PRINCIPLES OF DESIGN OF SINE-WAVE VOLTAGE GENERATOR-CALIBRATORS

M. S. ROITMAN

(Presented by the scientific seminar of the Department of Radio Engineering)

A generator-calibrator (GC) is a source whose output quantity is known with high accuracy. Such sources are finding ever wider application for calibration purposes, for carrying out critical investigations, and for accelerating and automating testing.

All autonomous sources of alternating voltages, depending on the response of the system to a change in a parameter of the output signal, are divided into:

1 — non-reflex, i.e., insensitive to the signal parameters;
2 — reflex, i.e., responding to a change in a parameter.

The requirements imposed on the metrological characteristics of sine-wave voltage generator-calibrators are so high that, in the overwhelming majority of cases, we are compelled to construct GCs in the form of reflex systems.

This means that a precision indicator of deviation (ID) of the controlled parameter (in the present case, the level of the output voltage \(U_{\text{out}}\)) from some prescribed value is introduced, and this information is used to act on the system (i.e., for automatic regulation of \(U_{\text{out}}\)).

The basic element of the ID, ultimately determining the potential*) accuracy of the GC and its structure, is the measuring transducer (MT).

Measuring transducers are subdivided depending on:

1) the conversion function—into MTs of the amplitude, mean, and rms value of voltage;
2) the degree of constancy of the MT over time—into “stable,” “moderately stable,” and “low-stability”;
3) the magnitude of the input resistance—into “low-consumption” and “high-consumption.”

Amplitude-value transducers, primarily because of their high sensitivity to harmonics and interference, have not found application in precision calibrators. In GCs of the low- and audio-frequency ranges, mean-value MTs based on amplifiers with rectifiers in the negative-feedback circuit are widely used, especially abroad [1, 2, 3]. They are the most advanced mean-value MTs; they have high time stability and a large

*) By potential accuracy we shall understand the limiting value of the attainable accuracy.

input resistance. But they also have a number of disadvantages—sensitivity to even harmonics, the impossibility of direct “self-verification,” and, most importantly, a limited operating frequency range. These disadvantages are almost absent in many rms-value measuring converters (thermoelectric, photoelectric). However, they belong to the category of “medium-stability” (and even “low-stability”) and “high-consumption” measuring converters.

The generalized block diagram of a generator-calibrator with a stable “low-consumption” measuring converter is shown in Fig. 1, where: ZG is the master oscillator, M is an adjustable or matching four-terminal network, PA is a power amplifier, D is a divider, RV is the reference quantity, CC is the comparison circuit, AP is the amplifying-converting device, \(a, b, c, d\) are possible paths of action of the stabilizing system, and I is the indicator of the built-in verification system, connected at the output of the CC or in the AP path. The MC, CC, and RV units form a precision indicator of the deviation of the output voltage from the set value.

Fig. 1.

In a specific implementation, some units may be absent and only one of the paths of action may be used. The main advantage of a generator-calibrator with a stable low-consumption MC is the possibility of stabilizing any voltages without additional power consumption, and of reducing the output resistance on all ranges above the main one.

It is precisely according to the above structural diagram that the “Absolute Voltage Standard” [1] and other generator-calibrators with mean-value MCs [2, 3, etc.] are constructed. We have already discussed the shortcomings of generator-calibrators with a mean-value MC. Unfortunately, generator-calibrators with an rms-value MC cannot be built according to the diagram of Fig. 1, since wideband (in frequency) and high-quality rms-value MCs suitable for automated systems consume considerable power. Low-ohmic MCs cannot be connected at the output of the divider; they can only be connected at its input, i.e., directly at the output of the generator-calibrator power amplifier. For example, to stabilize a level of \(1\ \mathrm{kV}\) with a thermoconverter having a current of \(10\ \mathrm{mA}\) (TVB-3), the power consumed by the MC is \(10\ \mathrm{W}\), which is completely undesirable.

A generator-calibrator with an rms-value MC can be constructed according to the diagram of Fig. 1 only when a matching measuring power amplifier (MMPA) is connected ahead of the MC. Using the capabilities of modern electronics and the principles of invariant systems, it is possible to create high-quality MMPAs for the audio- and ultrasonic-frequency range (for example, [5]). But we still do not have precision MMPAs capable of operating over a wide frequency range. Taking into account the insufficient time stability of the MMPA and of the existing rms-value MCs, it is necessary to use the control-signal method (a particular case of which is the modulation method). Thus, in the case of a low-stability and high-consumption MC connected to the output of the divider, we are forced to construct the generator-calibrator according to the structural diagram shown in Fig. 2,

where:
GOP — reference AC voltage generator;
SUM — matching measuring power amplifier;
P — switch;
UP — amplifier-converter.
The above-listed links and IP form the subsystem of periodic conversion.

