As noted in Chapter 1, the low-frequency or 'conventional' circuit theory is ordinarily the best model when the wavelength is large in relation to the dimensions of the system components. With increasing frequency and decreasing wavelength, however, the ability of the circuit model to explain the observed behaviour is no longer adequate for many practical problems. In principle at least, and assuming that the phenomena of interest fall within the scope of electromagnetic theory, these limitations can be overcome by recourse to a different model, namely the Maxwell or field equations. Unfortunately, however, this model is also an extremely complicated one. Here the low frequency concepts of 'voltage' and 'current' are, to a large extent, replaced by the electric and magnetic fields associated therewith and these are then described throughout all space! Although there are applications where this detail is useful, if not essential, there are many others where it is not and the problem is how to modify the model so it represents only the details of interest. The conventional low frequency circuit theory is one such simplification. Moreover, it can be shown that this circuit theory is a consequence of Maxwell's equations under the given conditions.

The chapter is about the properties of the media which is included in the multi-port structure, it should be noted that these are well satisfied for a large range of materials. Thus reciprocity ordinarily holds. On the other hand, these criteria are not satisfied by magnetised ferrites, for example, and an important family of nonreciprocal microwave devices exist which exploit these phenomena. These include circulators and isolators.

The generator or source model which is implicit in Chapter 3 eqn. 3.10 may in reality be a poor representation of a given source of microwave energy. On the other hand, with appropriate techniques, it is possible, in principle at least, to achieve any desired degree of conformity to the postulated model. In general there are two problems: frequency instability and amplitude instability. In its simplest form, frequency instability may be corrected by sampling the signal of interest, comparing its frequency against some reference and the introduction of appropriate feedback as required to achieve the desired result. This chapter will attempt to develop an understanding of amplitude stabilisation and consider some related topics.

In recognition of its all pervading nature and widespread application, equation 7.3 in Chapter 7, which may be wriiten as b₁ = A₁a₂ + B₁b₂, is given the title 'The fundamental equation of microwave metrology'. It is the primary objective of this chapter to provide its derivation under a fairly general set of initial conditions. This task, in turn, will become a vehicle for developing certain analytical techniques which find repeated application in microwave metrology. In the process, the equation will actually be obtained via several different methods, each of which provides a different perspective.

Even though it has been largely replaced by other methods, no text on microwave metrology would be complete without at least a brief treatment of the 'slotted line'. The basic form of the measuring instrument includes a section of the appropriate transmission line. In any case, the perturbation due to the slot can be minimised by keeping it narrow and parallel to the current paths. The other source of perturbation is due to the probe itself. In general this is minimised by the use of a sensitive detection system and by keeping the depth of penetration, or coupling, to a minimum. From the viewpoint of careful metrology, its sources of error include impedance discontinuity introduced by the slot, inability to maintain uniform probe coupling to the fields due to imperfections in the (nonideal) transport mechanism, and violation of the 'uniformity requirement', and distortion of field patterns due to probe 'loading'. Unfortunately, however, the success of this scheme is contingent on a known value for the 'shunt admittance', or probe coupling, and this is not easily determined.

The technology described in this chapter at one time represented a highly developed art and provided much of the basis for precision microwave metrology. Although failing to yield phase information, the need for this was in some cases circumvented by 'generalised reflectometer techniques' which specifically addressed certain problems of major interest, particularly in the area of power measurement. At best, however, these methods were both frequency sensitive and time consuming. With the increasing interest in broadband systems, and the advent of digital technology, these methods have to a large degree been replaced by others which are more amenable to automation. In the process, a major shift in measurement strategy has also occurred. In particular, the foregoing tuning methods were largely developed in an era where the key to better accuracy was an improved item of hardware. As such, they represent a set of highly developed techniques for making in-situ adjustments of the hardware parameters such that the 'ideal' response is more nearly realised. By contrast, the more recent strategy, as reflected by the 'vector' network analyser, for example, is that of specifically identifying and characterising the hardware imperfections such that they may be eliminated by software corrections. Although much of this earlier technology is still embodied in the 'scalar' network analyser, in other cases the shift in emphasis has been from trying to build a better piece of hardware, to 'smarter' use of that already in existence.

Microwave power meters fall into the categories of 'terminating' and 'feed-through' types. The bolometric technique, as described in the preceding chapter, (ideally) terminates the line in its characteristic impedance. It thus absorbs the energy which it measures and falls into the first category. By contrast, the feed through type may absorb only a fraction of the available energy and pass the remainder to a termination. The first type tends to be the basis for calibrating or assessing the operation of the second type and its applications will be described first.

The application of automation to the field of microwave measurements is perhaps best illustrated by the (vector) automated or automatic network analyser (ANA). The large reduction in measurement time and operator effort which it provides are well known, but this feature may tend to obscure a more fundamental aspect of its impact. Although techniques for measuring complex microwave impedance, scattering coefficients etc. have been known since the inception of the art, until the advent of the ANA much of microwave metrology was focused on the more easily determined scalar parameters such as power, attenuation and reflection coefficient magnitude. In the absence of phase information, it was common practise to determine 'worst case' limits for certain phase interactions which were labelled 'mismatch errors'.

Because the microwave art is characterised by the use of a variety of waveguides or transmission lines, the subject of adapters and their evaluation is an important one to the microwave metrologist. A typical problem is that of using a power or noise standard in a waveguide (for example) to calibrate an 'unknown' in coaxial line. Today, as will be described below, an explicit and complete adapter evaluation is possible using the automated vector network analysers. In addition, however, the prior art also includes techniques which provide an approximate elimination of the adapter parameters from the problem of interest. It is the purpose of this chapter to provide, with the help of specific examples, a brief survey of the existing methods.

The existing techniques for evaluating the noise contributions from microwave amplifiers, mixers etc. call for the use of noise generators or sources of known parameters or characteristics. This chapter will focus on noise sources, standards and their intercomparison.