant source oferror. Since thermistors are typically high-resistance devices, large currents will pro-duce large self-heating errors. Pay closeattention to the manufacturer’s recommenda-tions in this area.
MOTION SENSORSMotion sensors are useful in a number of dif-ferent applications. The two most commonmotion sensors are based on changes in
resistance. These are the slidewire and straingage. Another fairly common motion sensoris the accelerometer, which is used to meas-ure dynamically changing motion. All ofthese sensors are very broad in their applica-tions. For brevity, they will be treated froman interface and signal conditioning view-point only. Specific measurements and tech-niques will not be discussed.
SlidewireA rotary or linear resistor may be used tosense motion. This motion can be the smalldisplacement of a pressure diaphragm, thestroke of an hydraulic piston or shaft rota-tion. If voltage excitation (5 - 10VDC) and apotentiometer connection are used, the out-put will be a fraction of the excitation voltage
Figure 11. Typical Thermistor Characteristic
Figure 12. Linear Thermistor Network
Figure 13. Performance of Figure 12 Thermistor Network
9
and the measurement is referred to as aslidewire measurement. These signals usual-ly are 0-5VDC or 0-10VDC. If the motionbeing measured is non-linear, the relation-ship between motion and resistance of thesensor can be manufactured to provide a lin-ear signal.
If constant-current excitation and a rheostatare used, the resistance change can be scaledinto almost any engineering unit. For exam-ple, linear resistors are available in 1 to 20inch lengths. The resistance of these devicesis usually a constant value per inch. As anexample, if that value is 1kΩ/inch, and a0.25mA current is used, linear measurementscould be made for distances between 1 and 20inches. The scale factor would be a constant0.25VDC/inch regardless of the physicallength of the linear resistor. See Figure 14.
Slidewire sensors provide an inexpensivesolution for many measurement problems.Their main faults are noise introduced by thesliding contact and wear of the resistance ele-ment. If the measured variable changes bysmall amounts, the sliding contact is fairlystationary. The small changes that do occurcause a scrubbing action at a single point onthe element. A clear example of this problemappears in the data sheet of any slidewire-based pressure sensor where the sensor willbe lifetime rated in numbers of pressurecycles. Since most slidewire sensors measureslowly varying variables, the low bandwidthof the SCM5B36 potentiometer module isparticularly useful in eliminating contactnoise. This module also supplies constant-current excitation suitable for these sensors.Mechanical wear can be addressed only byreplacement of the sensor.
STRAIN GAGESThe strain gage is a special adaptation of theresistive motion sensor. Rather than movinga contact along the resistor, the strain gageitself is lengthened or shortened. The resist-ance of a conductor is proportional to itsresistivity and length and inversely propor-tional to its cross sectional area. As the wireis stretched, its length increases and its cross-section gets smaller. Both of these changesincrease its resistance.
If the wire had been placed in tension andthen relaxed, its resistance would decrease.The change in resistance for small changes inlength is directly related to the tension in thewire. This effect is not precisely linear but isvery repeatable. Gages made with stretchedconductors have measured the elongation or
contraction of bridge structures, the flexureof aircraft wings, and the stress and strain inthe airframe of the Space Shuttle. They havealso been used for more than a half century tomake pressure, force, weight, and accelera-tion measurements.
Practical Gage ArrangementsThe strain gage converts all of these physicalquantities into a changing resistance byresponding to the stress or strain induced in amechanical member to which it is attached.For example, a pressure diaphragm stretchesinward when pressure is applied. This motionstretches or contracts a strain gage attached tothe diaphragm and increases or decreases itsresistance. Measuring the resistance providesa corresponding measure of the pressure.
The change in resistance of a strain gage issmall and a method of detecting the change is
more useful than measuring the total gageresistance. Figure 15 shows the most com-mon way to detect the change. The straingage is connected as one arm of aWheatstone bridge. An excitation voltage orcurrent is applied and the bridge is initiallybalanced for zero output. As the gage isstressed or strained, the bridge unbalances bythe change in gage resistance. This can bedetected by a sensitive voltmeter connectedacross the center arms of the bridge. Furthersensitivity (more output) results when two ormore of the bridge resistors are ‘active’.Also, using two- or four-arm active bridgesresult in a more linear signal. Figure 16shows quarter, half and full bridge arrange-ments when connected to SCM5B38 straingage modules. Opposing arms are placed intension and compression (stressed andstrained). The full bridge will produce four
Figure 14. Slideware Example
Figure 15. Quarter Bridge Strain Gage
10
times the signal of the quarter bridge and itsoutput will be electrically linear with thechange in resistance of the identical arms.The success of the sensor designer in produc-ing identical stress/strain in each of thebridge arms determines how linear the signalwill be with changes in the physical quantity.Practically, strain gage manufacturers areable to achieve this goal within 0.1-1.0%accuracy.
Gage TypesThe original wire-based sensor has not beenused industrially for a long time. It has beenreplaced by a number of strain elements thatwork in the same way but offer advantages ofcost, accuracy, lower power consumption ora combination of these factors.
Bonded strain gages are metal film resistorsmounted on carriers or substrates. These sub-strates attach to the stress member with adhe-sives. Because the adhesive does not perfect-ly affix the strain gage to the stress member,bonded gages are subject to a phenomenoncalled ‘creep’. Minute shifts of the gage(s) onthe stress member appear as small drifts inzero and full scale calibration. Bondedgauges require relatively large amounts ofpower (5-10VDC, 25-125mA). These are notinsurmountable problems, and the manufac-turing and signal processing technologyavailable for these types assure a qualitymeasurement. Nonetheless, these and othertraits have led to greater use of piezo-resis-tive silicon and vapor-deposited metal filmstrain gages.
