UDC 621.317.725.088

DIFFERENTIAL INDICATORS OF VARIABLE STRESSES

N. P. Kalinichenko

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

The merits and shortcomings of single-channel differential indicators (DI) and DI of simultaneous comparison with various measuring transducers are considered.
3 illustrations, 6 references.

Differential indicators (DI), in combination with a generator-calibrator, make it possible not only to determine the transfer coefficients of active and passive three-terminal networks, their instability, and amplitude-frequency distortions, but also to measure variable voltages with high accuracy. This makes it necessary to seek ways of further improving the metrological properties of DI and, above all, of increasing their resolving power. The latter is restrained by the insufficient temporal stability of the DI links and, first of all, of the differential comparator. Two ways of solving the indicated problem are possible.

1. Construction of a DI in the form of a single-channel system using the widely known modulation-switching principle.

2. Search for differential comparators of the rms and quasi-rms values of voltages, which make it possible to obtain the desired sensitivity and stability comparatively easily.

In single-channel DIs of non-simultaneous comparison, the influence of instability of the links is sharply reduced, but new difficulties arise, caused by the presence of an automatic commutator. Even in the case of an ideal automatic commutator, some of these difficulties remain. As an example, let us consider a single-channel DI using photoelectric transducers (Fig. 1). Photoelectric transducers were chosen for the reason that they possess the highest sensitivity. The duration of the half-period of commutation \(T_k\) must be greater than the equivalent time constant of the transducer \(\tau_e\). The more commutation periods fit into the measurement time \(T_i\), the better the result is averaged and, consequently, the greater the noise immunity we have and the greater the sensitivity that can be realized. Thus, the condition must be fulfilled

\[ T_i \gg T_k > \tau_e . \tag{1} \]

But the smaller \(\tau_e\) is, the higher the level of the second harmonic in the low-frequency region at the output of the photoelectric transducer [6]. The ratio of the voltage of the second harmonic to the constant voltage on the photoresistor \(R_{\phi}\) is equal to

\[ K_{2\Gamma} = \frac{K_{\mathrm{fp}}}{2\sqrt{1+(2\omega \tau_l)^2}\cdot \sqrt{1+(2\omega \tau_f)^2}}, \tag{2} \]

where \(K_{\mathrm{fp}}\) is the transfer coefficient of the photoconverter;
\(\tau_l\) is the time constant of the incandescent lamp;
\(\tau_f\) is the time constant of the photoresistor.

For example, at \(\tau_l = 200\) msec, \(\tau_f = 50\) msec, \(K_{\mathrm{fp}} = 10\), and \(f = 20\) Hz, \(K_{2\Gamma} \approx 0.5\%\). The filter must suppress the second harmonic and must not distort the shape of the envelope at a commutation frequency of \(1 \div 2\) Hz. Owing to an insufficient separation of frequencies, even a multisection filter cannot be considered satisfactory if maximum sensitivity is desired. The problem is solved by using an active filter, but in this case fluctuations of the active elements (DC amplifiers) acquire substantial importance.

Fig. 1

The second drawback of single-channel DUs is the need, at low frequencies, for strict synchronization of the commutation period with the frequency of the voltages being compared. Otherwise, impermissibly large overshoots may occur. The influence of transient processes can be sharply weakened by using strobing [6]. However, in this case the requirements on the filter and on the stability of the time intervals of the commutators become more stringent.

The third, fundamentally important drawback of single-channel DUs is that they have a lower input resistance than a DU for simultaneous comparison.

In those cases where the accuracy of voltage comparison is of decisive importance, preference should be given to simultaneous-comparison systems. Differential comparators using photoelectric converters have the highest sensitivity, but the strongly pronounced “breathing” effect of photoresistors in the nonstationary regime complicates operation with the DU and increases the measurement time. As photoresistors are improved, this drawback will diminish; for the time being, however, it compels the search for other elements and possible solutions. Photoelectric converters with photodiodes have somewhat better characteristics in the nonstationary regime [1]. Of interest is a version of a DU design with a measuring converter based on low-power incandescent lamps (NSM \(9 \times 60\)) and germanium photodiodes (FD-3A). The sensitivity of the converter can easily be brought to the level of 5 mV to \(0.005\%\). Figure 2 presents the functional diagram of the DU, where

\(P\) is the input follower;
\(ZG\) is the commutating generator;

U — preamplifier;
K — switch;
SD — synchronous detector;
MA — mismatch amplifier.

One of the basic requirements imposed on indicators is the requirement of high input resistance. Otherwise, shunting of individual elements of the circuit by the input resistances of the indicator leads to additional errors, which may

