1955 Hemphill: Magnetic Radio Compass Antenna Having Zero Drag 17

apparent that for signal-to-noise comparisonis we would Later, theoretical studies of the sense antenna couplingbe more interested in the product of the effective height problem emphasized the need for some form of capaci-and the square root of the antennia capacitance so the tance transformer to convert from the high-impedanceterm "hi-rootcaps" camc into being. As the airframe low-capacitanice antenna to the higher-capacitance lowmanufacturers gradlually became more open-mindeled on impedance transmission line. Thuis, the "Suscepti-the requirement of the users for a large hi-cap antenniia, former" came into being.they agreed to make the "postage stamp" antenna a Here then in the history of the airline industry's at-"little larger" and extend it a "little further away" from tempt to obtain some so-called "very simple" answersthe airplane skin as a means of increasing the hi-caps. to what was believed to be-when started-a simpleTo reduce the frontal area of the antenna and reduce undertaking of developing a consolidated set of require-drag, they suggested a long slim metallic structure over ments for an airline airborne automatic radio directiona Fiberglas housing. They called this a rowboat or canoe finder. Could it be that we are really making a difficultantennia, but later it was made smaller and they referred problem out of a simple undertaking? Or could it be thatto it as the "banana peel" antenna. When the discussions we and the experts who have worked with us so pains-centered on the need for dual antennas for use with dual takingly on this problem are not really as smart as weADF' the antennia suddenly became a "banania split." thought we were?

A Magnetic Radio Compass AntennaHaving Zero Drag*

A. A. HEMPHILLt

Summary-The use of high speed aircraft has accentuated the it arrives at the back leg of the loop. The Bendixrequirements for low drag antennas and has made the precipitation MN-36 loop showni on the left of Fig. 1 is an 8-inch airstatic problem much more severe. A radically new loop design using core coil having eight turns. The effective height is 0.13ferrite materials makes it possible to exceed the performance of thepresent external loop antenna in a submerged magnetic antenna that cm at 200 kc. The phase difference between the fronthas zero drag. Among its important characteristics are lower weight, and back legs is approximately 1/36 degree at the lowless mechanism, and rejection of precipitation static interference as end of the operating range. These figures are for the loopwell as reduction of the drag to zero.

The antenna consists of a small ferrite-core goniometer with fourradial collector bars of the same material. The design includes a nov-el quadrantal error compensating scheme in which the loop is com-pensated by attaching selected end pieces to the collector bars.

THE Loop ANTENNA

N THE following, a discussion of the magneticantenna follows a brief summary of the factorswhich determine the pickup of a loop antenna.

The voltage induced in an air core loop is dependentupon the number of turns and the height and width ofthe loop in wavelengths. The induced voltage is also afunction of the orientation of the loop with respect tothe signal source. The effectiveness of the loop as a Fig. I-Photograph of various loop designs for the ADF. The LPA-means of extracting energy from a passing wave is low 70A flush-mounted design described here is at the far right.because of the tendency of the effects of the opposite in maximum pickup position. When it is used in thelegs to cancel each other. That a loop has any pickup atall is attributable to the phase delay of the signal be-tween the time it arrives at the front leg and the time near the null. The pickup at 20 from the null is maxi-

mum pickup times sine of 20: an effective height of 0.13* Original manuscript received by the PGANE, September 2, cm times 0.035 or 0.0045 cm. Phase difference between

1955. two legs is about 1/1000 degree. Although effectivet Principal Engineer, Bendix Radio, Division of Bendix Aviationheght of device is only two one-thousandths of an inch,Corporation, Towson, Md. hih fdvc sol w n-huadh fa nh

18 IRE TRANSACTIONS-AERONAUTICAL AND NAVIGATIONAL ELECTRONICS December

resultant drag, though small, is yet too great for aircraft differenices among the results for various shapes areuse and had to be reduced still further. small, and since we only want to get an idea of theseThe MN-60 iron core loop was a step in this direction. characteristics to guide us in development, the error due

The iron core has a permeability of about 50 and an to using a spheroid is of little consequence. Note theeffective permeability for the shape factor of about 3; three form factors at the bottom of the chart and theyet it gives the same pickup as an 8-inch diameter air formula for the demagnetization factor G given for each.core loop having twice the area. The powdered iron core For a sphere (L/D = 1) g= 3 and when 3 is substitutedloop has more pickup for the same area because the high in the formula on the next line above, m will be seen topermeability results in an increase in flux. The induct- have a maximum of three when permeability is in-ance is also increased. This requires that the number of finite. This shows that little gain is gotten by the use ofturns be reduced if the same inductance is to be main- high permeability materials in the form of a sphere. Thetained. However, with the introduction of the magnetic conclusion is similar in the oblate spheroid case. Thematerial, the pickup increases faster than the inductance prolate spheroid formula in the center is only the oneincreases so that there is a net gain in effective height. used in subsequent calculations and charts. Fig. 3 shows