The silicon gage combines the stress memberand strain element in one single structure.The silicon gage has no intermediate materi-als and pure silicon makes a very good stressmember: It has nearly perfect elasticity andhas about 10 times the output of a bondedgage while using less power. It also is aninexpensive gage to manufacture.
The disadvantages of silicon strain gagesarise from material limitations and tempera-ture sensitivities of the resistors. Siliconfunctions well as a gage material between -100°C and 200°C. Silicon cannot withstandexposure to many chemicals or corrosiveatmospheres and must be protected by othermaterials. This raises the cost of some sen-sors based on the silicon gage. Silicon gagesdisplay strong zero and full-scale tempera-ture dependencies. Sensor compensationtechniques can largely make up for thesedependencies, but add a small cost penalty.
Vapor-deposited strain gages offer a compro-mise between the monolithic structure of thesilicon gage and the advantages offered bythe bonded film gage. This gage is formeddirectly on the surface of the stress memberby first vapor-depositing a substrate on thestress member and then vapor-depositing ametal film pattern onto the substrate. Thismethod provides a structure that is nearly as
monolithic as the silicon gage. Also, thematerials used withstand higher tempera-tures. By controlling the amount of metalfilm deposited, the resistance of these gagescan be adjusted to between 3-15kΩ for lowpower consumption. This makes the vapor-deposited gage strong competition for the sil-icon gage, especially in 4-20mA transmitters.
Figure 16. Strain Gage Connections to Signal Conditioning Modules (SCM5B38)
11
The disadvantage of this technology lies pri-marily in its low output voltage — a trait itshares with the bonded strain gage. However,because it can be fabricated in batches andlaser-trimmed for final zero and full-scalesensitivity, this gage offers many of the low-cost advantages of the silicon gage andexhibits smaller temperature effects than thesilicon gage.
The signal conditioning used with straingages must provide exceptionally stableexcitation and high sensitivity. Some bridgesprovide as little as 6.667mV full scale outputwhen excited with 3.333VDC (20mV at10VDC excitation). The stability of the exci-tation source direcly controls the stability ofboth zero and full scale output. The DataforthSCM5B38 offers the excitation stabilityrequired, and provides stable gain for pro-cessing of these low-level signals. An addi-tional bonus is the isolation of the excitationsource from the rest of the module. It isimmune to line-voltage and transient dam-age. It is also less sensitive to bridge loadsthan most available modules.
AccelerometersAccelerometers fall into two general classes.The first is based on the strain gage in one ofits many forms. In this sensor a mass isattached to the stress member and exerts aforce proportional to the acceleration whichit experiences. A strain gage then convertsthis stress into an electric signal. The outputof the strain gage is handled just as it is for apressure or force measurement. Accelero-meters made in this fashion are able to meas-ure extremely low rates of change in acceler-ation, but tend to be limited in bandwidth.This limit may be as high as 25kHz.
The second type of accelerometer makes useof the strong piezo-electric effect displayedby some crystalline and ceramic materials.Under deformation, these materials develop avoltage. This effect can be used to measurethe force experienced when a mass attachedto these materials accelerates. Because thesematerials are good insulators, they behaveelectrically like a capacitor in series with avoltage and little or no current can be drawnfrom them without losing accuracy. Thecapacitance of a long cable can severelyaffect their measurement characteristics.Therefore, piezo-electric sensors are usuallyamplified directly at the sensor. Since thevoltages available from these devices can bevery large, the amplifier is usually an imped-ance-matching device. That is, it provides a
very high impedance (resistance and capaci-tance) to the sensor and provides a low-impedance output able to drive long cables.See Figure 17 for a charge amplifier whichcould be used for bearing vibration (seismic)monitoring.
High frequency response (100kHz-300kHz)can be achieved with piezo-electricaccelerometers, but they tend to be very lim-ited at low frequencies. Constant improve-ments are being made, so be sure to checkmanufacturers’ specifications.
LOOPS AND ANALOG SIGNALSSensors are connected to signal conditioningequipment, data acquisition or control sys-tems, and control devices. This collection ofequipment is known in the industry as ameasurement loop or, simply, a loop. Anexample of a such a loop was shown inFigure 1.
A loop may be classified by the measurementit is making, by its use, or by the type of elec-tronic signal carrying the measurementbetween the sensor and the rest of the equip-ment. Except for purely digital systems suchas FIELDBUS, loops make use of voltage,current or frequency signals. Most use volt-age or current. The amount of power requiredto make the measurement or to power thesensor influences the choice of loop configu-ration. For example, a gas chromatographused for chemical analysis requires morepower than is available from a 4-20mA loopand is usually mains-powered in a four-wireconfiguration. Other measurements are self-excited or can be powered by low energysources. In this case, two-wire transmission
offers cost advantages regarding wiring costsand simplifies the use of intrinsic-safetytechniques, saving further in installed cost.
MEASUREMENT LOOPCONFIGURATIONSMeasurement loops are configured in one ofthree ways: 2wire, 3wire and 4wire. Someloops do not require excitation or local signalconditioning or have outputs which may behandled directly by the data acquisition sys-tem. Examples of the former are thermocou-ples, while the latter are represented bypotentiometers driven by pressure elements.The greater number of measurement loops dorequire some form of local excitation andsignal conditioning. Merits of each signalconditioning configuration vary according tothe needs and cost-sensitivity of the applica-tion.
