ON IMPROVING THE COMMUTATION OF AN ELECTROSTATIC GENERATOR

A. M. Kuptsov, V. M. Fedyakin

(Presented by the scientific seminar of the Department of Theoretical Foundations of Electrical Engineering and the laboratory of rotary electrostatic generators of the Scientific Research Institute of Nuclear Physics)

One of the main problems that must be solved in creating a reliable industrial electrostatic generator with conducting carriers (ESGP) is the problem of commutation [1, 2, 3, 8]. Work carried out at Tomsk Polytechnic Institute on the study of ESGP with a large number of carriers per pole pitch made it possible to obtain a sufficiently complete understanding of the main commutation phenomena in them and thereby to create the prerequisites for solving this problem [3, 8].

In the present work the results are set forth of an experimental investigation of several methods for improving commutation in the above-mentioned electrostatic generators.

By improvement of commutation is meant the creation of a number of conditions bringing it closer to the optimal one, i.e., ensuring the maximum generator current with the least destruction of the brushes, carriers, and rotor dielectric.

Reduction of erosion of the carriers and brushes

It has been shown in [3, 8] that in the discharge system of the generator, during commutation, compensation of the charges of the carriers and inductors takes place. In this process energy is released that is proportional to the magnitude of the compensated charge and to the potential difference between the commutating carrier and the inductor. Owing to the small value of the active resistance of the connecting conductors of this circuit, practically all the energy stored by the carrier–inductor capacitance is released in the nonlinear resistance of the gas discharge in the carrier–brush gap, causing destruction of the latter. The intense gas discharge accompanying the deposition of excess charges on the carrier leads to destruction of the carriers and the brush in the charging system.

The questions of destruction of electrodes in strong electric fields for purposes of electric-spark machining of metals have been studied in sufficient detail. Therefore, when considering the conditions that reduce the destruction of the commutating elements in an ESGP, it makes sense to dwell on the principal factors determining these destructions.

It has been shown in [4–6] that the magnitude of erosion of the electrodes of a discharge gap is determined, first of all, by the magnitude of the energy expended—

dissipated in the given discharge gap. For one and the same energy, the destruction is greater the greater the power of the discharge. It is also known [4, 5] that electrode erosion increases if the energy of the discharge gap is released in a limited space. Conversely, dispersal of the released energy of a single spark channel over several channels leads to a noticeable reduction in erosion.

Taking the foregoing into account, the main directions for obtaining acceptable commutation in an electrostatic generator may be:

1. Dissipation of the energy stored by the capacitance of the conveyor—inductor outside the conveyor—brush discharge gap or, at least, reduction of the portion of the energy dissipated in the gas-discharge resistance.

2. Reduction of the discharge power.

3. Dispersion of the energy released in the discharge gap over several channels.

A reduction in the fraction of energy released in the spark channel can be achieved by introducing an additional active resistance into the discharge circuit.

The discharge power will decrease when an additional lumped inductance, connected in series with the active resistance, is included in the discharge circuit.

Finally, dispersion of the energy over several channels can be achieved by selecting a certain brush configuration. This may be, for example, a comb of points of different lengths.

Below are the results of an experimental verification of the stated premises on two specimens of electrostatic generators described in [3, 8].

Inclusion of Ballast Resistances in the Discharge Circuit

In accordance with [3, 8], in the discharge system during commutation of the conveyor the relation

\[ W_{\mathrm{з}} = W_{\mathrm{p}} + \Delta W, \]

holds, where \(W_{\mathrm{з}}\) is the energy stored by the conveyor by the moment of commutation;
\(W_{\mathrm{p}}\) is the energy released in the commutation zone in the resistance of the gas discharge;
\(\Delta W\) is the energy loss in the connecting conductors.

Introducing a ballast resistance into the connecting circuit and taking into account the energy released in it, we obtain

\[ W_{\mathrm{p}} = W_{\mathrm{з}} - W_{\mathrm{б}} - \Delta W. \]

This relation establishes the interconnection between the energy dissipated in the brush—conveyor discharge gap and causing destruction of this gap, and the energy dissipated in the ballast resistance. By estimating the value of \(W_{\mathrm{б}}\) for various ballast resistances, it is easy to trace the regularities in the change of \(W_{\mathrm{p}}\), and consequently the degree of electrode erosion.

