concerned, but may be critical to the device producer for the following reason: For the better grade diodes, the yield falls off precipitously with reduced limits on overall average noise figure, so that most diodes of a given grade (suffix) are close to the acceptance limit. Therefore, small change in the limit can change the yield of that grade diode far out of proportion to the ratio of the change to the 0.5-dB width of the acceptance "slot." A tightening of the limit by 0.1 or 0.2 dB may very well reduce to zero the yield of the best grade subtype. The economic impact of this yield reduction could be considerable, since sales price varies considerably with diode grade. (As an extreme example, a 35-GHz mixer diode type was advertised at $150, $200, and $340 apiece for noise figure limits of, respectively, 5.5 dB, 5 dB, and 4.5 dB; the 0.5-dB improvement offered by the best grade thus commanded a premium of $140.) The measurement uncertainties that have been determined for the NBS apparatus and techniques as described are sufficiently large that even if all parties used the same design of apparatus and the same techniques, the interlaboratory variance is likely to give rise to the following difficulty: With diodes ranked in grades with incremental improvements in noise figure limit between grades of 0.5 dB, consider the outputs from two suppliers. Each supplier designates the diode grade as a result of measurement. Because the difference between supplier measurements can be a substantial fraction of the 0.5-dB increment (in fact, at the limit of twice the uncertainty, equal to the increment), two diodes of identical performance are classed in two different grades according to their suppliers. Put another way, if the incremental difference between diode grades is to remain at 0.5 dB and is to be meaningful, independent measurements will not suffice unless or until the basic measurement state of the art is improved to reduce the uncertainty. (5) The uncertainties established for the attenuation and power calibrations can probably be reduced in the near future. Further improvements in power stability may also be possible. It seems unlikely, however, that the total measurement uncertainty can be reduced much below +0.1 dB, which may still be too large for independent producer measure ments. Note: The r-f noise measurements which were to have been investigated later in this program would perhaps have resulted in a somewhat lower measurement uncertainty. This uncertainty, however, would have been at least as large as the uncertainty in calibration of the r-f noise source, which is presently in the range of +0.06 to +0.08 dB over X-band and Kuband (and greater in other bands). It is unlikely that a total uncertainty much below +0.1 dB can be achieved, even if special care had been used to reduce the calibration uncertainty somewhat below this range. ACKNOWLEDGMENTS Essential to this work has been the skill of Leonard M. Smith, who modified and rebuilt the modulation attenuator, constructed the waveguide clamping system, and performed all other machine work. of the many people who helped to create and further this program, special recognition is due Judson C. French and Myron G. Domsitz of NBS; Dr. Leon Podolsky, former chairman of the NAS/NAE Advisory Panel, Ronald Wade of the Naval Electronics Systems Command, and most particularly, the late Eugene J. Feldman, who, before his untimely death, was chairman of the JEDEC Committee on UHF and Microwave Diodes. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. IRE Subcommittee 7.9 on Noise, Description of the Noise Performance Kenney, J. M., The Simultaneous Measurement of Gain and Noise Using Only Noise Generators, IEEE Trans. on Microwave Theory and Techniques MTT-16, No. 9, 603-607 (September 1968). Kenney, J. M., Microwave Device Measurements, Methods of Measurement Torrey, H. C., and Whitmer, C. A., Crystal Rectifiers, M.I.T. Rad. Taub, J. J., Crystal Characteristics, Handbook of Microwave Mea- Southworth, G. C., Principles and Applications of Waveguide Transmission, pp. 374-376 (D. Van Nostrand Co., Inc., New York, 1950). Hand, B. P., A Precision Wave Guide Attenuator Which Obeys a Mathematical Law, Hewlett-Packard J. 6, No. 5, feature article (January 1955). Larson, W., Gearing Errors as Related to Alignment Techniques of the Rotary-Vane Attenuator, IEEE Trans. on Instrumentation and Measurement IM-14, No. 3, 117-123 (September 1965). Larson, W., Table of Attenuation as a Function of Vane Angle for Rotary-Vane Attenuators (A = -40 log10 cose), NBS Technical Note 299 (January 1965). 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Larson, W., Table of Attenuation Error as a Function of Vane-Angle Error for Rotary-Vane Attenuators, NBS Technical Note 177 (May 1963). Military Specification, Semiconductor Devices, General Specification for, MIL-S-19500 E, par. 40.2 (1 April 1968). Military Specification, Semiconductor Devices, Diode, Silicon, Mixer Military Specification, Semiconductor Device, Diode, Silicon, Mixer, Military Standard Test Methods for Semiconductor Devices, as revised by Notice 1, MIL-STD-750B (13 May 1970). Price, W. A., A Treatise on the Measurement of Electrical Resistance, pp. 153-155 and 192-196 (Clarendon Press, Oxford, 1894). Roller, F. W., Electric and Magnetic Measurements and Measuring In- Northrup, E. F., Methods of Measuring Electrical Resistance, pp. 218-219 (McGraw-Hill Book Inc., New York, 1912). Gray, A., Absolute Measurements in Electricity and Magnetism, pp. 415-417 (Macmillan Publishing Co., Inc., London, 1888). Gerard, E., Mesures Electriques, Second Edition, p. 415 (Gauthier- Thomson, W., Modification of Wheatstone's Bridge To Find the Kenney, J. M., Semiconductor Measurement Technology: Permanent Danage Effects of Nuclear Radiation on the X-Band Performance of Silicon Schottky-Barrier Microwave Mixer Diodes, NBS Special Publication 400-7 (April 1976). Quarterly progress reports on this work appeared in the following NBS Technical Notes (Methods of Measurement for Semiconductor Materials, Process Control and Devices) in the section entitled Microwave Measurement Methods: Quarterly progress reports on this work also appeared in the following NBS Special Publications under the title, Semiconductor Measurement Technology: Microwave Diodes: APPENDIX A DERIVATION OF EQUATIONS FOR THE MEASUREMENT OF CONVERSION LOSS USING PERIODIC OR INCREMENTAL MODULATION OF THE LOCAL OSCILLATOR The wave travelling toward the load (incident wave) in a uniform transmission line of characteristic resistance Ro (however arbitrarily defined), and carrying a constant (unmodulated) power, Po at a single frequency, wo, can be represented by an equivalent instantaneous emf e = E E√2 cos(wt), r where Er is the Thevenin's equivalent rms open-circuit emf obtained from the available power expression (A1) (A2) so that e = 2 √2P R cos(wt) • Linear amplitude modulation at a single modulating frequency, wm, impressed upon this wave results in an instantaneous emf e = E √2 [1 + mcos (w_t)] cos(wt) r where m is the modulation factor. m Representing the product of cosines by a sum-and-difference identity: (A3) (A4) (A5) This familiar representation shows the two sidebands resulting from the modulation, each of which carries a power m2 Er2/16Ro, by analogy with eq (A2). The carrier (local-oscillator) power is unchanged by the modulation (for m ≤ 1). In the mixer, the intermediate-frequency output voltages resulting from these sidebands add linearly when m << 1, so that they are equivalent to a single-frequency r-f signal of, not twice, but four times the power in each side band: Relative to the carrier power, Po, this equivalent single-frequency r-f signal power, Pg, is, from eqs (A2) and (A6): |