Four-Wire LoopsFour-wire loops (Figure 18) provide thegreatest degree of flexibility because they aresupplied with power independent of the sig-nal return lines and are therefore not limitedin power consumption or by signal groundinduced errors. These reasons are the primaryones for choosing this configuration. They doimpose a cost penalty because they cannot beeasily installed using intrinsically-safe tech-niques and do require the running of extrawiring, which is a major cost in instrumentloop installation. However, for some meas-urement devices, the four-wire loop is theonly alternative. Some sensors and all ana-lyzers simply require too much power to berun in less costly configurations.
Figure 17. Charge Amplifier
12
Three-Wire LoopsThree-wire loops (Figure 19) offer some ofthe flexibility of the four-wire configurationand the savings and simplicity of one lesswire to install. They do suffer from a certaininflexibility in that both power and signalground are shared by the same wire. If sub-stantial power or long line lengths areinvolved, the voltage drop (I x RRETURN) in theshared ground return can substantially inter-fere with the accuracy of voltage-mode sig-nal transmission. In current-mode and fre-quency-mode signal transmission, this con-straint can be avoided.
Two-Wire Loops When considering two-wire loops, most peo-ple think of the 4-20mA loop, but othermeasurements fall into this category also.These other cases involve any sensor whichgenerates its own signal and could includethe thermocouple, pH, and ORP (oxygenreduction potential) sensors. These sensorsoperate in the voltage mode, usually haveoutput signals of moderately low-level (0-500mV), and are very susceptible to inducednoise. These sensors experience impedancerestraints regarding either leakage resistancein the instrument wiring or input characteris-tics of the signal-conditioning input.
The current-mode, two-wire configuration(Figure 20) overcomes many of these con-straints. It allows local signal conditioning,and because it operates in a manner that isrelatively immune to induced noise, hasbecome the de facto standard for most criti-cal process-control measurements. This con-figuration also enjoys economic advantagesarising from reduced wiring costs and theability to easily use intrinsically safe installa-tion practices.
It does impose constraints on the designer ofboth the transmitter and data collectionequipment. First, the transmitter must per-form all functions within a limited powerbudget. Sensor excitation, amplification andsignal conditioning functions must beaccomplished with 4mA or less at voltagestypically ranging from about 12 to 36 volts.At the receiver, the signal usually must beconverted into a voltage and level-shifted toa zero-based range so that subsequent con-version into digitized format may be doneusing the full input capability of the analog-to-digital converter. Both the current-to-volt-age conversion and level-shifting can intro-duce extra errors in both zero and span ifproper care is not given to circuit design.
ANALOG SIGNALS
Voltage-Mode LoopsVoltage-mode loops present to the dataacquisition system a voltage proportional tosome physical measurement. The nature ofthe voltage source, its impedance, length ofwiring, and the nature of the receiving instru-mentation all affect the accuracy of the final
measurement. Figure 21 shows some of theelements to consider when using a voltage-mode loop. The representation is general anddoes not reflect whether the voltage loop is 2-, 3- or 4-wire in nature.
The measurement voltage may be directlyfrom a self-exciting source (e.g., thermocou-ple) or from a signal conditioning amplifier.
Figure 18. Four-Wire Loop
Figure 19. Three-Wire Loop
Figure 20. Two-Wire Loop
13
Each voltage source will have some internalimpedance which acts as if it is in series withthe actual signal. The signal lines will alsohave some small impedance, and the receiv-er will present some load to the voltage sig-nal. These elements are shown as lumpedcomponents in Figure 21. The sensor, lineand receiver impedances form a voltagedivider that can affect the overall accuracy ofthe measurement. For this reason, the receiv-er impedance is made as high as possible. Forthermocouple, RTD, or strain gage measure-ments (about 10-300mV) or for precondi-tioned signals, the required impedance is ofthe order of 100MΩ. For pH, ORP,accelerometer or photo diode applications,the receiver impedance must be greater than200MΩ. In these latter applications, theeffects of humidity and integrity of wiring oncable leakage resistance make signal condi-tioning at the point of measurement almostimperative.
Because voltage-mode loops operate at rela-tively high impedances, their chief suscepti-bility to noise arises from electric fields andground potential differences. Electric fieldsarise from nearby natural or man-madesources. The most common sources includepower equipment and switching power sup-plies. Use of receiver designs offering highcommon-mode rejection and common- andnormal-mode filtering offers a measure ofimmunity, while appropriate shielding of sig-nal lines decouples the signal lines from theinterfering source. Using point-of-measure-ment signal conditioning to raise the level ofsignal voltage improves the ratio of measure-ment signal to interfering noise.
Ground potential differences create poten-tially larger interference than electric fieldsbecause they cannot be easily shieldedagainst and because many sensors are tied tolocal ground. Typical sensors for which thisis true are thermocouples and pH or ORPprobes. This local ground can, on a transientbasis, be several hundred volts different fromthe ground potential at the receiver. A plantgrounding system designed at the time ofplant construction can be of benefit, but isoften not available in old or retrofit installa-tions. Use of receivers with high common-and normal-mode rejection also offer somehelp in these difficult situations, but trans-former isolation provides the greatest relieffrom this problem. When in doubt, isolate!
Among the chief advantages of voltage-mode loops is their ability to be multiplexeddirectly. The expensive signal-conditioning
and data conversion equipment can be sharedwith many loops, reducing the per-loop costof these functions. In most configurations,they do not require zero offsetting and don’tneed a precision resistor for conversion ofcurrent to voltage.
