r^OUTBUT CURRENT

i^PÖtI0|i8ßeNt

OUTPUT VOLTAGE

Figure 6. Load circuit of a resonant-circuit constant-current regulator is suddenly short-circuited

Figure 7 (r ight). A resonant-circuit constant-current regulator is suddenly switched on a ful ly loaded Mazda-lamp circuit

provided the input voltage is held constant. If an input and an output transformer are used, the constancy of the output current is decreased as a result of the input-trans-former-voltage regulation and the change in current transformation ratio caused by the magnetizing current of the output trans-former. The magnetizing current increases with increasing output load. Thus the out-put current is somewhat less at full load than a t no load (load circuit short-cir-cuited) . For these reasons the Air Forces permitted a current variation of plus or minus two per cent.

The highest constancy of the output cur-rent occurs for noninductive or slightly capacitive loads. Heavy inductive con-stant-power-factor loads decrease the ratio of full-load current to the short-circuited current (no load). But constant power-factor capacitive loads increase said ratio, and it may reach undesirable values for ex-tremely capacitive loads. Figure 3, A and B, explain said relations.

An inductive or capacitive output load not only effects the constancy of the output current but also effects the voltages on the capacitors and reactors of the monocyclic square. For inductive loads the capacitor voltages are higher and the reactor voltages are lower than for noninductive loads of the same kilovolt amperes. For capacitive loads, the capacitor voltages decrease, and the reactor voltages increase.

Calculations based on the analytic methods developed by Steinmetz are very accurate. However, in practice a quicker inside view of the operating characteristics is obtained by vector diagrams and particu-larly by so-called circle diagrams. Figure 4 shows a circle diagram for an L-connected circuit. This diagram together with dia-grams for a monocyclic square and other special cases appeared in a paper by H. Weichsel.1

Mr. Kevern touched in his paper on the effect of frequency changes on the output current. He cites a test on a regulator oper-ating at 50 and 60 cycles.

The theoretical relation between the out-put current and impressed frequency for a monocyclic square without input and output transformers is given by Figure 5. With input voltage held constant, the output cur-rent varies only slightly over a very large range of frequencies. I t is a simple matter to consider the effect of input and output transformers if used.

Figure 6 shows the change in useful out-put current and voltage when the useful load circuit is suddenly short-circuited.

nompvr VÖLTAOE ;.

OUTPUT VOLTAGE

Figure 8. A ful ly loaded resonant-circuit constant-current regulator is suddenly discon-

nected from its supply circuit

The effective value of the output current changes only slightly during the very short transient period. The output voltage dropped to zero at the instant the short cir-cuit in the load circuit occurred.

Figure 7 shows the output current and voltage as well as input current to the regu-lator when regulator is suddenly switched on a fully loaded circuit. Only a small change in the output current occurs during the very short transient period. The out-put voltage increases gradually due to the temperature increase of the Mazda lamps' filaments. The latter results in increased watt output. Consequently, the input current to the regulator increases nearly proportional with the output voltage.

Figure 8 represents, in the two upper pic-tures, the output current in the useful load circuit when the unit is suddenly discon-nected from its supply circuit. The lower picture shows the corresponding input cur-rent to the regulator. The complete ab-sence of oscillations or transients is note-worthy.

The opening of the input circuit of a fully loaded monocyclic square results in only a small arc, contrary to the experience when opening the input circuit of moving-coil transformers. This difference is partly brought about by the high power factor of the input current to a monocyclic square, whereas the input circuit to a moving-coil transformer is highly inductive.

The foregoing gives some of the funda-mental laws which must be considered in the design as well as the operation of constant-current resonant regulators. There also exist certain problems in the manufacturing of these units. For instance, it is readily realized tha t neither the required capacitors nor the reactors can be manufactured to ex-actly the specified values. Theoretical study carried out on this very subject has shown that , in spite of said manufacturing variations, it is possible by suitable means to obtain a surprisingly high degree of con-stancy of the output current.

Mr. Kevern's paper forms a very impor-tant contribution to the problem of constant-current a-c circuits and doubtlessly will re-sult to a large extent in a better understand-ing of the great field in which constant-cur-rent resonant circuits can be used very ad-vantageously.

R E F E R E N C E

1. C O N S T A N T - C U R R ^ T REGULATORS, H. Weich-sel. Wagner Electric Corporation, St. Louts, Mo., 1941.

Electric Power in a Steel Plant

Discussion of paper 45-140 by R. Ø. Graham, published in A IEE TRANSAC-T I O N S , 1945 , December section, pases 8 1 6 - 8 .

