ENGINEERING RESEARCH INSTITUTE UNIVERSITY OF MICHIGAN ANN ARBOR A TOROIDAL SAMPLE HOLDER FOR MEASURING VHF PERMEABILITY AND LOSSES Technical Report No. 35 Electronic Defense Group Department of Electrical Engineering By: Paul E. Nace Approved by: I4{) Lt) H. W. Welch, J D. M. Grimes Project 2262 TASK ORDER NO. EDG-6 CONTRACT NO. DA-36-039 sc -63203 SIGNAL CORPS, DEPARTMENT OF THE ARMY DEPARTMENT OF ARMY PROJECT NO. 3-99-04-042 SIGNAL CORPS PROJECT NO. 194B July, 1954

TABLE OF CONTENTS Page LIST OF ILLUSTRATIONS iii ABSTRACT iv 1. INTRODUCTION 1 2. MATHEMATICAL ANALYSIS 1 2.1 Section One 1 2.2 Section Two 5 2.3 Section Three 7 2.4 Wall Losses and Internal Inductance 7 2.5 Accuracy 8 5. DESIGN 9 4. MEASUREMENTS 10 4.1 Impedance Measuring Apparatus 10 4.2 Measurements with the Three Coaxial Inductors 11 4.3 Calibration 11 4.4 The Value of Characteristic Impedance, Zo 18 4.5 Procedure 18 4.6 Z-G Chart 21 4.7 P-Contours 21 5. CONCLUSION 24 ACKNOWLEDGEMENTS 24 ii

LIST OF ILLUSTRATIONS Page Figure 1 Coaxial Inductor No. 1 2 Figure 2 Coaxial Inductor No. 2 3 Figure 3 Assembly, Coaxial Inductor No. 3 12 Figure 4 ItL1 and I/L2 Measured by Different Inductors 13 Figure 5 IL1 and AL2 Measured by Different Inductors 14 Figure 6 Calibration Curves: 12 + 13 vs 2 15 Figure 7 Calibration Curves: 7 Vs 2a) 17 Figure 8 Sections II and III of Coaxial Inductor No. 2: The Transmission Line 19 Figure 9 AL-Curves Using Two Different Values of Zo 20 Figure 10 P-Contours 23 iii

ABSTRACT A toroidal sample holder suitable for measuring the permeability and magnetic losses of ferrite toroids is described. Equations describing its operation are derived and design problems are discussed. The approximations used and the sources of error are evaluated. Calibration curves are presented. A graphical means of determining the permeability and losses of the ferrite from the measured value of impedance is described. The sample holder has been used over a frequency range of 30 Mc/s to 500 Mc/s.

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN A TOROIDAL SAMPLE HOLDER FOR MEASURING VHF PERMEABILITY AND LOSSES 1. INTRODUCTION The coaxial inductor is a toroidal sample holder suitable for use at "very high frequencies" for the measurement of the permeability and the magnetic losses of a ferrite ring independent of associated circuit parameters. Used with suitable impedance measuring apparatus it allows rapid measurements on large numbers of cores. The coaxial inductor is composed of three sections: 1. A oneturn toroidal inductance which takes the form of a short length of coaxial line terminated in a short circuit; 2. A taper section which transforms the inner and outer radii of section 1 to the radii of a standard coaxial transmission line; 3. A coaxial transmission line joining section 2 to the point at which an impedance measurement can be made. Figure 1 shows a coaxial inductor which was used at the University of Michigan. Figure 2 shows an improved design which is currently in use. The three sections of the inductor are considered separately in the mathematical analysis which follows. 2. MAITHE4ATICAL ANALYSIS 2.1 Section One A length of transmission line which is much shorter than a radian in electrical length may be treated as a lumped impedance. A longer length must be treated as a distributed impedance. The impedance ZA of a shorted section of line of length 11 and inner and outer radii of ri and r is:l 1 Ramo, S. and Whinnery, J. R., Fields and Waves in Modern Radio, John Wiley and Sons, New York, 1944, p. 47..... 1

1.5 DIA..093 DIA- TYPE N CONNECTOR 1/4 9/32 21/32 1/4 1/2 DOWEL PI Nl /8 DIA I D I/2 LONG l-' 1.25 DIA - 1.82 DIA 2 1/8 DIA FIG. I COAXIAL INDUCTOR NO I DIMENSION IN INCHES SCALE 2x 2