Current-Mode LoopsCurrent-mode loops offer relatively highimmunity to electrostatic interference, butare susceptible to magnetically-inducederrors. The simplest way to avoid these errorsis to use twisted pair signal lines and to main-tain separation from power or relay controllines or other sources of magnetic induction.As with voltage-mode loops, current-modeloops can also suffer from ground potentialproblems. Again, transformer isolation is thebest remedy.
While current-mode loops offer the betternoise immunity of the two types, they dorequire additional conversion into a voltageat the receiver, and most often require level-shifting to return them to a zero-based volt-age signal. In particular, the two-wire 4-20mA loop is difficult to multiplex directlybecause of the need to keep power on the
loop to avoid amplifier settling-time prob-lems. Figure 22 shows a typical current loopconfiguration. Also, most multiplexers expe-rience difficulty with the power supply volt-ages used in these loops. The conversion andlevel-shifting is often performed before mul-tiplexing occurs. This imposes an additionalcost at the receiver. Even so, for many criti-cal control applications this loop type hasbecome standard because of its superiornoise immunity.
Frequency-Mode LoopsFrequency signals are used less often thanvoltage or current signals, but do offer somedistinct advantages. A frequency signal isessentially a digital signal and is relativelyinsensitive to noise and interference fromoutside sources. It is sometimes used for thisreason alone. However, there are some sen-sors that inherently generate a repetitivepulse train. Such sensors as turbine meters,flowmeters based on vortex shedding, ortotalizers are ‘natural’ for frequency trans-mission. Dataforth’s SCM5B45 FrequencyInput modules can accept either TTL level orAC signals and convert their frequency into aproportional 0-5VDC or 0-10VDC output.
Figure 21. Typical Voltage Loop
Figure 22. Typical Current Loop
14
Full scale inputs as high as 100kHz are sup-ported.
Choice of Signal LevelWhile there is no universal rule for choosing thetype and level of signal input to the receivingequipment, there are some practical considera-tions where large numbers of measurementsmust be dealt with. First, the wiring costs,including labor, are a major part of the cost ofownership of instrumentation. Second, the abil-ity of the system to detect abnormal wiring situ-ations such as opens or shorts is often crucial.Third, with lost-production costs in modernplants running from $20,000 to $500,000 perhour, it is prudent to repair by replacement andthen fix the defective piece of instrumentationoff-line. This is possible only if spares are athand. Standardization on one or two input typesreduces the costs associated with carrying thespares inventory of instruments and input mod-ules. Table 3 lists many of the standardized sig-nal levels in use today.
SIGNAL INTEGRITYThere are many pieces of equipment in boththe field and control room that can interferewith measurement signals. This equipmentand other sources of man-made and natural‘noise’ are a part of the ‘electronically haz-ardous’ environment in which instrumenta-tion exists. Since the equipment and naturalevents themselves cannot be eliminated, theireffects upon the instruments must be under-stood and minimized or eliminated. Byunderstanding the ways in which this noisegets into the instrument system, one can takerational steps to avoid problems in a new sys-tem or eliminate them in an existing one.Table 4 list the most common problems andpossible solutions.
SOURCES OF ERROREach of the following influences can serious-ly degrade signals. Methods to avoid or min-imize their effects are discussed in later sec-tions.
Capacitive Coupling Any piece of plant equipment can develop anelectric charge. So long as this charge doesnot change, it has little or no effect on aninstrument system. However, all electricallypowered equipment has a varying charge orvoltage. It can vary in a smooth or erratic(transient) manner. When it does so, theequipment creates a changing electric fieldthat can capacitively couple into the sensor,
its signal conditioning equipment or itswiring. Man-made static discharges andlighting have been known severely damageinstruments.
Magnetic CouplingAn electric current produces a magneticfield. If a conductor moves through the mag-netized field, a current is produced in the
conductor. Similarly, if the electric current ina conductor varies, a current is induced innearby stationary conductors. The resultinginduced current can be a disturbing influenceor can produce a voltage across the ends ofthe conductor, also producing an influence.Since most sensor wiring is fixed in place,varying currents are the usual cause of mag-netic coupling in instrumentation systems.
Table 3. Common Signal Types and Levels
Table 4. Sources of Error and Possible Solutions
SIGNAL TYPE STANDARD RANGESVoltage 0 through 500mV (non-standard ranges for TC’s, RTD’s and others)
0-1V0-5V1-5V0-10V±1V±2.5V±5V±10V
Current 0-1mA0.2-0.5mA0-20mA4-20mA10-50mA (obsolete, but still in use)
Frequency 0-500Hz throught 0-100kHz Full Scale(no standards adopted)
Often these current influences result fromplacing wires too close to power conductors.
Ground LoopsGround is an elusive and often misunder-stood electrical concept. Its very nameimplies that the soil we walk on is the placeto which all currents and voltages are some-how referred. In an electric power distribu-tion system, a rod driven into the earth or aburied metal pipe is ‘ground.’
Unfortunately, that is not the entire the story.The local ‘ground’ where you are now locat-ed can be several volts above or below that atthe nearest building or structure. If there is anearby lightning strike, that difference canrise to several hundreds or thousands of volts(Figure 23). Even in a home, different partsof the electrical grounding system can varyby several volts. This arises not only from theresistance of wiring but its inductance. If thecurrents change very rapidly, the voltagedrops in the ground system will approachseveral hundred volts for short periods oftime. For example, studies conducted bypower companies have shown that the opera-tion of an oil burner ignitor can produce tran-sient differences up to 2000 volts routinely.Imagine the potential for similar voltagespikes in other industrial environments.
