ON THE MEASUREMENT OF PULSE VOLTAGES WITH A SPHERE GAP

S. S. PELTSMAN, B. V. SEMKIN

(Presented by the scientific seminar of the Research Institute of High Voltages)

The breakdown voltage of a sphere gap over a large interval of times does not depend on the duration of voltage application or on the law of its variation in time; however, at short exposure times, the discharge delay plays an essential role. Existing standards permit the use of calibration tables when the voltage rise time to the maximum is not less than 1 μsec and the fall time from the maximum to half-value is not less than 5 μsec; in this case the measurement error must not exceed 3%. Many studies have been devoted to the relation between the breakdown voltage of a sphere gap and the prebreakdown time [1—5].

Bellashi and Teague [1] studied the breakdown of a number of gaps between spheres of 6.25, 25, and 200 cm and showed that, for times to breakdown of up to 1 μsec, the measurement errors exceed 3%.

Hagenguth [3—4], for 25 cm spheres in a gap 4 cm long and with a rate of rise of 1000 kV/μsec, showed that, relative to a prebreakdown time of 2 μsec, the overvoltages required for breakdown in 1, 0.5, and 0.2 μsec are respectively 13, 54, and 91%. Illumination of the gap by a mercury lamp in these experiments did not affect the results, and Hagenguth suggested that the overvoltages during breakdown at short times are determined rather by the process of spark-discharge formation than by the statistical time lag.

Park and Cones [5] investigated the volt-second breakdown characteristics of a 6 cm gap for spheres 12.5 and 25 cm in diameter under negative-polarity impulses over a wide interval of prebreakdown times (0.03—100 μsec).

The deviations of the breakdown voltages from the values specified in the standards, for a time from the moment of voltage application to the occurrence of breakdown of 2 μsec, reached 10%.

Summarizing the foregoing, it should be noted that measurement of impulses with amplitudes amounting to hundreds of kilovolts and with a prebreakdown time of less than 2 μsec by means of sphere gaps cannot guarantee errors within 3%. The systematic data available in the literature [1] on overvoltages during breakdown with a short prebreakdown time ($t < 1$ μsec) are given relative to the voltage causing breakdown after 2 μsec, when the discharge-delay time is still quite substantial [5]. Data on breakdown voltages over a wide time interval are limited to individual values.

tions of the interelectrode distances [3—5]. As was shown in [1], the impulse coefficient depends on the ratio of the gap length to the sphere diameter, which does not allow the results of studies [3—5] to be extrapolated to other interelectrode gaps.

In the present work, results are given for systematic measurements of impulse breakdown voltages of a number of gaps for spheres of diameter 12.5 cm for both impulse polarities.

The rate of rise of the voltage was regulated by means of the generator voltage and by changing the parameters of the discharge circuit. To maintain the linearity of the voltage rise up to breakdown, the amplitude of the impulse generated by the source was approximately twice the breakdown voltage of the gap.

Fig. 1

For recording, a capacitive voltage divider and an OK19-M2 cathode-ray oscilloscope were used. As a source of ionizing radiation, discharges in the generator spark gaps were used as the most effective means of ionizing illumination [6—7]. Five spark gaps 2 cm long were arranged one above another at intervals of 40 cm. The distance from the gap under investigation to the nearest ionizing gap was 150 cm, in accordance with IEC requirements. The arrangement of the spherical measuring gaps relative to the ground and to extraneous objects is shown in Fig. 1 and corresponds to standard requirements. Grounding was carried out with a copper busbar 50 mm wide.

Compliance with standard conditions and introduction of a correction for the temperature and density of the surrounding medium ensure an error in calibration of the divider of not more than ±3%. The total error of the investigated values of the breakdown voltages of the sphere gaps did not exceed ±5.5% for impulses of negative polarity and ±5% for positive polarity.

The relative error of the values obtained at different durations of the voltage up to breakdown was not more than 2%.