0.1450D 0.2000D SLIP FIT TO BE SOLDERED 0 0 < O05 3 ID5D -.060 r. 1.~~~~~~~~~000 ID ~~~FITS TYPE N CONNECTOR FIG 2 COAXIAL INDUCTOR NO 2 DIMENSIONS IN INCHES SCALE 4x

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN ZA Zotanh yEl Zoy _, for, <<-, i.e., for I al < 0.1. (1) 4 where2 = a + j = j A/)z = propagation constant and2 Z =-2- o/E loge ~r = characteristic impedance. ri FL and E are the permeability and dielectric constant of the dielectric medium. (,. = /I 1ge ro:jWL ZA-; 2i rg =;~LA (2) Here LA is the inductance of section one. Since iL=ALo =. hy/m for the air dielectric, there are no dielectric losses. That wall losses and internal inductance are negligible will be established later. This short length of shorted line is equivalent to a one-turn toroid and we shall show this latter approach leads to the same equation. But first consider a one-turn toroid with a two-media core, a ferromagnetic ring that partly fills the region plus air that fills the remainder of the region within the single turn. We require the ferromagnetic ring to be positioned concentrically and to have the dimensions rl and r2 for inner and outer radii and If for axial length. It has a permeability of -L = (F1 - jiz2) -o% H = 27r (amps/ m) The flux through the area A, the cross-section of the toroid, is: = jA webers(3) 2 tbid, p. 332 4

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN Taking the integral in two parts - over the entire region with the air permeability,o and over the ferromagnetic region with the contribution to, over and above what an air medium would contribute - gives: o'r2 f - u LoHQ, dr + (L - o)HIf dr ri r, = _ ILo + o [o ogse r + (4) di 2re Z, = jiL, = ji d(~ = i 2_r {-~I Ioge ri + Jf [- 1] rog r If the ferromagnetic ring is removed, i.e., If = Of Eq 5 becomes identical with Eq 2 as predicted above. 2.2 Section Two If the taper section is short enough or if the change in - is ri small from the input to the output of the section, we can treat it as a transmission line section with a constant characteristic impedance. Otherwise one must calculate the impedance transfer function of this taper section. Consider a linear taper of length 12 and with radii: ro = ko(x + xo) and ri = ki (x + xi) r x + x k~ — = k - where k = ri x + Xi Let us require 11 + 12 << X. Then lumped-circuit techniques can be used. We treat the section as a single-turn toroid. 5

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN dO 02 ro LT di Bf dA = f LoHdrdx A 0 ri 12 ~ log (k X + X )dX k x2 + xx zok(xi - xO) [ loge y dy] 2 + Xi 2 7r (Y-k) 2 k X + Xo i where y = k x + Xi and H = 2 Integrating by parts: /lo k -19e Y dy k 02+ Xi 2ir (yX X - k JY(Y- k) k X xi Yk 2 + xo aok (xi- XO oge(y-k) Ioge ] 12 + xi 2i7r ( -xk k k k XO 2r{ I92 loge kX-i + 12 log1e x-o 12 + X- ] + (7) 2 Lo-92 + xi xlog - XO IOge xO + xi loge'2 + xi 2 + X0 If x0 = xi in Eq 7 it becomes Eq 2. The error introduced by treating the taper as a constant impedance transmission line is given by the last three terms in Eq 7. If this error is small enough the taper section can be treated as a constant impedance transmission line. If we do this we introduce an error by neglecting the taper, but counteract the error introduced in calculating Eq 7, namely, using lumped circuit methods. This will become apparent in the next section.

