MICROWAVE DEVICE MEASUREMENTS

To add greater rigidity and thermal mass to the harmonic pad, it was remounted (in the course of the waveguide rearrangement) on a ten-inch long aluminum block which supports the pad for its full length. A top plate of the same size is used for clamping.

(L. M. Smith)

Responding to proposed changes in JEDEC Publication No. 77, a letter was sent to the Electronic Industries Association urging that a distinction be made between (spot) noise figure and average noise figure, as is done in IEEE standards.

The feasibility of using "noise measure" as a mixer parameter was briefly explored. Because of the behavior of this parameter for values of output noise ratio close to unity, no noise measure limits could be found that even roughly corresponded to typical limits on overall average noise figure. (J. M. Kenney)

Plans: Work will continue on the random uncertainty determination and on the summary of progress to date. If its stability appears satisfactory, the modulation attenuator will be sent to the NBS Electromagnetics Division for calibration. A limited revision of IEC document 47 (Central Office) 376, an international standard for microwave mixer measurements, will be prepared at the request of the U.S. National Committee.

5.3. CARRIER TRANSPORT IN JUNCTION DEVICES

Objective: To improve methods of measurement for charge carrier transport and related properties of junction semiconductor devices.

Progress: The delay-time correction technique was successfully applied to measurements on the Sandia bridge of transistors subjected to gain degradation,

The data on the passive devices obtained by the five round-robin participants who used automatic network analyzers for measurement have been examined for both within and between-laboratory variability.

Sandia Bridge Delay-Time Instrumentation - Previous work (NBS Tech. Note 743, pp. 39-40) has shown that for the transistor represented by the small-signal model in figure 21 the term to be added to the measured delay time to correct for the effects of extraneous signal pickup is predicted to be

where hfe is the small signal, common-emitter current gain, R2 is the characteristic impedance of the bridge (100 ), r is the transistor base resistance, re is the transistor dynamic emitter resistance, and (T

T) is the delay time zero shift measured

S

rs

by the bridge when an emitter-collector short-circuit is replaced by a resistor of

value R The quantity T is the negative of the delay time error which arises because

of extraneous pickup.

*

To check this prediction, measurements of delay time of a 2N2219 transistor were made as a function of emitter current and frequency before and after degradation of the hfe by neutron irradiation. Signal frequencies used were 3 to 15 MHz; collectorbase voltage was 5 V. This frequency range was chosen for the measurements since it is in this range that the delay-time measurements previously made on R-C networks showed the greatest frequency dependence (NBS Tech. Note 733, pp. 41-43). Results for an emitter current of 20 mA, typical of results in the 5 to 35 mA range explored, are listed in table 8. The pre-irradiation delay time, 1 is relatively independent of frequency. After irradiation, a significant frequency dependence of delay time,

m2'

is observed. The quantity ( - Trs) T) was measured for a through resistor, R = 20.

S

S

For the case under discussion r ≈ 1.3 and is negligible compared with both R and

R_.

e

S

For a high-frequency transistor, such as the 2N2219, the base resistance is much less than 100 . Therefore eq (5) can be quite well approximated as

the correction to T is not significant before irradiation when hfe ≈ 200.

ml

c2

However,

after irradiation, when hfe ≈ 10, the correction shown as T in the table significantly affects the result as can be seen by comparing the columns headed t

(τ

m2

+ Tc2) in the table or the curves in figure 22. Note that the frequency variation shown by T alone is almost completely eliminated when t is added. It should also be noted that there are no adjustable constants in the expression for c

m2

C

(D. E. Sawyer)

We are indebted to Capt. P. J. Vail of the Air Force Weapons Laboratory, Albuquerque, New Mexico, for performing the irradiation.

ments of the S parameters of the transistors and passive devices circulated for the round robin. Five laboratories used automatic network analyzers for the measurements; one used a manual system. While analysis of the data is not complete, some trends can be noted from data obtained on the passive devices by the five laboratories which used automatic network analyzers.

An analysis of variance of the data was made to determine the variability to be expected from repeated measurements within the same laboratory as well as the differences to be expected when the same devices are measured in different laboratories. The variances from all laboratories were averaged to obtain the within-laboratory variance, which procedure is valid if all laboratories are equally capable of reproducing their results in repeated measurements [1]. The between-laboratory variance is a measure of the dispersion of the mean of each laboratory's measurements from the grand mean determined from the measurements of all the laboratories. From these two variances, the within- and between-laboratory sample standard deviations were calculated. Since the magnitudes of the S parameters vary with frequency, the coefficient of variation, which is the sample standard deviation expressed as a percentage of the mean, is used to indicate the variability of the magnitude of the S parameters.

The results of the analysis of the measurements of R-C Network 2 are summarized in tables 9 and 10. The within- and between-laboratory variability is given at three frequencies representative of the range of frequencies covered in the measurements. In table 9 the coefficient of variation is used to describe the variability of the

11

magnitude of S. and S 21; in table 10 the sample standard deviation is used to describe the variability of phase measurements. The columns labeled A show the results obtained when all participants used the same transistor fixture to adapt the transistor socket to the coaxial output of the network analyzer, and the columns labeled B show the results when each participant used his own transistor fixture of the same construction.

12

Values for $22 and S. are not shown; they are almost the same as those and S21, respectively, since the networks are symmetrical.

CARRIER TRANSPORT IN JUNCTION DEVICES

In almost every case, the variability is greater when each laboratory used its own transistor fixture than when all used the same transistor fixture. This would be expected unless the calibration procedures completely compensated for the characteristics of the transistor fixtures. In most cases there is less difference in measurements of than in measurements of S. which indicates that the calibration procedures (NBS Tech. Note 773, p. 35) are more effective in compensating for the transmission characteristics of the transistor fixture than in precisely locating the reference planes.

11'

11

The use of a percentage to express the variability of the magnitude of the Sparameter measurements, table 9, helps to emphasize that the variability does increase with frequency independently of any changes in the magnitude of the S parameter itself. This change is also evident in table 10 where it is seen that the variability of the phase measurements increases with frequency even though the phase angles of both S. and S, decrease with frequency. Variability increases with frequency because errors in locating the reference planes represent a larger fraction of the signal wavelength and hence a larger phase error as the frequency increases. The between-laboratory variation in the phase of the S parameters tends to be larger than the within-laboratory variation because additional errors are introduced by the differences in the transistor fixtures and calibration procedures. (G. J. Rogers)

21

Plans: Additional measurements of delay time of transistors before and after gain degradation by radiation will be made. The analysis of the S-parameter round robin data will be completed. The within- and between-laboratory variability determined from measurements on the passive devices will be used to assess the extent to which variability in the transistor measurements can be attributed to variability in the measuring systems.

1.

Dixon, W. J., and Massey, F. J., Introduction to Statistical Analysis, pp. 119-127 (McGraw-Hill Book Co.,. New York, 1951).