To calculate \(W_{\mathrm{б}}\), as well as the rate of its change, we oscillographed the current in the brush—conveyor circuit. From the results of processing the current oscillograms in the short-circuit regime, curves 1, 2, and 3 in Fig. 1 were constructed.

Curve 1 characterizes the law of variation of the energy released in the commutation zone during one commutation act as a function of the magnitude of the ballast resistance. Curve 2 characterizes the chan-

decrease in amplitude, and curve 3—the change in the steepness of the leading edge of the discharge-current pulse.

All curves are given for unchanged values of the excitation voltages, which ensured an almost unchanged value of the current recorded by the microammeter.

Fig. 1. \(A\)—ratio of the amplitudes of current pulses with and without ballast resistance connected

Figure 2 gives the short-circuit current characteristics obtained with different ballast resistances for rubbing brushes (curve 1) and for brushes made in the form of points (curves 2–5).

Since, with rubbing brushes, the insertion of ballast resistances from 0 to 5 MΩ caused no changes in the characteristic, Fig. 2 shows only one curve.

As follows from the curve shown in Fig. 1, the sharpest decrease in the energy released in the discharge gap is achieved when resistances up to 300 kΩ are connected. In this same interval the amplitude and the steepness of the front of the current pulses change most sharply. Since the discharge power is directly proportional to the amplitude and the steepness of the current pulses [7], on the basis of the behavior of curves 2 and 3 we conclude that the insertion of ballast resistances effectively limits the discharge power as well.

Fig. 2. 1—brushes rubbing at \(R_b\) from 0 to 5 MΩ; 2–5—point brushes
(\(R_b = 50,\ 510,\ 1000,\ 2700\) kΩ)

The curves of Fig. 2 establish the relationship between the generator current at different excitation voltages and the magnitude of the ballast resistances. It follows from the figure that, with rubbing brushes, the insertion of balla-

…contact resistances within broad limits does not cause a decrease in the generator current. This means that in this case the reduction of the discharge energy and power is achieved without worsening the generator parameters. This is in full agreement with the concepts of commutation processes as processes of charge compensation [3, 8].

Indeed, assuming the process of charge commutation to be analogous to the discharge of a capacitor closed through \(R_b\), we find that

\[ R_b \ll \frac{t_k}{5C}, \]

i.e., \(R_b\) is limited only by the commutation time \(t_k\). Since with rubbing brushes \(t_k\) can vary over broad limits up to \(T_k\)—the commutation repetition time [3]—\(R_b\) can be increased, without reducing the generator current, to the value

\[ R_b \ll \frac{T_k}{5C}. \]

The situation is different with brushes made in the form of points, which provide only gas-discharge commutation. The commutation time here is limited by the lifetime of the gas discharge, which also depends on the ballast resistance. Although it increases as \(R_b\) increases, the generator current at the same \(U_v\) becomes somewhat smaller.

This can be explained as follows. The insertion of a resistance leads, as is known [9], to a decrease in the rate at which energy enters the discharge gap and, consequently, to an increase in the extinction voltage of the gas discharge [10].

The latter determines the magnitude of the charge uncompensated during commutation. As \(R_b\) increases, this magnitude grows, which explains the change in the generator current. For example, at resistances greater than \(20 \div 40\) MΩ the generator current decreases practically to 0.

It is important to note that when the excitation voltage is increased, the reduction in generator current due to the ballast resistances becomes less significant. Thus, in the saturation region of the current this difference lies within the measurement error, amounting to \(5 \div 10\%\).

This is in full agreement with the concepts of gas-discharge development at different overvoltages [9].

A study of the generator operation in the load mode showed that the inclusion of ballast resistances up to 500 MΩ practically does not change the load characteristics.