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN 2.3 Section Three The impedance Z3 at the input of a transmission line of length 13 terminated in an impedance Z2 is:3 Z2 cosh yl3 + Zo sinh 3 (8) Zo cosh a'3 + Z2 sinh (.8 Z2 + jZo tan/3.3 Z+ jZ2 tan 83 for a low loss line (9) This equation treats the transmission line as a distributed impedance. If it is sufficiently accurate to treat the taper as a constant impedance transmission line, Z1 can be substituted for Z2 and 12 + 13 for 13 in Eq 9. In this way the error introduced by the lumped circuit approximation is avoided. 2.4 Wall Losses and Internal Inductance These are negligible since the wall impedance is negligibly small compared to the impedances calculated above. The depth of penetration is:4 8 2 ] 1/2 LWLLoaw J meters (10) where aow = wall conductivity. The wall impedance ZW is:5 I I 4ZW ( + ) - R + j(Lw ohms/meter (11) 3 Reference Data for Radio Engineers, 3rd Edition, Federal Telephone and Radio Corporation, 1949,. 311 4 Page 204 of Ramo and iWhinnery, op.zit. 5 Pages 332-333 of Ramo and Nwhinnery, op.cit., ~ ~ ~ ~ ~ ~ 1 M -

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN Using Eqs 2, 10 and 11 and letting tA = go gives: I I Lw RW ro r+ LA IZAI /2oWw 0o' Ioge (ro/ri) Using the dimensions of Fig. 1, a frequency of 50 Mc/s, and the conductivity of silver, 6.2 x 107 mhos/meter. Rw 0.07 /o Clearly, wall resistance and internal inductance are negligible. 2.5 Accuracy The greatest error is in treating the taper section as a constant impedance transmission line. We use the dimensions of Fig. 1 which is admittedly a poor design for this approximation. This will give an extremum in error. k = 2.04 x0 = 0.2 inches xi = 0.17 inches Using Eq 7, a LT = 6.32 ohms at 500 Mc/s and the error is o.66 ohms, a 9.5% error. This is a serious error, but it is negligible if one makes difference impedance measurements as will be demonstrated later. It is interesting to compute the resulting error in Z3. For this calculation the maximum percentage error will occur at the lowest value of Z1, namely Z1 = ZA (Eq 2). At 500 Mc/s, ZA j 4.5 ohms. Z2 = j c LT + ZA = j 10.82 ohms if we use Eq 7 and Z2 = j l.48 ohms if we use only the first term in Eq. 7. Using Eq 9 the two values of Z3 obtained are j 65.0 and J 66.7 respectively, a 2.6% difference. 8

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN This is roughly the per cent error one expects in the impedance measuring apparatus. Therefore, our approximation roughly doubles the possible error. An additional, but small error is introduced by treating section one as a lumped impedance. The percentage error in ZA is:,3i, - tan $,, tan 1, - = 0.2 % at 500 Mc/s and ~ = L. 3. DESIGN One can alter the length 11, 12, and 13 in Fig. 1 to obtain a better design. The dimensions of section one should be adjusted such that the ferrite ring will occupy a large portion of volume and still accomodate the range of ring sizes desired. In order to maintain negligible shunting capacitance between the faces of the toroid, ~3 must not be decreased too much. This requirement is met if the capacitive reactance is very much less than WL1. Here I If I than ct1. Here E' co 77 LRe(E) E0 where E is the dielectric constant of the ferrite ring. A taper much shorter than a quarter wavelength in length is equivalent to an abrupt or discontinuous change in the inner and outer radii. Since the greatest error is introduced in the taper section, one should hold the length of this section to a minimum. However, one must be careful that at no point along the taper section does the ratio ro decrease enough to produce an ri appreciable shunting capacitance. C 2farads/meter for a conIolge (ro/r1) stant impedance line. This capacitive reactance must be less by orders of magnitude than the series inductive reactance of 9

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN 2/ l~og rF henrys/m Finally, it is desirable to maintain 13 as small as possible as this simplifies Eq 9. Also, the smaller 13, the less the impedance Z1 is isolated from the point of impedance measurement. This means that the error in Zl due to an error in measuring Z3 is decreased as 13 is decreased. This is important because as computed eallier a 2.6% error in Z3 corresponded to a 9.5% error in Z2 for a large ~3. Figure 2 shows a better design. The major fault with this design is the small ri in Section One. However, a steeper taper in the inner conductor might have introduced appreciable capacitive shunting. 4. MEASUREMENTS 4.1 Ipedance MeasuringApparatus The impedance measurements may be accomplished by any of several methods:6 1. Standing-wave ratio measurements with a slotted line 2. Byrne Bridge method 3. Admittance Comparator method 4. Directional-coupler methods 5. Probe methods 6. Hybrid Junction methods Methods 1, 2, and 3 are available commercially. We chose the Byrne Bridge method and purchased the Hewlett-Packard VHF Bridge which allows a 6 These methods are discussed in Section 4-10 of Electronic Measurements, Terman, F.E. and Pettit, J.M., McGraw-Hill Book Co., 1952, pp 157-165 10