Spheroid of magnetic material placed in uniform magnetic field. T I , 150Let: SOLID PROLATE SPHEROID- E----- D=2a=diameter of spheroid. OF FERRITE IN A UNIFORM v 13OO 130 L

= < = L=2c=length of spheroid. MAGNETIC FIELD 120Ho = magnetic field before spheroid is present.H = resultant magnetic field acting on o

material. __ _ __ 100 2_1K =susceptibility of material. _ -f 90 °p = l+47rk =permeability of material. - - - - ji>4OO - - - - 80°0rB= IAH, =flux density. L-

I=KH1 = intensity of magnetization. _ _ 0_ 70

g =de-magnetization factor defined by Hi =Ho -47rgI._

Z -100 60m=B/Ho =ratio of flux step-up over original field. 50

/ D2 4e= 41- -=factor determined by shape of spheroid. 0 3

Then: I\Since Hi =Ho -4rgI, and 042

JU ( 2 4 6d 10I)I2 14 6 18320 22 24 26 28 30 i32 pI=KH1= 1 H1 RATIO= LENGTH

DIAMETERH1 = Ho- gHo0a - 1) = H-- -- Fig. 3-Curves showing the dependence of the step-up of flux density

1 + g(1- 1) for a solid prolate spheroidal core on its length-to-diameter ratioBut Hi also equals B/u so that and the permeability.

-1+ 110 m (the ratio of flux step-up over the original field)or plotted for various length-to-diameter ratios for a solid

B _ prolate spheroid in a uniform magnetic field. This chart- =-+_____ -=)-nm (ratio of flux increase) shows that if the L/D ratio is small, i.e., 2 or less, theHo 1 +gG -1)flux step-up is almost independent of permeability for

For a sphere (LID = 1.0) g =1 so that m has maximum value of 3 if,u is infinite. permeability ranges from 10 to 500. It shows that in

For prolate spheroids (L/D >1t.0) order to take advantage of high permeability materials,g I

I- { lo°g, 1 + e _ I the L/D ratio must be large. Our problem was to designe2 J\2e 1- e an antenna for high speed aircraft where drag is a very

For oblate spheroids (LID < 1.0) important factor. A thin flat shape lends itself to this ap-1g4 - e- plication when drag and pickup are taken into consider-

e2 e3 sin-e ation. Such an antenna can be mounted external to theFig.'2-Summary of the results of the analysis of the step-up of flux aircraft, in a very shallow blister than is within the

through a loop antenna achieved through the use of a core of mag- boundary layer where the air flow is much slower thannetic material. it is some distance from the skin, or it can be recessed

into the skin and covered with a suitable nonconductingTHEORY material. We find that high permeability materials, such

A study was made to determine what happens when as ferrite, are most efficient when in long, thin bars anda magnetic material is inserted in an electromagnetic this is the form that is most easily mounted on an air-field. A number of charts are presented here to show the craft. The loop used by the Germans in the last war in-practical aspects of this effect. Fig. 2 shows the effect of corporated these features. It was a long, thin loop coila spheroid of magnetic material placed in a uniform using ferrite rods and recessed into the aircraft skin.magnetic field. A spheroid was used, rather than a This type of loop was built by Bendix Radio and othercylinder or bar, for analytical simplicity. The practical manufacturers and is being used to day.

1955 Hemphill: Magnetic Radio Compass Antenna Having Zero Drag 19

We decided that we could take advantage of this form Leakage Inductancefactor in another way; one that would allow the rotor One of the problems that arose in the course of thisto be very small. In this design, we proposed to use development was an undesired circumferential flux pathradial collector bars of high permeability ferrite to col- through the pole pieces. This is shown in Fig. 5. Whenlect the magnetic energy and concentrate it at the center the rotor is lined up with pole pieces (a) and (c), thewhere a ferrite goniometer would extract the signal and leakage inductance due to the flux path in pole piecesdirectional information. The design of the collector bar (b) and (d) greatly increases the inductance of the loopand pole pieces would be such that a highly concen- but contributes nothing to the pickup. If the rotor istrated, but very accurate reproduction of the radiated turned from this position, the additional air gap be-field would be created in the center where the pickup tween the pole pieces will be in this path and will resultloop is located. By the use of this principle, we hoped to in a change in the loop inductance. There are a numberbe able to get the effective height of our present loop in Of xvays to reduce this leakage inductance, but thea device that was very thin and easily mounted on an method used consists in placing a shorted turn radiallyaircraft. around the pole pieces at their center. The shorted turnWith the general concept outlined, the next step was introduces a high reluctance in the center of this path

