was therefore terminated before the principal objectives were met. report describes the work actually carried out, including the successful demonstration of an improved incremental modulation technique. Not included were any of the contemplated noise measurements, nor, as a consequence, the much-needed intercomparison between the modulation and noisesource methods. (Appendix E constitutes a distillation of unpublished work on mixer noise measurements conducted by the author prior to his joining the NBS.) 2. Design and Analysis of Measurement System The immediate problem was not one of device characterization although characterization of mixer diodes does leave much to be desired - but of refinement and evaluation of the quality of the measurements made of those parameters generally accepted for characterization. Accordingly, the decision was made to construct an experimental measurement system at NBS to study the practical problems faced in measuring the traditional mixer parameters. (Certain alternative parameters for the same basic operational, single-frequency characterization were also to be considered.) The measurement methods to be used were preferably to include all those that had been devised to date, for comparison purposes, but the principal goal was to be the minimization of measurement uncertainty, whether by improved engineering practices in implementing previously used methods, or through the development of new methods. It was recognized that the measurement methods in use are numerous and have large numbers of possible variations in their implementation, so that the possible number of intercomparisons is also large, but it was hoped that the main routes to the best techniques using current technology could be explored. In particular, it was hoped that the discrepancies that have been noted between measurements made using a 30-MHz intermediate frequency and those made using a very low (audio or d-c incremental) intermediate frequency could be resolved. It has been the general experience of those working in this area that these two types of measurements frequently yield substantially different results even though there is no theoretical reason for a measurable difference (no methodical investigation of this discrep ancy has been reported, however). A second intended purpose of the NBS system was to enable absolute measurements of high quality to be made for the development of calibration standards or for evaluating diodes for experiments. With respect to this goal, a study of Schottky-barrier diode radiation hardness was pleted, but it was not possible to begin work on calibration standaris prior to project termination. 2.1 General Nature of Measurement System It may be seen from the equation for overall average noise figure given in 1.2 that one mixer parameter may be calculated if certain others are known; there are only two independent parameters if the intermediatefrequency average noise figure is held constant. The mixer is measured This work is described in reference [22]. as a two-port network, with the image response assumed equal to the signal response, and combined with it in a manner dependent upon the nature of the signal (quadratically, or powerwise, for uncorrelated inputs and linearly, or voltagewise, for correlated inputs). The other spurious sources are combined with the mixer to form a two-port network as shown in figure 1. (The input and output signals shown are noise temperatures - see Appendix D.) This two-port is essentially linear for signal powers much smaller than the local-oscillator power. (The local oscillator has been assumed to be a part of the mixer.) By definition, such a linear two-port has a linear relationship between input and output, as shown in figure 2 (again, using noise temperatures) [2]. The straight line representing the signal transfer characteristics is completely determined (so long as it remains linear) by only two points, or by one point and the slope of the line, the slope being the power gain (or loss) of the network. (Where the gain is less than unity, as in the case of a mixer, it is usually expressed by its reciprocal, which, for mixers, is conversion loss. For the noise temperature representation of figure 2, the gain or loss is for available power, the ratio of the power available from the network to the corresponding power available to the network.) The calculation of one mixer parameter from two others requires that the latter be measured under identical conditions. If the calculated parameter is to be directly measured for comparison, then it too must be measured under those same conditions. For any of these measurements to be duplicated elsewhere using the same device, with different equipment, these conditions must be sufficiently defined and reproducible. These elementary requirements for objective measurements have unfortunately not always been met in measurements on mixer diodes. Frequently, the different parameters are measured in different holders at different stations of a production line. Occasionally, even the operating frequencies are different, and different criteria are used for setting the local-oscillator drive, as when millimeter-wavelength diodes are measured for output noise by inserting them into an X-band test set using an adaptor and adjusting the local-oscillator drive for a given rectified current (the conversion loss being measured at the nominal operating frequency with a given available local-oscillator power). Even when every attempt is made to obtain identical measurement conditions, discrepancies have been noted by various investigators. These are frequently attributed to a difference in the r-f environment of the diode at frequencies above the operating band, at and near harmonics of the local-oscillator frequency. Adequate control of this "harmonic environment" is exceedingly difficult because the frequencies involved are above the normal operating range of the waveguide, thereby permitting multimodal transmission. Multimodal transmission is difficult to analyze precisely, because every waveguide discontinuity tends to transfer power between modes. The behavior of these high frequencies in the waveguide components designed for lower frequency, single-mode propagation has neither been established nor controlled by the component manufacturers. Because of the experimental difficulties involved, the degree of dependency of the mixer parameters on the harmonic (and near-harmonic) terminations has not been established, as noted in 1.2. Until proven other wise, it is well to consider these harmonic terminations as having significant effect on precision measurements. Since the harmonic environment is difficult to reproduce, it is useful to keep this factor constant by performing all of the measurements using the same holder and the same measurement system and with all switches and attenuators that are operated in the course of the measurements isolated from the mixer under test by means of low-pass filters or pads. Other factors that can add uncertainties to the intercomparisons are also eliminated or reduced by this approach: lack of repeatability in the diodeto-holder contacts; differences in match at the signal, image, and localoscillator frequencies; differences in local-oscillator power or frequency; difference in d-c and intermediate-frequency load, etc. The price to be paid for these advantages is system complexity and, as will be seen, certain compromises with the optimum location of the waveguide components for each type of measurement. The same arguments for a common measurement system for the various parameters apply equally well for measurements of a single parameter using different methods; all factors affecting the measurements should be kept constant except for those unavoidably changed by the difference in the methods. 2.2 Choice of Measurement Frequency The choice of an operating frequency for the NBS system resulted from a consideration of diode usage, availability of fixed-tuned standard diode holders, and the availability, performance, physical size, and cost of the other microwave components. Two types of point-contact diodes are produced in the largest volume: the 1N21 S-band diode, measured with a 3060-MHz local oscillator, and the 1N23 X-band diode, measured with a 9375-MHz local oscillator. The distribution of use for Schottky-barrier diodes is not known, but may be expected to follow a similar pattern. (There is a tendency to measure them at the same frequencies as the point-contact types, even using the same holder when they have identical external dimensions or using adaptors when they do not.) X-band waveguide components are much smaller and lighter than S-band waveguide components and also less expensive. Coaxial components could be used for parts of an S-band system, but the best performance still requires waveguide, despite the recent improvements in coaxial connectors. Frequencies higher than X-band would yield an even smaller and lighter system, but at greater cost and with reduced component availability and perfor mance. 2.3 Details of Measurement System The final form of the microwave circuit of the mixer measurement system is shown in the schematic block diagram of figure 3 and in the photos of figures 4 through 17. Overall views of the system are shown in figures 4 and 5. Somewhat greater detail can be seen in figures 6 and 7, each of which shows half the system. These can aid in locating the detail views shown in the remaining photos (figs. 8 through 17), which are sequenced from left to right across the system, seen from the front as shown in figures 4 and 5. This progression also corresponds in a large degree |