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN coverage from 50 Mc/s to 500 Mc/s. A H. P. signal generator and a H. P. detector were used in conjunction with the bridge. 4.2 Measurements with the Three Coaxial Inductors Figure 3 shows a third coaxial inductor. Each of the three inductors have different dimensions. Measurements were made on two ferrite rings with each of the inductors. Figures 4 and 5 show the resulting curves of 1l and 2.' Inductors No. 2 and No. 3 give identical readings. Inductor No. 1 in Fig. 1, being of the poorest design, gives values a little different from those obtained with the other inductors. However, as noted earlier for inductor No. 1, a small error in the Z1 measurement corresponds to a higher error in Z3 and, therefore, a high error in ~1 and p2. Obtaining essentially the same values of Cl and ~2 with the three different geometries serves as a partial verification of the foregoing analysis. 4.3 Calibration One must determine 12 + 13 for use in the modified form of Eq 9. To do this, one inserts in section one a brass toroid of dimensions ro, ri and 11. The measurements made are ZB = j Zo tan P(12 + 13). The lowest curve in Fig. 6 shows the value of 12 + 13 versus frequency for inductor No. 2. The curve shows roughly a 1% increase over the frequency range which seems to be typical of measurements with our bridge. All of the points deviate within experimental error from a straight line. Rather than compute by Eq 2 the value of ZA, it is measured by removing the brass shorting ring and making impedance measurements. Solving the modified form of Eq 9 for ZA gives: 11

s'c3G~I ad-1 J LI-.LV-V OL6-Wi 0. 120OD CONNECTOR THD FITS TYPE N CONNECTOR SOLDER CENTER 0.277ID CONDUCTOR 0.443 1.0201D CAP 0.250 SOLDER FIG 3 12

24. 20 16 FIG 4,"I a,u2 MEASURED BY DIFFERENT INDUCTORS COAX IND' I N1 COAX IND'2, u 2 COAX IND'3 CORE A-61-1 Ti CORE A-61-I

100 90 80 70 60 -so \\ 250 30 _ 20 N _I I 50 100 200 300 400 5( FREQUENCY (Mc) FIG 5 F.i a F'2 MEASURED BY DIFFERENT INDUCTORS A COAX IND #2.. v COAX IND #3 - CORE A-127-1 14

J.5260 0 0 z~ * T I.50.465 / 0 t.44 1.34__ _1 Z, =47.30.30 Zo 50a 0IO 0 050.08.06.04 - _ 9 ____.02 -0 50 100 200 300 400 500 FREQUENCY (Mc) FIG 6 CALIBRATION CURVES: 42 +.0 VS 2r' 15

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN Z3 - jZo tan P3(2 + 13) Z3 (12) Z - jZ3 tan 8(12+ 3) (Zo )- 3 ZB) zo The lowest curve in Fig. 7 shows ZAversus frequency where o = (c/ (o 2:. 100 Mc/s. This curve shows a 10% increase over the frequency range. Deviations from a straight line are again within experimental error. The average value of z gives an inductance LA that is about 10p below the theoretical value. The value of LA was 9% below the theoretical value for both of the other inductors. The errors indicated by these calibration curves are not serious because we use the values actually measured for computing l1 and p2. We believe that they are inherent in the bridge. Since repeated measurements give the same curves it is reasonable to use the values actually measured. Then only the change in impedance is used in calculating 1-4 and p2. On inductor No. 2 the calibration measurements were made twice; once with the shorting ring as described above and once with the short in effect placed closer to the point of impedance measurement. The latter method resulted in an average value of 12 + 13 of 3.50 cm instead of 4.03 cm and an average value of Z of 1.39 instead of 0.833 ohms. Ferrite rings gave Tofwo identical Pl and t2 curves versus frequency regardless of which calibration was used. This indicates that the values of calibration data themselves are not important so long as difference measurements are used in calculating p1 and 12. 16

II).87 3- 1 I - - IL - i r I:.86 _ _ _ _ _ _ _ _ _ _.85 =- 50~ __.84.81 z".eo CsoP- I ~ -B I I i i i i _{ Zo- 45,, 50 I 00 200 300 400 500 FREQUENCY (Mc) FIG 7 Za CALIBRATION CURVES: VS FOR DIFFERENT VALUES OF Z0