The voltages themselves can obviously pro-vide a great source of interference for ameasurement loop, but the currents whichcause them can also induce significant cur-rents and voltages in the signal wires locatednearby. Circulating currents in ground loopsmay also be periodic in addition to beingtransient events. Consider the case when aground loop is formed with AC power linessuch that a 50 or 60Hz signal is imposed onthe gound loop. If the measurement signalground is part of the ground loop, theunwanted AC signal will appear as an errorvoltage or perhaps as a common mode signalto the system inputs. Figure 24 shows a10mV thermocouple signal which is corrupt-ed by a 60Hz and a 180Hz (3rd harmonic)ground loop voltage. The original signal hasbeen corrupted beyond recognition. Propergrounding practice and use of a signal condi-tioner with high common mode rejection isvital in preserving signal integrity (Figure25).
Over Voltage and TransientsBeside the transient voltages caused byground potential differences, large voltagescan appear in the signal wiring directly. This
can arise from capacitively or magneticallyinduced sources, from accidental electro-static discharges or from nearby high-voltagearcs such as welding. Some of the instrumen-tation installed in today’s industrial plant iswired in close proximity to power wiring.Although this is very poor instrument wiringpractice, it does happen. Accidental connec-tion of signal conditioner inputs to 110 or240 VAC is not an uncommon event.
Many manufacturers of industrial equipmentprovide input circuit protection to preventdamage from transient voltages and the mis-application of power line voltages. Suchabnormal voltages will render the measure-ment useless during the time they exist, but
will not damage the equipment if it has thisprotection. When specifying signal condi-tioning equipment, choose equipment thatprovides SWC (surge withstand capability)protection and line voltage protection.Compliance with ANSI/IEEE C37.90.1-1989standards is a good indicator that the equip-ment supplier has provided adequate protec-tion (Figure 26a, b). All Dataforth SCM5Bproducts are designed to meet this specifica-tion.
EMI and RFIElectromagnetic interference (EMI) is a gen-eral term for induced errors signal loops aresubjected to. It includes most of the sourcesdiscussed before. However, it has come to
Figure 23. Differences in Ground Potential
Figure 24. Ground Loop Signal Protection
Figure 25. Ground Differences Cause Uncertain Results
16
have a somewhat more restricted usage in theinstrument business. It has become synony-mous with the term RFI (radio frequencyinterference). The usual source of this partic-ularly annoying disturbance is a radio trans-mitter. If it is from a nearby radio or televi-sion station, it can at least be diagnosed withsome ease. More often it appears as a ‘ran-dom’ or intermittent change in the measure-ment signal. In this situation, the usualsources are CB transmitters or two-wayradios used in the plant. The normal signalconditioner will not amplify these signalsbecause they are too high in frequency.However, the input stages of some signalconditioners will rectify the RF voltage in asimilar to the old ‘crystal’ receivers. The rec-tified voltage appears as a drift or suddenshift in the measurement signal (Figure 27).Two-way radio conversations between thefield and control room locations can causeseemingly random changes in signal levels.Look for signal conditioning equipment thatspecifies EMI/RFI immunity. A typical spec-ification might state “less than 0.5% shift forRF fields of 30V/M between 27 and500MHz”. This RF field strength is roughlyequivalent to a 5 watt transmitter at 3 feet or1 meter.
Aliasing‘Aliasing’ becomes an issue when signalconditioning passes from the purely analogworld to the digital world. This, of course, isthe norm in today’s systems. The notion ofaliasing is confusing to some users of dataacquisition equipment. The need to preventaliasing arises from the use of analog-to-dig-ital conversion in the signal conditioningpath.
To understand the problem, consider the dataconversion system shown in Figure 28a. Thisis not a real system, because one does notusually re-convert a digital signal as shown.However, reconstructing an analog signalafter it has been converted allows us to seethe aliasing phenomenon directly in the fre-quency domain. The input signal containstwo frequency spectra centered around f1 andf2, as shown in Figure 28b. The sampling fre-quency is fs. The output spectrum for thissystem is shown in Figure 28c. The term‘Nyquist frequency’ comes up often in dis-cussions about data conversion systems. It’ssimply the sampling frequency divided bytwo and represents the highest signal fre-quency which can be processed in a perfectsystem without aliasing.
Figure 27. Electromagnetic Interference
Figure 26. ANSI/IEEE C37.90.1-1989 Test Waveforms
17
Signals, such as f2, that are above the Nyquistfrequency are present in the output sprec-trum, but also appear in the output at a lowerfrequency. These output signals are false rep-resentations or ‘aliases’ of the higher fre-quency input signals. Obviously, this is not afaithful reproduction of the input signal andcan lead to large signal distortions if not han-dled properly.
WAYS TO PRESERVE SIGNAL INTEGRITYSignal integrity begins with a quality meas-urement and good signal conditioning equip-ment. Despite the best equipment, however,signals can be degraded if the equipment isnot properly installed. Signal integrity is pro-foundly affected by shielding and wiringpractices. It can also be affected by systemdesign choices.
ShieldingCable shielding is used to minimize or elim-inate capacitively coupled interference andto aid in lowering RFI-induced errors.Shielding is arguably the most controversialsubject faced by instrument users. Some sayground everywhere; others say ground onlyat the source end and others defend ground-ing at the receiving end. The most constantly‘correct’ place to ground an instrumentwiring shield is at the receiver end. However,some manufacturers of transmitters andreceivers supply cables with shields connect-ed at the source and, short of remaking theinstrument, there is no choice but to acceptthis ground for the shield. One important ruleis that there should be only one ground forthe shield. Violating this rule automaticallycreates a ground loop and all of the problemsassociated with it.