From a comparison of the results of analyzing the generator characteristics and measuring the energies in the discharge gaps, it is seen that the optimal value of the resistance \(R_b\) is of the order of 150–300 MΩ. A further increase is ineffective, since the reduction of energy in the commutation gaps is insignificant. Confirmation of the effectiveness of reducing erosion of the commutating elements by means of ballast resistances was obtained after inspection of the commutating elements of a generator that had operated for 200 hours in a nitrogen atmosphere at a pressure of 10 atm. The generator, with a rotor diameter of 140 mm and with two pairs of poles, operated in the short-circuit mode with brushes made of steel sewing needles. The gaps between the needles and the rotor were set using a special gauge. Excitation of both pairs of poles was carried out from a single excitation source, and the high-voltage circuits (one of which included \(R_b = 100\) MΩ) were separated in order to record the currents. Inspection showed that the brush in the circuit with the ballast resistance

burning was significantly less. The gaps, relative to those initially set, changed by a factor of 0.3 in the circuit with ballast resistance, and by a factor of 2 in the circuit without resistance.

Thus, regulation of the energy and power of commutation discharges by switching in ballast resistances appears very promising for achieving acceptable commutation in an electrostatic generator with conducting conveyors.

Switching in Inductance and Ballast Resistance

When switching inductance and resistance into the circuit of the commutating conveyor, we assumed that this would lead to a substantial decrease in the amplitude and steepness of the leading edge of the discharge-current pulses. At the same time, a sharp reduction in discharge power was expected. Experimental verification of this assumption did not confirm it.

We switched inductances of the order of \(50 \div 300\) mH with various ballast resistances into the conveyor circuit. The oscillograms obtained show that, owing to the switched-in inductance, the shape of the current pulses changes only insignificantly. Only a slight decrease in the steepness of the pulse front is observed, with a small increase in its duration.

The reduction in the power of the commutation discharge when inductance is switched in is accompanied by a very slight decrease in the brightness of the discharge glow.

It should be noted that a further increase in inductance (from 300 mH to 1 H), with the ballast resistance kept unchanged, does not lead to any substantial changes, despite the fact that approximate calculations for this value of inductance indicate the possibility of charging (or discharging) the conveyor over a “half-wave,” i.e., with a current waveform close to sinusoidal. Evidently, this is explained by the large interturn capacitances of the inductance used. Breakdown of the inductance is also not excluded, since at the moment of commutation a voltage

\[ U = L \frac{di}{dt}, \]

commensurate with the excitation voltage, is applied to the inductance.

Taking the above into account, one may conclude that regulating the power of commutation discharges by switching in inductances is inadvisable. More effective is the switching in of ballast resistances, which simultaneously reduce both the energy and the power of the commutation discharges.

Dispersal of the Energy of Commutation Discharges by Brushes Made in the Form of a Comb

It was noted above that, when the energy released in a single discharge channel is dispersed among several channels, the erosion of the electrodes decreases. To verify this during generator operation, brushes made in the form of a comb were installed instead of sharp-point brushes. The points of the comb were arranged at different distances from the end surface of the rotor; in this case, the point that first commutates with the conveyor has the greatest distance from the rotor surface. The next one has the smallest.

The length of the comb was varied up to a value equal to twice the distance between the carriers. The number of points in the comb was set from 5 to 15. By selecting the spacing between the points, multiple commutation of the carrier with each of the comb points was achieved. In each act of commutation, only part of the charge of the carrier and of the inductor was compensated. The form of the gas discharge was determined by the excitation voltage, the operating mode, and the distance of the comb points from the end surface of the rotor.

Figure 3 shows an oscillogram of current pulses during commutation of the carrier by means of the comb. It is evident from the oscillogram that commutation occurs repeatedly. Each act of commutation in the brush circuit produces current pulses whose amplitude is much smaller than the amplitude of the pulses in single commutation [3, 8]. The pulses are irregular and of different shape, which is explained by the nature of the gas discharge from each of the comb points. Indeed, visual observations of the commutation zone show that both weak spark discharges and corona discharges develop from the points. By changing the operating mode of the generator and the position of the comb, a weak corona discharge uniformly distributed over all the points of the comb can be obtained.

Fig. 3

Analysis of the current oscillograms shows that the power of each individual discharge is incommensurably small in comparison with the power of a single discharge during commutation by rubbing brushes or in the form of single points. This, consequently, accounts for the smaller erosion of the commutating elements, despite the fact that the total amount of energy released in the commutation zones has not changed as a result of dispersing the commutation.

One-hundred-hour operation of the generator in a nitrogen atmosphere (10 atm) showed that burning of the comb needles is insignificant. It was not possible to establish changes in the gaps between the rotor and each of the comb needles by means of a gauge.