to determine the physical size and material. We had de- and reduces the coil inductance. After the shorted turntermined that the optimum pickup for a radio compass was added and the air gap between the pole piecesloop antenna for use with our ADF equipment was that matched, the winding configuration was developed toof an 8-inch diameter air core loop. A chart of ferrite give minimum octantal error and to make the induc-loop dimensions that would have the same pickup as a tance constant with rotation. The sensitivity of the loop9-inch diamneter air core loop was then made. This chart was increased materially by this change.is shown in Fig. 4. On it are plotted rod lengths vs roddiameters for various effective permeabilities. Thischart was very valuable in designing the loop because Bit allowed us to evaluate the effect of varying one param- _eter on the other two. For instance, if the air gap isdoubled, the effective permeability is approximately cutin two and this chart shows how much increase in lengthand/or diameter is necessary to get the gain back.

A C

PROLATE SPHEROID LOOP I - 1i1- - - F20 \

DIAMETER AND LENGTH OF LOOPS IWITH PICKUP OF A 9 INCH DIAMETER - -- - - 18

-AIR LOOP.- -II= foo ~~~~11=200

A ei~4OO -XA m 14 E Fig. 5-Sketch showing flux linkages giving rise to undesirable varia--.I 3 SI vv1 1 1 ,, = tion of inductance with rotor orientation.

-- " -12

- - - .I1 Quadrantal-Error Correction- a m m -..§| | [ | 110 This loop could be corrected for quadrantal error by-..F_ I I I I I I ~9 the conventional mechanical means such as cut cams,

3.6 3.4 3.2 3.0 2.8 2.6X2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 .8 .6 4 adjustable tape cams, eccentrics, etc. The use of fourDIAMETER IN INCHES radial collector bars allows a number of corrections, such

Fig. 4-Chart showing the range of core dimensions vs permeability as changing the position of the bars, tailoring theirfor an iron-core loop yielding the same pickup as a 9-inch diam- length or area, or by adding additional pieces of ferrite.eter air loop. These methods have the advantage of being accom-

plished without electrical measurement or disassemblyThe ferrite material is a ceramic and, therefore, very of the loop. A further advantage is in the reduction of

difficult to work. For this reason, the preliminary work the amount of mechanism in the loop, resulting in awas done by using models made of soft iron in a 400 smaller power requirement from the loop drive motorcycle induction field. This use of models allows us to and a consequent extension of its life.make detailed investigations in pole piece shapes and Our first models of this loop had collector bars thatrotor winding investigations on inductance variation were pivoted at the pole pieces and positioned so thatwith rotation as well as octantal error. When these re- their nominal position was at 450 from the line of flight.suits were checked with the loops using ferrite materials, A normal quadrantal error was compensated by forthe correlation was found to be fairly good. moving the bars toward the line of flight. This method

20 IRE TRANSACTIONS-AERONAUTICAL AND NAVIGATIONAL ELECTRONICS December

resulted in two great a loss in sensitivity when the cor- Quadrantal-Error Compensationrection had to be large. Fig. 6 shows why this occurred. Fig. 7 is a graphical representation of the bearing

In this sketch, with the station at a distance as shown, compensation that can be obtained with various lengthsthe magnetic lines of the field would travel down the of compensator bars. To maintain maximum sensitivitypage and be horizontal. Therefore, other things beingpquag,thelanderbhebarinthehorizontal.ht ercthiong bhei of the magnetic antenna, compensator bars 10 inches ingequal thelonger thhb n thehorioal direction th lengthl are attached to the collector arms which lie paral-greater-ythepickup. Whnen the barszotar dislanced as lel to the line of flight. Bearinig compensation is then ob-shownebyfthe tte lines te hit ditacei s tamed by varying the length of the compensator barsless; therefore, the fluxi s i ei a mre attached to the transverse collector arms. This pro-