Shielding consists of a metallic sheath sur-rounding the instrument wires. This isintended to be a Gaussian or equi-potentialsurface on which electric fields may termi-nate and return to ‘ground’ while leaving theinternal wires uncoupled to these fields. But,if two grounds exist, ground currents willflow through the shield which generate mag-netically-induced voltages in the signal leads.The shield meant to reduce capacitively cou-pled errors has become a source of magneticfields and introduced an additional source oferror. Avoiding this is as simple as followingthe rule above: use only one ground in thesystem (Figure 29). Sometimes the installa-tion and mechanics of the wiring do notallow the one-ground approach. In that case,
two steps help: use twisted signal wire pairsand isolation.
Twisted PairsUsing twisted wire pairs is the simplest wayto reduce magnetically induced interference.This applies for shielded or unshielded wiresand and for cases where the field is caused byshield currents or from other field sources.The induced voltage is proportional to themagnetic field strength and to the wire looparea through which it passes (see Figure30a). Twisting the wires forces them closetogether, reduces the loop area and lowers theinduced voltage.
Another benefit of using twisted pair is moresubtle. By alternating the positions of thewires in an electric field, the induced volt-ages in each incremental length of wire tendsto cancel the voltage in the adjoining lengthsof wire. Figure 30b shows an untwisted pairwith all induced voltages portrayed as onevoltage source, while Figure 30c shows thetotal effect of twisting the wires. Theenclosed magnetic loop area has beenreduced dramatically and magneticallyinduced errors are much reduced. Pickupfrom an electric field has also been reduced.Each incremental length of wire has its ownvoltage error from this source, but theseerrors oppose each other and tend to cancel.
Figure 28. Aliasing Errors
18
Cable SpacingWhen cables are located in close proximity,coupling can cause unintended pickup fromthe voltages and currents running nearby. It isoften tempting to use the same conduit orcable tray to run instrument wiring withpower wiring. Don’t do it! This practice is asure formula for added difficulty with instru-mentation signals. Keep the instrument wirestogether and apart from all power wiring.
IsolationIsolation is a universal way to eliminateground loop problems. Isolation simplymeans using one of a number of electronictechniques to interrupt the connectionsbetween two grounds while passing thedesired signal with little or no loss of accura-cy. Without a path for ground currents toflow, these currents cannot induce signalerrors. Isolation also solves the other prob-lem encountered with ground loops — volt-age differences which cannot be rejected bythe signal conditioner. Isolation has becomevery cost effective for solving many signalconditioning problems and is an integral partof each SCM5B module.
Anti-Aliasing FiltersEarlier, the phenomenon known as aliasingwas examined. The aliased signals resultedfrom higher-frequency signals in the input. Afront-end low pass filter would prevent high-er frequencies from being processed and pre-vent aliases from appearing in the outputspectrum. The sharper the cutoff of the low-pass filter, the smaller the amount of aliasing.Practically, there will always be some alias-ing. It just needs to be reduced to insignifi-cant levels. The six-pole low-pass filters usedin the SCM5B series exhibit superior per-formance in rejecting higher-frequency sig-nals and minimizing aliasing.
The sampling frequency of a data conversionsystem also has a profound effect on aliasing.It should be significantly higher than that ofthe desired input signals — usually 3-10times the spectrum of the desired signals.How much higher is strongly influenced bythe characteristics of any input filtering.
Note that digital filtering in the data acquisi-tion system will not reduce or eliminate thealiasing. The aliased signal is converted intothe spectrum of the desired signal and thedigital filter cannot distinguish between thereal and aliased signals. Once aliasingoccurs, the original input signal has been lostand cannot be recovered by any means.
DESIGN EXAMPLESThe following design examples illustrate oneor more common signal conditioning orinstallation problems and show howimprovements can be made. The examplesshown are real installations. The remedies
were achieved using readily available SCM5B modules and common techniques.SERVO CONTROL The manufacturer of an automated piece ofmachinery chose 0-10VDC signals to sendcontrol and feedback signals between thecontrol center and the motor controller. The
user had installed these components in a fair-ly normal environment on the factory floorand took reasonable care in running theinstrumentation wiring between the equip-ment — a distance of about 250 feet.However, these components were part of acritical work piece positioning system on amanufacturing line. Noise pickup from adja-cent machinery and an undesired groundloop caused erratic positioning of the workpieces and considerable scrap and reworkexpense.
Figure 31a shows the original configurationalong with the discovered sources of interfer-ence. Figure 31b shows the re-configuredsignal wiring and signal conditioning addedto the two signal paths.
The SCM5B39 module provides 0-10V to 4-20mA signal conversion and ground loopisolation. The 4-20mA signal is less suscep-tible to noise pickup and the isolation elimi-nates common-mode noise and ground-loopinterference. Use of twisted, shielded cablingfurther rejected noise pickup.
The SCM5B32 at the controller end of theloop restores the 0-10VDC signal and com-pletes the control input path. It also isolatesthe input and rejects both common-mode andground noise pickup. The control looprequires a feedback signal, so identical signalconditioning equipment was installed in thefeedback path.
The SCM5B32 and SCM5B39 modules pro-vide an additional benefit. Their bandwidthwas modified to closely match the dynamicrequirements of the control loop. The addi-tional rejection of higher-frequency noisegreatly reduced the erratic positioning prob-lems while still providing sufficiently fastresponse time to provide accurate feedbackcontrol.