Reduction of Destruction of the Rotor Dielectric

Along with the destruction of the carriers and brushes, the discharges accompanying commutation also destroy the rotor dielectric. Destruction of the dielectric always occurs only on the side of the carrier running onto the brush. This is explained by the fact that the spark discharge between the brushes and the carrier moving toward it partially passes over the surface of the dielectric adjoining the end of the carrier. Owing to the high temperature of the spark channel, the dielectric partially evaporates and contaminates the working volume of the generator, which ultimately leads to a reduction in its energy characteristics. An example of such destruction is shown in Fig. 4, where a part of the rotor with one carrier is shown. The region of dielectric destruction shown in this figure has a cone-like shape of dark color. The base of the cone adjoins the end of the carrier. Since the destruction is caused by spark discharges, the principal measure for reducing destruction of the rotor dielectric should be the reduction of the energy and power re-

occurring in the spark channel. For this purpose one may use the methods indicated above. However, destruction of the dielectric can also be eliminated by selecting a certain configuration of the end surface of the rotor. For example, by making the carriers protrude above the surface of the dielectric.

In this case the discharge path between the brush and the carrier passes at a sufficient distance from the dielectric and, consequently, the dielectric is not destroyed. However, such a rotor design is not advisable from the standpoint of the electric strength of the gap between adjacent carriers. In order not to impair the electric strength of the indicated gap and to eliminate melting of the dielectric, on one of the rotors of the experimental generator we milled out the part of the dielectric adjacent to the commutating ends of the carriers in the form of crescents.

Fig. 4

With carrier diameters of 3.5 mm and distances between their centers of 9 mm, the dielectric was removed at the surface from one third of the carriers to a depth of 2 mm, from another third to a depth of 4 mm, and from some of the carriers the dielectric was not removed at all. An additional check established that the electric strength of the gap between adjacent carriers did not deteriorate as a result.

After one hundred hours of operation of the generator in a nitrogen atmosphere under pressure, it turned out that the surface of that part of the rotor around whose carriers the dielectric had been removed remained unchanged, both for 2-mm and for 4-mm recess depths. In the control part of the rotor the dielectric was melted in a manner analogous to Fig. 4.

Thus, the experimental data indicate the possibility of eliminating destruction of the dielectric by giving the end surface of the rotor a certain configuration.

Conclusions

The investigations carried out to improve commutation in an electrostatic generator showed the following:

1. The inclusion of ballast resistances in the generator discharge circuit effectively reduces the energy and power of the commutation discharges and, as a consequence, leads to a reduction in electrode erosion.

2. The inclusion of an inductance in the generator discharge circuit is inadvisable.

3. Performing commutation by means of a comb leads to the distribution of the commutation discharges over the entire length of the brush, with a sharp decrease in the power of each of the discharges.

4. In order that the dielectric surface of the rotor not be destroyed under the action of commutation discharges, crescent-shaped recesses should be made around the carriers.

References

1. N. I. Felici. Journal de Physique et le Radium, S. 8, v. 10, 1949.

2. V. I. Levitov, A. G. Lyapin. Electrostatic Generators with a Rigid Rotor, Part I. Tsintipriboelektroprom, Moscow, 1963.

3. A. M. Kuptsov. Physical processes in electrostatic generators with carrier-conductors embedded in a solid dielectric. Dissertation, Tomsk, 1968.

4. B. A. Zolotykh. Physical Principles of Electrospark Machining of Metals. GITTL, Moscow, 1953.

5. N. F. Pobezhimov, A. T. Chepikov. Breakdown of Solid Dielectrics and Semiconductors. Energiya, Moscow, 1964.

6. Electrospark Machining of Metals. Transactions of TsNIIEL-ELEKTROM, Academy of Sciences of the USSR, Issue 1, 1957.

7. G. V. Zeveke, P. A. Ionkin et al. Fundamentals of Circuit Theory. GEI, Moscow–Leningrad, 1963.

8. A. M. Kuptsov, V. V. Patsvich. Elektrichestvo, No. 7, 1968.

9. A. I. Gavrilin et al. Breakdown of Solid Dielectrics. Energiya, Moscow, 1964.

10. I. S. Marshak. Pulsed Light Sources. GEI, Moscow–Leningrad, 1961.