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN 4.4 The Value of Characteristic Impedance, Zo Figure 8 shows a cross-section of the transmission line of length 13. It is not a simple straight pipe. The ratio ro varies along the length ri of the line, all of which is not visible. Values of characteristic impedance for that part of the line, which is visible have been calculated and an average value of 46 ohms has been obtained. The manufacturer quotes a value of 50 ohms for his bridge and the latter value has been used. However, if one uses slightly lower values the curves of 12 + 13 and Z are affected in a favorable manner. Figures 6 and 7 show curves for which the values of characteristic impedance used were 45 and 47 ohms. The result is to decrease the positive slope of these curves. In Fig. 6 the slope is zero. In Fig. 7 the increase over the frequency range is 7.5% for Zo = 47 ohms and 5% for Zo = 45 ohms. The effect of using Zo = 46 ohms on the v-curves is negligible as Fig. 9 shows. 4.5 Procedure Once the calibration is made, a ferrite ring is placed in the inductor and Z3 is measured. Z1 is then calculated using Eq 12. Then Pl and 12 are calculated using the following equations which are derived from Eq 5: k [I(Z,) IZAI] if W/ o (13) k Re(Z,) AL2 k (14).f GWo where k = a constant since for our cores r2 is a constant. F2 r1 8.

1.125 1 1.000,.310 ~.102.312o / //.833aD.34000.6351D 7000OD POLYSTYRENE SPACER.4180Dvg 850D CAN'T SEE BEHIND IT FIG 8 SECTIONS'18' IIm OF COAXIAL INDUCTOR #2: THE TRANSMISSION LINE (DIMENSIONS IN CENTIMETERS)

Ok 1-19-V ON 3103'Z# CNI'X VOO,/d....... =97 -o~Z v ~Z JO S3nlVA LN3H3-1jaI OM1 9NISn S3AhIn3 71 6 91J (DIN) AON3nf03_J OOt 002 00 z 001 OS' - -. I I~ I: t.. I II.. 0 01 01 OL \ 06 OL

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN 4.6 Z-@ Chart The use of Eq 12 may be avoided by use of the Z-@ chart.7 Then the calibration procedure must include the computation of 2 + 13. We computed and drew up a Z-Q chart of 12" radius. It was found satisfactory at the higher frequencies and over the entire frequency range for cores of high permeability. For the lower frequencies data on low permeability cores could not be used on the chart because of the difficulty in reading the phase angle accurately on the chart for low values of Z1. 4.7 P-Contours Since the calculation of 1l and p2 is tedious even using the Z-@ chart, P-contours were compiled which greatly speeded the calculations. Computations, heretofore not mentioned, are corrections of the bridge's readings according to the calibration curves of a and b supplied by the manufacturers. These computations are: 3 corrected 3 measured (15) 3 corrected o @3 measured +b (16) These corrections plus Eqs 12, 13 and 14 are all included in calculating the P-contours. Let us define P, = af(hL I I); P2 -= OfL2 Then from Eqs 12, 13 and 14: 7 Pages 152-157 of Ramo and Whinnery, op.cit. 21

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN Z3sinO3 PW1k + ZA + ZB zo and- Pa o and 2 P2 W/k Z3 COS 03 = zB W (18) [P, /os + ZA P /kP o tan 03] In Eqs 17 and 18 all values of Z are absolute values. @3 is the phase angle of Z3. These implicit relations allow the calculation of Z3 and G3 corresponding to a given set of values of P1 and P2. The procedure is: 1. Choose a value for P1 and a value for P2 2. Guess a value for 93 3. Calculate the right-hand sides of Eqs 17 and 18 4. Take the ratio of these two computations and equate to tan 03. Thereby determine 03. 5. Compare this value of @3 with the value guessed earlier. If they are not equal, use the computed value as a more intelligent guess and repeat the computation until the computed value of @3 equals the guessed value. 6. Calculate Z3 from left side of either Eq 17 or 18. The values of Z3 and 03 just calculated refer to the corrected values. So one must use Eqs 15 and 16 to find the measured values. The P-contours thus calculated are plotted in the Z3 measured - @3 measured - plane. P-contours must be calculated at each frequency of interest. Figure 10 shows P-contours for a frequency of 200 Nc/s. Roughly 60 computations are necessary to establish this chart. But it will be used for hundreds of measurements. In addition, 22