Experimental Results

Voltage pulses with different rates of voltage rise were applied to the gap between spheres of diameter 12.5 cm. The volt-second characteristics presented in this work characterize average values obtained from 20 points. The dependences were obtained with the ambient temperature varying within 18–21°C, atmospheric pressure 754–762 mm Hg, and relative humidity 50–60%. More than 90% of all points differed from the constructed characteristics by no more than ±2.5%. Figure 1 shows the volt-second characteristics of a spherical gap with sphere diameter 12.5 cm, for positive polarity of the high-voltage electrode and for gaps 8, 6, and 4 cm long (curves 1, 2, and 3, respectively).

Figure 2 gives the volt-second characteristics of analogous gaps for negative pulse polarity. The dashed line gives the characteristic of a 12.5 cm gap with a distance between electrodes of 6 cm, obtained in [5].

Fig. 2

From the volt-second characteristics, dependences were plotted of the relative overvoltages at breakdown of the sphere gap in the range of prebreakdown times (0.06–10 μsec), relative to the breakdown voltages indicated in the calibration tables and to the corresponding values toward which the volt-time characteristics tend as the prebreakdown time increases. These dependences are shown in Fig. 3. Curves 1, 2, and 3 are given for gaps of 4, 6, and 8 cm with pulses of negative polarity; curves 4, 5, and 6 are for the same gaps, but with positive pulse polarity.

The pulse coefficient \(K\) for negative pulse polarity is substantially higher than for positive polarity. This fact is explained by the field in the gap between the spheres not being strictly uniform, and the polarity may influence the development of the spark discharge.

The impulse coefficient of a sphere gap varies in a characteristic way when the ratio of the gap length—\(S\)—to the sphere diameter—\(D\)—is changed. With positive polarity of the high-voltage electrode and a small distance between the spheres (\(S = 4\) cm; \(S/D = 32\%\)), the impulse coefficient is relatively high owing to shielding of the gap from the radiation source. The influence of irradiation in such gaps has been proven

Fig. 3

by Meek [6], who showed a distinct difference in the breakdowns of irradiated and nonirradiated gaps 3 cm long for spheres 12.5 cm in diameter. As the ratio \(S/D\) is increased, the statistical time lag decreases, while the formation time increases. In this case the impulse coefficient, reaching a minimum value at \(S/D = 50\%\) (\(S = 6\) cm), rises again and at \(S = 8\) cm is close to the value of \(K\) at \(S = 4\) cm.

With negative polarity of the impulse, as \(S/D\) changes the impulse coefficient varies within narrower limits than with positive polarity. This is due to the weaker influence of ionizing radiation [6] for negative polarity of the electrode. Only when the time to breakdown is reduced below 0.2 μsec does \(K\) at \(S = 4\) cm increase somewhat more rapidly than for gaps of 6 and 8 cm, since the statistical lag in small gaps, as the prebreakdown time decreases, still plays a certain role.

The results of the experiments we carried out confirm that a voltage rise time to breakdown of 1 μsec, and even 2 μsec, does not ensure preservation of the accuracy guaranteed in the standards. Thus, for negative polarity of the impulse, the additional error in voltage measurement caused by the discharge lag amounts to as much as 10% for a prebreakdown time of 2 μsec (according to [5], as much as 11.5%). With positive polarity, the value of the error due to discharge lag for a voltage application time of 2 μsec is about 4%.

Measurement of high-voltage pulses, with breakdown on the front of a linearly rising voltage, with a total error of 3.5%, is possible for pulses of negative polarity only when the time to breakdown is \(t > 20\ \mu\text{s}\), and for pulses of positive polarity when \(t \ge 8\ \mu\text{s}\). In this case, the error due to the discharge lag will be no more than 0.5%.

References

1. Bellaschi P. L., McAuley P. H., Electr. Journ., 31, 228 (1934).
2. Bellaschi P. L., Teague W. L., Electr. Journ., 32, 120 (1935).
3. Hagenguth J. H., Trans. Am. Inst. Electr. Eng., 56, 67 (1937).
4. Hagenguth J. H., Trans. Am. Inst. Electr. Eng., 60, 803 (1941).
5. Park J. H., Cones H. N., Spark-gap flashover measurements for steeply rising voltage impulses. “J. Res. Nat. Bur. Standards,” 1962, C66, No. 3, 197—207.
6. Meek J. M., Journ. Inst. Electr. Eng., 93, 97 (1946).
7. Garfitt D. E. M., Proc. Phys. Soc., 54, 109 (1942).