C. G. Archibald (Westinghouse Electric Corporation, Buffalo, N. Y. ) : This paper presents in a condensed form the problems involved and the methods pursued by Mr. Graham and his associates in solving a very difficult problem in continuity of service. This paper shows tha t very thorough studies have been made of the system load requirements, the method of relaying, switching, and other problems associated with large load in a very small area, handling processes that require the highest degree of continuity.

The magnitude of this load may call to your attention that this is no small load; being, on the order of the amount of power consumed by central station load at peak, by a city of 350,000 to 400,000 population. However, it does have two unusual features about it that makes it a more interesting load. These are—first, the very high load factor from 24-hour operation, on the order of 70 to 75 per cent; and second, practically unity power factor at all times.

994 Discussions AIEE TRANSACTIONS

The solution of operating problems by using fast relaying (both phase and ground), and sectionalizing of the plant, with double feeds wherever possible, appears to be the most economical and practical method of Tapid sectionalizing and maintenance of continuity to critical loads.

If all industrial plants placed an equal amount of emphasis on their electrical engi-neering problems, and solving these as we1l as this property has, greatly improved service would be maintained.

Capacitors for High-Frequency Induction-Heating Circuits

Discussion and authors* closure of paper 45-157 by F. M. Clark and M. E. Scoville, pub-lished in AIEE TRANSACTIONS, 1945, November section, pages 791-6.

Philip Bliss (The New Britain Machine Company, New Britain, Conn.): The writers have presented the problems of timing capacitors for high-frequency indus-trial use very well. The new oil and new method of construction make a much smaller, neater capacitor than those in present use employing mineral oil dielectric. Incidentally, the 15-kw industrial heater shown appears to be of far superior con-struction than previous models.

L. W. Foster (General Electric Company, Schenectady, N. Y.) : There are two points that I should like to make in connec-tion with this paper. First, Lectronol is a synthetic dielectric liquid whose character-istics may be controlled very closely in the manufacturing process; whereas mineral oil and castor oil are natural materials which will have a considerable variation in elec-trical properties depending upon their source of supply.

Secondly, the difference in dielectric strength a t high frequency of solids and liquids can be attributed mainly to thermal effects. The dielectric liquid is a good thermal insulator but in the case of the parallel-plate capacitor free circulation is possible, therefore, hot spots will not be readily formed. In the case of solid insula-tion like styrene or mica the hot spot-temperature may be very high compared to the capacitor case temperature, there-fore, the solid dielectric will have lower breakdown strength than the equivalent liquid dielectric because heat cannot be transmitted rapidly enough through the solid dielectric material.

L. J. Berberich (Westinghouse Electric Corporation, East Pittsburgh, Pa . ) : Every manufacturer of high-frequency heating equipment, during the war especially, has felt the need for a capacitor suitable for application in this type of equipment. Messrs. Clark and Scoville have presented a very satisfactory solution to this problem in the liquid-filled capacitor which promises to hold its own even in the postwar period when an abundant mica supply is again ex-

pected. Our analysis of the problem led us to the same conclusion reached by Messrs. Clark and Scoville; namely, that a liquid-filled capacitor should be the solution. We want to emphasize, even more strongly than the authors of the paper have, tha t liquid impregnated paper, despite its enviable rec-ord as a dielectric in low-frequency capaci-tors, is completely unsuited for high-voltage operation in the range from 100 kilocycles to 1 megacycle. The power factor of impreg-nated paper1 in this frequency range is from 20 to 50 times the somewhat arbi-trarily chosen maximum of 0.05 per cent which appeared feasible for this type of capacitor.

Liquid-filled capacitors are not funda-mentally new. Every radio amateur prob-ably has had the experience of submerging air capacitors in oil to cure breakdown difficulties. We agree with the authors that mineral oil, however, is not satisfactory for this application chiefly because its dielectric constant is so low tha t the size of the capaci-tor for a given rating would be larger than even the compressed gas capacitors which have come into use in the last few years. This meant tha t the solution had to be in a polar liquid whose dielectric constant is at least twice tha t of mineral oil. Lectronol obviously is such a polar liquid. We would add to the requirements for the liquid listed in the paper, tha t it must have a sufficiently low viscosity so that the increase in loss due to rotating dipoles is not serious in the frequency and temperature ranges of operation. Evidence of this type of loss is given in Figure 4 of the paper a t frequencies above one megacycle. This loss is lowered at a given frequency by an increase in temperature because of a reduction in vis-cosity, which is also shown in Figure 4 by the relative positions of the 25 degree centi-grade and the 75 degree centigrade curves at the higher frequencies. This is entirely normal and the expected behavior of polar liquids.