According to the block diagram shown in Fig. 2, the GK 745A of the firm Hewlett Packard, 1968, was built.

Creating a source of sinusoidal reference voltage operating over the entire required frequency range is a difficult task, and, in general, an incorrectly posed one, since a vicious circle results: in order to construct a stable source of AC voltages, a stable source of the same voltages is needed. If the requirements on the waveform of the GOP curve and its range are excluded, then the task of creating it is sharply simplified. The most realistic and comparatively easy to implement is the construction of a square-wave voltage generator at a fixed frequency [5].

Such a generator is used as the GOP in [6]. But here we are placed in unequal conditions with foreign developers, since our industry still does not mass-produce precision temperature-compensated reference diodes with a long-term guarantee. This fact was one of the reasons why the staff of the Department of Radio Engineering of the TPI did not adopt in its developments the path of constructing calibrators according to the structure of Fig. 2. But this is not the only reason, and not even the most important one. The potential capabilities of such calibrators are determined by the metrological characteristics of the reference AC voltage generator (GOP), by the stability of the course of the frequency-phase characteristic of the SUM*), by the sensitivity of the IP, and by errors caused by switching of the converter. A substantial shortcoming is that the possibility of direct self-checking is excluded. In fairness it should be said that work on creating reference AC voltage sources (even at a fixed frequency and with a nonsinusoidal waveform) is very timely, since the availability of precision GOPs considerably broadens the possibilities of instrument designers.

Fig. 2.

The indicated shortcomings are absent in calibrators with a “multipurpose” IP, made according to the block diagram of Fig. 3, where OP is a source of reference DC voltage. An indispensable condition for the high-quality operation of such calibrators is the absence of short-term fluctuations in the IP. The time drift of the IP and the required stability of the output voltage determine the interval between two connections to the OPI. It is obvious that in GKs with “medium-stability” IP it is necessary—

*) The problem of creating amplifiers with a bandwidth of the order of 1 MHz and wider, with a stable course of the frequency characteristic, is very important and timely.

it is permissible, for periodic correction, to have in the system an indicator of the built-in verification system (I). In the case of a “low-stability” IP, the permissible intercorrection interval is reduced, making normal operation of the GK difficult (or even impossible). Therefore, when high long-term stability of \(U_{\text{out}}\) is required, it is necessary to use automatic periodic correction. The principal difficulty that arises in this case is the lack of highly reliable relays with low contact resistances \((r \leq 0.01\ \Omega)\) and parasitic reactances.

Fig. 3.

For many practical cases, high voltages must be present at the output of the GK (300 V and more). As was already noted, it is inadvisable to apply such output voltages to the input of a “multi-consumption” IP. Let us consider possible implementations of a GK with a high-voltage output.

In Fig. 4, the part outlined by the dashed line is completely analogous to Fig. 3. Added are a step-up transformer Tr, high-voltage divider \(D_2\), and differential level indicator DU. All voltages below a certain value (for example, 10 V) are taken from \(D_1\), and those above it from \(D_2\).

Fig. 4.

Fig. 5.

Obviously, the voltage at the output of the high-voltage divider can be stabilized by connecting a SUM to a particular tap (see Fig. 2). Finally, a voltage and power amplifier with a strictly known transmission coefficient (UV) can be used—Fig. 5. The latter variant is the most preferable.

First, it is flexible; second, it makes it possible to use only one low-voltage divider (which substantially simplifies the GK) and to have, on any range, a low output resistance (if deep voltage negative feedback is used in the UV).

Its practical implementation comes down to solving an extremely urgent and important problem—the creation of a reference nondistorting high-voltage amplifier.

References

1. H. Galman. Average value transfer standart. Instruments and Control Systems. No. 1, 1963.

2. J. Monrol. Precision A. C. Supplies. Instruments, and Control Systems. No. 12, 1961.

3. P. Richman. “A New Absolute A. C. Voltage Standard.” Session 26, IEEE Record of convention held March, 1963.

4. M. S. Roitman. Precision Repeater. Author’s certificate, 1971.

5. M. S. Roitman, G. P. Trofimov, L. A. Naumov, A. V. Sokolov. Report on R&D “Standardizing generator of rectangular voltage with adjustable parameters.” VNIITRI, 1970, No. 68371510.

6. Calibrator 745 A. Prospectus of Hewlett Packard.