length in hevetialdiecio (o te ag) lne'~cedure should be followed because the sensitivity of the"captured" earlier and released later. Since a line cuts..; ,, ~~~antenna iS related to the mass of ferromganetic ma-the loop wires as soon as it is "captured" and does notcut the other wire until it is released, the phase dif- .te dwth an compensation is.' . ~~obtainled with no compensator bars attached to theference between the wires on each side of the loop 1S transverse collector arms, and succeedingly less com-increased. This increased phase difference partially com- p i o apensates for the reduction in flux density. The pickup bars are increased in length. In typical installations,for a signal at right angles to the one described is greater whre the foreaft andtroswis ionstoflthopen-where the fore-aft aind crosswise dimeD)SIions of the openl-with the bars moved, because of the longer magneticpath in line with the flux lines, and it is reduced by the i f t awill serve as a guide in selecting compensator bars.slightly less phase difference as explained above. Thisdata indicates that to get niecessary quadrantal compen-sation by changing ratio of fore-and-aft to sidewayspickup, bars have to be moved so far that the effective Since this is a four pole device, it has a number ofarea, and therefore pickup, is materially reduced. possible sources of octanital errors. The coupling to

space is accomplished by means of four collector bars.This is onie possible source of error. The rotor is coupled

_- . . _______tothe collector bars through four pole pieces consti-tuting another; while the chanige in inductance andcoupling in the rotor is yet a third source. The lattercan be compensated for by proper distribution of thewinding of the rotor. This is similar to the octantal error

DIRECTION problem in goniometers. The pole piece shape is heldTO to very close tolerance in manufacturing to reduce

STATION errors. The error caused by the use of four collector barsis fairly small because the length is very small comparedto the wavelengths used. Even a small error from thissource would be of considerable consequence, however,because in a system having the bars parallel and at right

- ----' - -- - ---- - - - -- - - angles to the line of flight, the error occurs withlinl a fewFig. 6-Sketch showing effect of quadrantal error degrees of the four cardinal points. The direction of thecompensation by ulse of pivoted collector bars. error is such that it tends to increase the width of the

null at these four points. The ADF takes a bearing on astation almost directly ahead of the airplane a great

The present loop has the bars both parallel and at deal of the time; therefore this error of 20 or 30 is ofright angles to the line of flight. Quadrantal compen- much more consequence than a similar error occurringsation is accomplished by changing the length of addi- at the 450 points.tional cross pieces that are added at the ends of the col- The error occurs at the four cardinal points becauselector bars. By making these fairly long, good sensitivity there, two of the bars are parallel to the magnetic fieldis obtained because most of the effective area of the and have a high flux density, whereas the other twoopening in the aircraft skin is being utilized. These collector bars have no longitudinal flux. A small changecross pieces are being stocked in a number of lengths so in direction of signal arrival will not produce the re-that two pairs that will give the required compensation quired increase in flux in these two bars because theircan be selected for each installation. The antennas are reluctance is not zero, and an appreciable "in line"interchangeable in all installations, except for these length must exist to make the reluctance of this rela-cross pieces which are labeled for each different airplane tively long bar lower than the reluctance of the air path.and added when the antenna is installed. The result of this is that, in this region, a small deviation

1955 Hemphill: Magnetic Radio Compass Antenna Having Zero Drag 21

12-

14_

ie2- 20 30 40 50 60 0 80

zX 2030 40//9070 0 100 110 120 130 140 O 160 110 I

<s -2t\X//CRETEAI6OEES

-S \ COMPENSATION EFFECTED BY VARIOUS LENTHS OF COMPENSATOR BARS_ COMINAN UNDISTORTED fIELD WITH THE TYPELPA-ETA MAGFNTIC ANTENNA

-10INA TYPICAL INSTALLATION

Fig. 7-Bearing compensation curves for various lengths of compensator bars. The legends refer tothe two compensator bar lengths used, given in inches.

in direction of signal arrival produces a much smaller The LPA-70 is a magnetic antenna. The MN-60 anddeviation in the null position of the pickup loop. other iron core loops could also be called magnetic

In the LPA-70, we have right angle compensating antennas; but in the new loop, the magnetic energy isbars on the ends of the collector bars so that this oc- picked up by the collector bars and conducted as mag-tantal error is practically eliminated; the pickup ele- netic energy to the small pickup coil in the center. Thisment is not a narrow bar but has large end area. feature excludes the electric field because only the mag-

netic component is conducted by the material.The pickup is very small because it is in a concen-