The first polar liquid investigated by us was mixture of chlorinated hydrocarbons which belong to the class of liquids known as the "Askerels." This liquid has long been used as coolant and dielectric in fire-proof transformers. By proper purification means the power factor could easily be re-duced to a reasonably low value in the 100 kilocycles to 1 megacycle operating range. Chlorinated aromatic hydrocarbons, as is well known, are quite stable to oxidation and have a dielectric strength at 60 cycles superior to tha t of oil. When tested a t 400 kilocycles, however, we found, to our great surprise, tha t the dielectric stability is ex-tremely poor. We further observed what we believe to be a new phenomenon.

This observation was made by placing the liquid in a test gap similar to the one used by authors, and applying 2,000 to 5,000 volts in the frequency range from 200 to 800 kilocycles with the gap separation set a t from 0.5 to 0.75 inch. Within a short time after the application of high-frequency voltage, one or more scintillating streamers were observed as they progressed gradually under the liquid from one electrode to the other. Within a very short time after the first streamers bridged the gap, numerous other streamers appeared. The electrode gap and the region surrounding it gradually became filled with more streamers, and fin-ally the field configuration of the gap was

beautifully outlined with scintillating lines of light. The liquid eventually darkened and became so hot due to increased losses that it boiled. Carbon was formed, result-ing in complete destruction of the liquid within a very few minutes, without break-down in the usual sense. The higher the frequency, the lower the voltage necessary to produce this phenomenon. We have never developed a satisfactory explanation for this marked instability of chlorinated hydrocarbons in a high-frequency field. We would like to ask the authors if they have also observed this phenomenon, and, if so, would they venture an explanation.

This poor stability is not only character-istic of the chlorinated hydrocarbons bu t is also true of certain hydrocarbon phosphates, nitrated hydrocarbons, and some other polar liquids. Esters of certain dibasic organic acids as well as a few other liquids, on the other hand, were found to be quite stable. As a matter of fact, it was found tha t if a substantial portion of a stable ester is added to the unstable chlorinated hydrocarbon, the streamer formation could be tempo-rarily inhibited. However, later long-time tests indicated that if appreciable quantities of chlorinated hydrocarbons are present in the mixture, instability always eventually results. Finally, of course, the chlorinated hydrocarbons were eliminated and a liquid developed which has very similar properties to those for the liquid described in the paper.

R E F E R E N C E

1. CHARACTERISTICS OF CHLORINATED IMPREG-NANTS IN D-C PAPER CAPACITORS, L. J. Berberich, C. V. Fields, R. E. Marbury. AIEE TRANSAC-TIONS, volume 63, 1944, pages 1173-9.

C. V. Fields (Westinghouse Electric Cor-poration, East Pittsburgh, Pa . ) : The authors have apparently developed a well integrated radio-frequency capacitor which is a definite contribution to both the capaci-tor and the high-frequency heating arts.

After trying a number of solid and com-posite dielectrics we, too, were forced to re-vive the liquid-filled capacitor which has long been dormant, primarily for lack of a suitable liquid. In addition to studying various liquids, we have encountered several interesting design problems touched on lightly by the authors.

In testing various liquids, it was found that even those considered suitable for use in radio-frequency capacitors show glowing and bubble formation prior to breakdown. If there are any sharp points in the structure, the stress may be such as to cause these phe-nomena to take place even a t working stresses. Thus it is necessary to round all metal parts adjacent to stressed liquid not only to increase the ultimate breakdown, but also to insure stability in normal operation.

The authors seem to imply tha t the bush-ing losses are large compared with the capacitor dielectric loss, which at 1,000 kva and 0.0001 power factor would be 100 watts. We should not expect tha t the two bushings would have more than a fraction of this loss. I t is obvious, however, tha t a suitably com-pact bushing for such voltages and currents at these frequencies must not only be liquid filled but must be designed with particular regard to the generation and elimination of heat. The problem is greatly simplified by the use of Zircon porcelain1 which not only

1945, VOLUME 64 Discussions 995