The changes in control and feedback signalsneeded to create the incremental movementsrequired at the work piece are extremelysmall. Some isolation amplifiers can intro-duce extra noise into the signal path.Isolation amplifiers operate by passing a sig-nal through some non-resistive media to theoutput. This media may be optical, magneticor capacitive. Each of these methods makesuse of a carrier signal which is modulated bythe input signal and transmitted across theisolation media. The signal is then demodu-lated to produce the output signal. Thisprocess can introduce substantial noise andcarrier ripple unless carefully controlled bythe design of the isolation scheme.
Dataforth’s SCM5B modules are specificallydesigned to provide extremely low levels ofoutput noise and ripple, providing the highdegree of signal integrity required for thisapplication. Additionally, by matching thefull-scale signal levels of the original equip-ment, the SCM5B32 and SCM5B39 modulespreserve the signal-to-noise ratio (SNR) ofthe system.
ALUMINUM SMELTINGThe electrolytic production of aluminumtakes place in reduction cells or pots. Theremay be 1,000 or more cells in a modernplant. The reduction cells are made of steelwith a carbon lining. Generally, two rows ofcarbon anodes are suspended overhead frombus bars, which carry the electric power.
The cells are filled with molten cryolitemaintained at a temperature of about 980°Cby passing direct current through the cells.Alumina is continuously added to the cells.The electric current passing through the cellsdrives off the oxygen atoms in the alumina,leaving the aluminum atoms to collect asmolten aluminum at the bottoms of the cells.Periodically the molten aluminum is drainedoff. The oxygen atoms adhere to the carbonanodes and gradually erode them. Spent
anodes are removed and replaced on a regu-lar schedule. Controlling the electrolysis atthe cells and determining when to replace theelectrodes are two major problems facing theplant operators.
Normally, the cells operate at about four tosix volts drop across the electrodes.Electrical current through the cells rangesfrom 50,000 to more than 150,000 amperes.The condition of the electrodes can beinferred from the nature of the voltage dropacross each cell. The process is controlled byadjusting the immersion depth of the elec-trodes in response to voltage changes acrosseach cell. High resolution (1mV) is requiredto maintain good control. However, thechemical activity and deterioration of theelectrodes of the electrodes produces electri-cal noise. Additionally, the high supply volt-ages present require that the measuring appa-ratus withstand up to 1500V common-modevoltages. These measurement problems sug-gest an isolation amplifier with high com-mon-mode rejection, low output noise, goodresolution and linearity. The presence of avery sharp low-pass filter response is neededto reduce the undesired electrode noise.Because of the need to resolve 1mV in thepresence of common-mode voltages as high
Figure 31. Servo Control Application
20
as 1500V, the same considerations of lowoutput noise and ripple apply to this meas-urement problem as existed in the previousservo control application. The DataforthSCM5B31 provids the right solution for thisdifficult but crucial measurement problem(Figure 32).
To preserve the benefits obtained with theSCM5B31, care in routing the signal wiring,placement of the signal conditioning equip-ment and use of twisted, shielded signalwires was of utmost importance. The mag-netic and electrical fields surrounding thepower busses are quite high and can producesubstantial normal-mode error voltages if notproperly accounted for. As in most instances,the installation practices are as important asusing signal-conditioning equipment with theright characteristics.
GROUNDED THERMOCOUPLESSome thermocouples used in process-controlare made in a unique way. Figure 33a showsthis construction. The dissimilar metals areassembled in a concentric arrangement, witha ceramic insulator separating them. They arewelded together at the end to form a junction.This provides an extremely robust thermo-couple, but the assembly is designed forinsertion into a thermo-well — a metal fittingon the vessel that allows removal of a TC orRTD without spilling the vessel’s contents. Athermo-well is always grounded and causesthe grounding of this type of thermocouple(Figure 33b).
As explained earlier, shielding and ground-ing are two of the most difficult problems todeal with. The TC is a low-level signaldevice and shielding is required. The signalconditioning system has its own ground andthe TC is grounded as well. Ground looppotentials and currents are a major problemfor this thermocouple type.
There is one completely satisfactory way tosolve these problems — insert an isolationstage between the signal and the rest of themeasurement system. Note that the isolatedsignal common can be connected to the sig-nal shield because it is not connected toanother earthed ground point, so no groundloop is created. Signal amplification and lin-earization should take place before the signalis input to the measurement system in orderto preserve signal integrity and relax over-head requirements. The TC’s are usuallyattached to large thermal masses such as stor-age tanks and therefore do not change outputrapidly. A low-bandwidth signal path further
reduces noise introduced by the TCs connec-tion to the plant process. An SCM5B47 lin-earized TC input module selected for the
appropriate TC type and temperature range isan optimum choice of signal conditioning forthis application (Figure 33c).
Figure 32. Aluminum Smelting Application
Figure 33. Grounded Thermocouple Application
21
PRODUCT SELECTION GUIDE
Analog-to-Analog Signal Conditioning ModulesDataforth Corporation, established in 1984, is a worldwide leader in the design and manufacture of signal conditioning and datacommunications products for industrial and institutional markets.
Dataforth products protect valuable measurement and control signals and equipment from the dangerous and degrading effectsof noise, transient power surges, internal ground loops, and other hazards present in industrial environments. We offer severalfamilies of analog and digital I/O products and accessories, plus full customer service and application support. We invite you tocontact us for complete technical information about our products and how they might help meet your current or future designrequirements.