af) 00 I) ~ ~ 0 0 0 0 _. _.: oi I*:.5.oL' to ~ ~ _N ot ~~~~~~~~~ii ii ii i i ii i ii i iiiiii a. - o C ~ ~ a.-J a_ - 0.a- 7a: a. a. 46 44 P-0 42 ___= ____ 00_M 40 38 pz~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 36 34 32 30 E CD28 26 24 22 20 13 14 I~ 16 17 I8 19 20 21I22 Zgm (ohms) FIG I0 P CONTOURS

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN computational errors are quickly isolated in computing the chart, whereas corresponding errors made in evaluating the data for each core according to the procedure of paragraph 4.5 would be extremely difficult to isolate. We have calculated P-contours for frequencies of 50, 100, 200 and 500 Mc/s. One merely takes the measured values of Z3 and @3 to the P-contours and reads the values of P1 and P2. Dividing by If gives A1L - 1 and t2 respectively. Thus U1I - 1 and /12 are now easily computed. 5. CONCLUSION The coaxial inductor permits the rapid measurement of tLl and HL2 of a large number of ferrite toroids over a frequency range of 50 to 500 Mc/s. A wider range is permitted if one has the necessary impedance measuring apparatus. The measurements are accurate to perhaps 5% if the inductor is judiciously designed and the impedance measuring apparatus is sufficiently accurate. The technique of measurement is rapid because an initial balance is not necessary and because the ferrite ring does not require a snug fit in the inductor. The only requirement is that it fit concentrically. We accomplish this by sticking the ring in place with vacuum grease which does not affect the measurements. The inductor packed with vacuum grease had the same impedance as the air filled inductor. Finally, the use of P-contours makes the computations very rapid and there is little source of error. ACKNOWLEDGEMENTS The author wishes to acknowledge the counsel of members of the Department of Electrical Engineering of the University of Michigan. In particular the encouragement and counsel of Mr. D. M. Grimes has been most helpful. 224

DISTRIBUTION LIST 1 copy Director, Electronic Research Laboratory Stanford University Stanford, California Attn: Dean Fred Terman 1 copy Commanding Officer Signal Corps Electronic Warfare Center Fort Monmouth, New Jersey 1 copy Chief, Engineering and Technical Division Department of the Army Washington 25, D.C. Attn: SIGJM 1 copy Chief, Plans and Operations Division Office of the Chief Signal Officer Washington 25, D.C. Attns SIGOP-5 1 copy Countermeasures Laboratory Gilfillan Brothers, Inc. 1815 Venice Blvd. Los Angeles 6, California 1 copy Commanding Officer White Sands Signal Corps Agency White Sands Proving Grounds Las Cruces, New Mexico Attn: SIGWS-CM 1 copy Commanding Officer Signal Corps Electronics Research Unit 9560th SU Mountain View, California 1 copy Mr. Peter Haas High Frequency Standard Section Central Radio Propagation Laboratory National Bureau of Standards Washington 25, D.C. - 25 -

UNIVERSITY OF MICHIGAN I111111 1 1II IIIIII rlIIIF11111111111 1111 1 3 9015 03483 1670 75 copies Transpoftation Officer, SCEL Evans Signal Laboratory Building No. 42 Belmar, New Jersey For - Signal Property Officer Inspect at Destination File No. 25052-PH-51-91(1443) 1 copy H. W. Welch, Jr. Engineering Research Institute University of Michigan Ann Arbor, Michigan 1 copy Document Room Willow Run Research Center University of Michigan Willow Run, Michigan l4 copies Electronic Defense Group Project File University of Michigan Ann Arbor, Michigan 1 copy Engineering Research Institute Project File University of Michigan Ann Arbor, Michigan 1 copy Dr. J. K. Galt Bell Telephone Laboratories, Inc. Murray Hill, New Jersey 1 copy Dr. G. T. Rado Naval Research Laboratory Washington 25, D.C. 1 copy Dr. R. M. Bozorth Bell Telephone Laboratories, Inc. Murray Hill, New Jersey