Fig. 2

substantially distort the measurement result. For this purpose, the input follower is assembled on a 6N6P double triode according to the circuit of a cathode follower with enhanced feedback, which makes it possible to obtain a low output resistance and sharply reduce nonlinear distortions. Changes in the voltages of power supplies, amplifiers, and in the parameters of tubes, transistors, and circuit elements may lead to the gain coefficients of the two amplifier channels differing from one another, which is inadmissible in the case of a DI. Therefore the amplifiers are made identical both in circuit design and in construction. The identity of the channels is achieved by introducing deep negative feedback, which not only stabilizes the gain coefficients of the amplifiers but also reduces nonlinear distortions and the noise level. In addition to frequency errors, errors may also arise that are caused by a change in the absolute level of the voltages being compared. They are associated with the nonlinearity of the characteristics of tubes and semiconductor devices used in indicator circuits. In the DI, nonlinear elements are used (PDs, low-power tubes), whose characteristics are practically identical. These measures make it possible to considerably reduce the zero drift of the DI. The photodiodes are supplied by a pulsed voltage with a frequency of 5 kHz, which makes it possible to amplify the difference of the measured voltages already in alternating current. In practice, the accuracy of indication is determined by the level of fluctuations of the amplitudes of the voltages being compared and by fluctuations of the parameters of the pre-detector part [5]. The presence of higher harmonic components in the photocurrent in the low-frequency region plays a negative role. The use of a passive filter, owing to sufficient frequency separation, makes it possible to eliminate the higher harmonic components. A single-section T-shaped filter gives a suppression of the second harmonic by a factor of 10, and a two-section filter by almost a factor of 100. As the switching frequency is increased, switching transients arise, which may considerably exceed in amplitude the extracted differential signal and lead to saturation of the MA. To reduce switching transients, coupling capacitors of the PD power-supply circuits are used. The amplified differential voltage is rectified by means of a synchronous detector. The SD operates in a switching mode. The control voltage is also rectangular in shape. In this case attention should be paid to the stability of the duty cycle

of this voltage. The SD circuit is implemented using IPPGK integral switches. The zero offset of the SD is affected not only by the residual parameters of the integral switches, but also by the parasitic capacitive coupling between the windings of the switching transformer. Therefore, the winding of the signal and switching transformers was carried out with tight inductive coupling (the wires were twisted into bundles and the winding was done turn by turn until the inner perimeter of the core was completely filled).

The measurement process takes place in three stages. First, the reference voltage \(U_x\) is compared with the reference dc voltage \(U_0\). Then mutual calibration of the channels is performed, for which the second channel is connected to the input of the first channel. The characteristics are equalized by adjustment until the indicator gives a zero reading. The third stage is the direct measurement of \(U_{x1}\) and \(U_{x2}\). The signals are amplified and converted. The rectified voltage is modulated. The voltage of the difference modulated signal is amplified, rectified, and recorded by a dc galvanometer. The absence of modulation \((m = 0)\) indicates exact equality of the amplitudes of the voltages being compared and can be recorded by the null method with high accuracy. As a result of the investigations carried out, a DU was constructed with the following technical characteristics:

dynamic range \(5\ \text{mV}—20\ \text{V}\);

frequency range \(20\ \text{Hz}—1\ \text{MHz}\);

sensitivity \(0.005\%\);

input resistance (at the mid-frequency) \(50\ \text{M}\Omega\);

input capacitance \(5\ \text{pF}\);

power supply—ac mains \(220\ \text{V}\), \(50\ \text{Hz}\).

We conducted studies of the stability and resolving power of converters based on field-effect transistors, UPTs, and semiconductor diodes. The studies showed that converters based on UPTs (1UT221) do not make it possible to obtain sufficiently high sensitivity and have a rather large short-term zero drift \((\pm 0.02\%)\). The time instability is \(0.06\%\) over \(30\ \text{min}\). At the same time, the amplitude characteristic is linear for input-signal values up to \(25—30\ \text{mV}\); moreover, at \(U_k \leq 70\ \text{mV}\) the SD operates with fairly good selectivity for the 2nd and 3rd harmonics of the input signal. The accuracy of a DU with voltage rectification is limited by the presence of diodes, their instability, and the substantial influence of harmonics.

The best stability results were obtained with rms-value converters based on 4B94 diode assemblies. Fast-acting integral diode matrices made it possible to create a quadratic converter based on an rms detector with sliding bias and piecewise-linear approximation of the current-voltage characteristic in a combined circuit (Fig. 3). A four-link approximation of the quadratic current-voltage characteristic of the converter is used. At the nominal level of the input ac voltage, the dc voltages across resistor \(R\) and capacitor \(C\) are equal in magnitude and opposite in sign, owing to which the converter output signal is set equal to zero. If there is a difference between the measured voltages \(U_{x1}\) and \(U_{x2}\), a dc signal will appear at the converter output. The stability of such a squarer without additional thermal-compensation circuits is \(0.01\%\) over \(20\ \text{min}\). A DU has been constructed on the basis of such a converter. The operating principle of this DU is analogous to the DU variant with a measuring converter based on low-power incandescent lamps and germanium photodiodes. The sens-

the converter sensitivity at a level of 10 µV is 0.005%. An undoubted advantage of a converter of this type is its insignificant power consumption in comparison with previous types. The DU is implemented almost entirely on microcircuits. MT8 microcircuits are used as input repeaters. As amplifiers, 1UT401B are used. All this makes the instrument compact and convenient in operation.

Fig. 3

REFERENCES

1. M. S. Roitman, V. R. Tsibulsky. Photoelectric converter with a silicon photodiode of the KFDM type. Izv. TPI, vol. 171, Tomsk, TSU Publishing House, 1969.

2. B. A. Perminov, M. S. Roitman, E. I. Tsimbalist. Alternating-current comparator based on photoelectric converters. Avtometriya, 1965, No. 5.

3. M. S. Roitman. Precision DU for rms voltage values. Izv. vuzov SSSR. Priborostroenie, 1966, No. 3.

4. Yu. A. Skripnik. Modulation differential indicator. Izmeritel’naya tekhnika, 1962, No. 2.

5. G. A. Shtamberger. Measurements in alternating-current circuits. Novosibirsk, Nauka, 1972.

6. Yu. G. Svinolupov, M. S. Roitman et al. Comparator of alternating voltages. This collection.