TEMPORAL STABILITY OF PHOTOVOLTAIC CONVERTERS

Yu. K. SHELKOVNIKOV, V. R. TSIBULSKII, A. I. LYSOV

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

Recently, sources with automatic regulation of the output voltage (generator-calibrators—GK) have been successfully used as calibration equipment for checking high-accuracy digital voltmeters. It is known that the principal unit of a GK, determining its metrological characteristics, is the converter, which compares an alternating voltage with a constant reference voltage. The instability of the converter, as well as of the reference source, is wholly and completely included in the instability of the GK output voltage. It is obvious that, for a generator with high long-term stability of the output voltage, it is necessary to have either a sufficiently high frequency of commutation of the alternating voltage with the constant voltage, which eliminates the long-term instability of the converter, or a converter with acceptable long-term stability. The requirement of linearity of the conversion coefficient over a wide dynamic range is not dominant in this case (the converter operates at one operating point of the range). The quality of the converter is determined by the stability of the transfer coefficient, the frequency response, and the speed of response.

In the present work, the temporal stability of a photovoltaic converter (FP) has been investigated; its advantages are a high transfer coefficient, the absence of galvanic connection between the control and controlled circuits, and good short-term stability. The best metrological characteristics of the FP were obtained when low-power incandescent lamps of the NSM type were used as the converter of electrical energy into light, and photoresistors of the SF 2-5 type were used as the light-sensitive element. The temporal instability of incandescent lamps is due to:

1) the presence, in some types of production lamps, of a supporting traverse, which, under shocks and vibration, short-circuits a varying number of turns of the spiral, inevitably leading to jumps in the resistance of the lamp;

2) fastening of the filament by clamping part of the spiral, which, under vibration, causes short-circuiting of turns near the current leads;

3) the semiconductor structure (CuO) of the transitional contact at the point where tungsten is clamped by platinite;

4) the aging process of the filament, which occurs not only from evaporation of tungsten, but also from internal structural changes.

Obtaining small temporal instabilities of incandescent lamps is associated with at least three conditions:

1) by carrying out preliminary conditioning of the lamps before installation in the apparatus (daily heating in the rated mode);

2) by choosing a rational operating mode for the lamp \((1.1 \div 1.2\, I_{\text{fil}})\);

3) by using the design most optimal for measuring purposes. In this case, in our opinion, this is a soffit-type design with the ends of the filament welded to the crosspieces.

The temporal stability of incandescent lamps of various types was recorded (about one hundred experimental samples of the NSM \(6 \times 150\), \(9 \times 60\), \(6 \times 20\), \(12 \times 5\) lamp types). The studies showed that the best temporal stability (on the order of \(0.001\%\) per hour after preliminary heating) is possessed by individual samples of NSM \(6 \times 150\) lamps of soffit construction, with straight ends welded to the current leads. For \(6 \times 20\) lamps with a hook, an instability of the order of \(0.01 \div 0.05\%\) per hour is typical. The criterion of stability for photoresistors is the invariability of their photocurrent (at constant applied voltage and luminous flux), since other parameters practically do not change during operation. Fluctuations of the photocurrent may be affected, directly or indirectly, by various factors: the temperature coefficient, dissipated power, the temperature and relative humidity of the surrounding medium, etc. The principal and, moreover, irreversible component of the instability of photoresistors is the aging process. Aging of light-sensitive elements is caused by a slow change in the properties of the material of the photosensitive layer and in the state of its surface. This process depends to a considerable degree on the fabrication technology and design features. A study of the photoresistors FSK-2P, SF2-2, SF3-2, FSK-G7, SF2-5, OS-4-681-010, and of KFDM photodiodes showed that it is desirable to use SF2-5 photoresistors as the light-sensitive element. Made on the basis of cadmium sulfide, they are distinguished by a small temperature coefficient at low illuminances and by acceptable temporal stability. An attempt was made to find the minimum instability of SF2-5 photoresistors by changing the operating mode of the latter (the applied voltage and the luminous flux). The averaged data from the study of a batch of photoresistors (10 pieces) are illustrated (Fig. 1),

Fig. 1

where the instability of the photoresistors at various applied voltages and luminous fluxes is shown. Curve 1 corresponds to an illuminance at which the resistance of the photoresistor is \(R_{\phi}=145\ \text{k}\Omega\), curve 2—\(75\ \text{k}\Omega\), 3—\(53\ \text{k}\Omega\), 4—\(25\ \text{k}\Omega\), 5—\(10\ \text{k}\Omega\), 6—\(8\ \text{k}\Omega\), and 7—\(2\ \text{k}\Omega\).

Because of the technological scatter of the parameters of photoresistors (in particular, for one illuminance the resistance values of the photoresistors may differ by a factor of \(2 \div 3\)), it is difficult to give practical recommendations (for example, a graph of the dependence of the applied

to the photoresistors on their illumination in order to obtain the minimum instability). Nevertheless, it may be noted that it is most expedient to apply a voltage of \(2 \div 4\) V to the photoresistors. But even in this range, to obtain minimum instability, additional adjustment of the mode is required. For the mode \(U_{\phi}=3\) V, \(R_{\phi}=30\) kΩ, the temporal instability over 5 hours without warm-up was measured for 70 SF2-5 photoresistors. The results showed that the time for the mode to become established (fatigue) for the SF2-5 is 30–40 minutes, and the short-term instability over 10 minutes in the established mode is 0.01%.

Consequently, with optimal adjustment of the PC, stability in the established mode is determined by both the bulbs and the photoresistors. In the first samples of GC, created at the Department of Radio Engineering of TPI, comparators of non-simultaneous comparison were used; the instability of their output voltage was determined mainly by the comparator, since the instability of the reference source was 0.001–0.003% per hour. The temporal drift of the output voltage of the indicated GCs was:

\[ \begin{aligned} \text{GC-4} &\ —\ 0.03\% \text{ in } 15 \text{ min};\\ \text{GC-5} &\ —\ 0.03\% \text{ in } 15 \text{ min};\\ \text{GC-6} &\ —\ 0.01\% \text{ in } 10 \text{ min};\\ \text{GC-7} &\ —\ 0.01\% \text{ in } 10 \text{ min}. \end{aligned} \]

The average instability per minute is less than 0.005%. Long-term instability of the PC was eliminated by comparing the alternating voltage with a constant voltage by means of manual balancing at the moment of measurement. One way of improving the metrological characteristics of the GC was the creation of differential comparators based on two PCs with identical monotonic aging characteristics. Having measured the time characteristics of incandescent lamps and photoresistors, it is not difficult to select two PCs with approximately analogous drift curves. In such a converter the output drift and warm-up time are smaller than in converters of non-simultaneous comparison. The GC-10 with a differential PC has a temporal drift of less than 0.01% over 10 minutes. The drift determined by the nonidentity of PC pairs can be reduced by changing the transfer coefficients of the pairs (lamp currents, illumination). After the PC has been mounted in the generator, final end-to-end adjustment for minimum instability is carried out. The GC-8 with a comparator adjusted in this manner, after three hours of warm-up, has an instability on the order of 0.005% over 4 hours of operation without additional manual readjustments.