One of the main advantages of this loop, design trated magnetic field. The new antenna has a very smallsecond only to the reduction in drag, is its anti-static effective height to the electric field, which is not con-operation. An insulator in an air stream develops centrated, so its pickup of static is very low. Both thischarges which produce static as they dischange into the loop and previous models have electrostatic shieldsair. This effect is greatly reduced in previous loops by which cannot completely eliminate the static field; butmaking the house slightly conductive but charges do this antenna, the area of which is about two squaredevelop under severe conditions, especially after the inches, will pick up far less energy than the MN-60 withloop has been in service for some time and the conduc- an area of about 12 square inches.tivity has been reduced by erosion. In cases where theloop has been placed under the canopy;, the problem is Inductance Adjustmentvery severe because the canopy surface cannot be made The ioop antenna is very closely coupled to the tunedconductive. The new loop is flush, hence the surface is circuit, hence its inductance must be accurately heldprotected by the boundary layer on the surface of the because of phase shift. In our previous loops we wereaircraft and charges will be small and can be conducted able to adjust the inductance of the pickup coil byoff by graphite impregnation. various means, such as changing the number of turns or

22 IRE TRANSACTIONS-AERONAUTICAL AND NAVIGATIONAL ELECTRONICS December

varying their spacing. In this loop this alteration is not jets on down has been accumulated. The reports of per-possible because there are so few turns and their spacing formance have been excellent. They also bear out ouris fixed by the octantal error requirements. We have belief that sine wave compensation is adequate for alladded an adjustable coil in parallel with the loop pickup installations and, while individual error curves maycoil to set the inductance. This shunt coil has much show a few degrees departure from the sine curve, themore inductance than the pickup coil so that the losses operational results are entirely satisfactory. Those casesare negligible. where the data shows deviation from a true sine wave,

still are in agreement in the vicinity of the line of flightProtective Coating. (fore and aft) where nearly all compass operation occurs.The ferrite material had to have a protective coating (A surprising thing has been that the widest use has been

for a number of reasons. It is very hard and somewhat in relatively slow executive aircraft where the looks orbrittle so that a relatively light blow by a hard object rather lack of looks, since it is flush, is the big sellingmight cause it to chip or fracture. It would eventually point.)absorb some moisture if used in a wet or damp location. The LPA-70 replaces the older radio compass loopsA mechanical support was needed, especially for the without any changes in the system except for some com-pole pieces since the inner surface of the stator assembly ponents which had to be added in the junction box be-must be ground in a very close tolerance after assembly. cause of the different motor characteristics. It has theThe ferrite, being a ceramic, has a very low coefficient following advantages:

of expansion. The coating material had to match this Appearance. The trim appearance of the modern air-low temperature coefficient so that it would not be sub- craft is enhanced by having one or two less pro-ject to excessive strains over the required wide tempera- tuberances.ture range. It must not crack or be overly brittle at No drag. The MN-60 drag is about 5 lbs at 400 MPH-65°C and must still be firm enough to give the re- and 25 lbs at 800 MPH. The MN-36 drag is 15 lbsquired mechanical support at +125°C. It must be im- at 300 MPH and 27 lbs at 400 MPH.pervious to moisture and must not be effected by alco- Less Weight. The LPA-70 weighs 8 lbs as comparedhol, aviation fuel, oil or hydraulic fluids. Epoxy resins to 15 lbs for the MN-60. However, in cases ofoffered the best possibilities, but existing materials did retrofit, this weight advantage may be lost in thenot meet all the requirements. Finding the right resin extra beefing in the ship needed to accommodate theformula and developing the process of encapsulation was new unit.the most difficult problem to be solved in getting the Same sensitivity. In a completely flush installation,unit into production. the sensitivity is the same-when a blister is used,

the sensitivity is appreciably better-than theCONCLUSIONS older unit.

An LPA-70 has been in use in the company plane for Easier maintenance. There is less mechanism and thewell over a year. The loop is recessed in an opening unit is gasket sealed for ease of dissassembly.about 17" square and I" deep; it is therefore completely Fleet interchangeability. Previous loops had to beflush. An MN-60 loop and a means switching from one stocked with compensation for each type of instal-to the other for direct comparison had also been pro- lation. This loop requires only that additional barsvided. The performance of the two loops is equal, be stocked and added at the time of installation.demonstrating the accomplishment of our objective Better anti-static operation. These characteristicsmaking a flush installation with the same sensitivity as were discussed earlier in this paper.an external loop. (Much of the material in this article was in a paperThe LPA-70 has been in production for nearly a year presented at the National Electronics Conference in

and field experience on aircraft of all speed ranges from 1954 and published in Tele Tech, June, 1955.

C~~5