DSCA High Performance DIN-Mount, Isolated Signal Conditioners
SCM7B Accessories 1-, 2-, 4-, 8-, 16-Channel Standard and DIN Backpanels; 19-inch Racks; Cables; Resistors
23
DSCL Loop and External Powered Isolators
Model Description
DSCL20 Loop Powered 4-20mA Isolator, Component Mount
DSCL21 Loop Powered 4-20mA Isolator, DIN Mount
DSCL22 Loop Powered 4-20mA Isolator, DIN or Panel Mount, 1, 2, or 3 Channel
DSCL23 Loop Powered 4-20mA Isolator, DIN or Panel Mount, Feeds DC Power to Transmitter
DSCL24 Jumper Configurable Isolator, DIN or Panel Mount, Universal DC/AC Power
DSCL25 Powered 4-20mA Isolator, DIN or Panel Mount, Input Signal Fault Detect
DSCP and SCTP Software Programmable Transmitters
Model Description
DSCP20 Loop Powered Temperature Transmitter, DIN Mount
DSCP80 Universal AC/DC Powered Temperature Transmitter, DIN Mount
SCTP20 Loop Powered Temperature Transmitter, Head Mount
• 24-60V or 85-230V AC/DC Powered Models
• Jumper and Software Programmable Models
• Single and Multi-Channel / Splitter Models
• Input Signal Fault Detection Available
• Isolation Protection to 4000Vrms
• DIN, Component, Panel or Instrument Head-Mount Packages
• Signal-Powered Passive Loop Isolators
Low Cost Industrial Loop Isolators and Programmable TransmittersThe DSCL, DSCP, and SCTP products are a complete family of loop or universal AC/DC powered models of isolators and trans-mitters in component, DIN and head-mount packages offering fixed-gain or hardware and software programmability. Dependingon the model, they accept a wide range of voltage, current, thermocouple, or RTD signals then filter, isolate, amplify, linearize, andconvert these input signals to high level analog outputs suitable for use in data acquisition, test and measurement, and controlsystems.
Key Features and Specifications
PRODUCT SELECTION GUIDE
PRODUCT SELECTION GUIDE
24
DSCT/SCT1P Two-Wire Transmitters
Model Description
DSCT/SCT1P30/31 Analog Voltage Input Transmitters
DSCT/SCT1P32 Current Input Transmitters
DSCT/SCT1P34 Linearized 2- or 3-Wire RTD Input Transmitters (4-Wire, other RTD on request)
Distributed Data Acquisition and Control SolutionsHigh quality SCM9B modules provide cost-effective protection and conditioning for a wide range of valuable industrial controlsignals and systems. Our extensive line includes fixed and programmable sensor-to-computer and computer-to-analog outputinterface modules, RS-232/RS-485 converters, RS-485 repeaters, and associated backplanes, accessories, and applicationssoftware. SCM9B modules are excellent solutions for distributed data acquisition applications such as process monitoring andcontrol, remote data logging, product testing, and motion and motor speed control.
Key Features and Specifications
Industrial and Process Two-wire TransmittersOur DSCT DIN rail mount and SCT1P head mount isolated, instrument class 2-wire transmitters are low cost, high performancesolutions for conditioning and sending analog signals from field sensors to monitoring and control equipment located thousandsof feet away in central control areas. They operate on power from a 2-wire output signal loop, modulating the supply current torepresent the input signal within a 4 to 20 milliamp range. These transmitters accept a wide range of inputs, and the 4-20mA sig-nal is virtually immune to noise pick-up and interference.
Key Features and Specifications • 95dB NMR (60Hz)• ± 0.05% accuracy • CE compliant
• Instrument head, panel, and DIN mounting options
• CSA Certification and FM Approval pending
• Only low-cost, twisted-pair wiring isneeded, operates with 10.8V to 100V loop supply voltages
• 1500Vrms transformer isolation • Input/output protected to 240VAC
continuous
SCM9B Isolated, Intelligent Signal Conditioning Products
RML100/101 14-slot Card Cage/Backpanel w/110/220VAC, ±12VDC Power Supply
RML035 Isolated RS-232 Current Loop Card, 2 Channels, 19.2K Baud at 0.5mi
RML070 Isolated RS-232 Current Loop Card, 2 Channels, 57.6K Baud at 0.5mi
RML485 Isolated RS-485/422 Card, 2 Channels, 57.6K Baud at 0.5mi
Industrial Communication Products
• Protects equipment from damage dueto power surges, transients, lightning
• 1500Vrms isolation with opto couplers and power DC-to-DC converter (6000VDC, 1 min)
• Extends RS-232 communications distances without expensive low-capacitance cabling
• Connects RS-232 devices to RS-422and RS-485 devices
• Data rates to 115K Baud • Distances to 12 miles (20Km)• CE Compliant
Industrial LANs and data communications systems stretch over long distances, inside and outside, with signals exposed to elec-trical transients, noise, ground loops, power surges, and lightning. Commercial communications equipment often is not designedfor use in these environments, which can lead to unreliable signal quality, damage to expensive CRTs, computers and other on-line equipment, and production downtime. Our heavy duty modems, rack-mount modem systems, and PC-based serial I/O cards"harden" and protect these systems, and can extend communications for many miles without expensive cabling.
Key Features and Specifications
PRODUCT SELECTION GUIDE
Call 1-800-444-7644 for Customer and Applications AssistanceDataforth Corporation Tel 520/741-14043331 E. Hemisphere Loop Fax 520/741-0762Tucson, AZ 85706 email: [email protected]
