forced into the slots, N, N. The very rapid mode of mounting keeps the agglomerations in good metallic contact with the support.

As to the zincs, they are supported by metallic plates, F, F, passing into rectangular holes pierced in the lid. To the upper part of these plates are fixed steel springs, H, H, which are capable of bending and so passing through the lid, and then resuming their former position; they support the weight of the zinc by bearing on the surface of the lid. In the small model, a conducting wire is fixed to the zinc, in the other models, metal clamps, 0, 0, 0, are placed on each of the zincs and connected by the negative conductor. Moreover, the zincs are kept at the required distance from the positive electrode by ebonite insulators, I, I, fixed on the supports of the oxide of copper. The ensemble of agglomerate slabs and zincs is encircled by an India-rubber band, K.

The zincs should be kept completely submerged on account of their liability to split if exposed above the surface of the liquid. Their supports are generally made of amalgamated copper or brass. Tinned iron can also be employed, being both more resistant and cheaper. In fact the tin is not attacked and does not give rise with the zinc to a disengagement of hydrogen. The electrodes are not submerged as far as the bottom of the vessels; the zincate of potash which is

778

The final resistance rarely amounting to more than double its initial value. Practically, the resistances certainly fall below the above figures, but a slight polarisation takes place.

When the work is intermittent we can obtain from the various elements with sufficient constancy, outputs of 4, 12 and 25 ampères.

The quantity of electricity available at any given moment, even several months after mounting, is according to the models 75, 300 and 600 ampère hours. Reckoning 6 of an effective volt per element, we should obtain respectively, before the solution of potash and the cakes of oxide of copper are practically exhausted 45, 180 and 360 watt hours.

These new models of the oxide of copper battery seem capable of rendering important service in the various applications in which it is essential that the current should be powerful, constant, and always ready to act.

formed during the working of the battery, separates pretty readily, on account of its great density, settling in a layer at the bottom of the vessel. The electrodes being brought near the surface, thus remain as long as possible in the solution of potash which is to attack them.

The general properties of the oxide of copper battery are well known. It is up to the present the only primary element mounted permanently which is capable of a large output and which only consumes its products in proportion to the work furnished.

The employment of slabs with a metallised surface causes a very large depolarising surface to act from the commencement, thereby increasing the constancy of the elements. The proximity of the surfaces diminishes the internal resistance. The outputs are consequently greater, and the constancy more complete for elements of corresponding dimensions in the new models than in the old ones.

The curves shown in figs. 5 and 6 represent the discharges of the small (I.) medium (II.) and large (III.) models respectively, over resistances of, ths and th of an ohm. The intensities are at first 1:18, 3:25 and 64 ampères. We can gather from these curves that the intensity does not, on an average, lose more than from 2 to 3 thousandths of its value per hour, when the battery is in action continuously for three whole days. If the discharge took place at successive intervals and not continuously, the constancy would evidently be still greater. If we attribute the variation in intensity during the discharge solely to the increase in the internal resistance, admitting that the electromotive force of

A STUDY OF AN OPEN COIL ARC DYNAMO.*

BY MILTON E. THOMPSON.

Of all the dynamo machines in use at the present day, perhaps the internal action of none is so little understood as that of the arc lighting machines of the open-coil armature class. Much concerning the regulation and general behaviour of these machines seems utterly at variance with what one would naturally expect from a superficial examination of the design and construction, and it was with the idea of throwing some light on this seeming mystery that the investigation to be described was undertaken. In carrying out this work no attempt has been made to determine the efficiency of the dynamo, or to determine an answer to the question, Is the machine well suited for arc lighting? The vast numbers of these dynamos in daily use in all parts of the world is a practical answer which must carry more weight than any which might be suggested by a theoretical study of the machine. The experimental work of this investigation was carried out under the direction of Prof. Nichols in the laboratories of Cornell University, during the past winter, by the writer, with the assistance of Mr. H. J. Hotchkiss and Mr. E. L. Morley, and due credit must be given to these gentlemen for any points of interest which may have been brought out by the investigation.

The object in view was to determine accurately, by taking * A paper read before the American Institute of Electrical Engineers, May 21st, 1891.

774

instantaneous values throughout the revolution, the character of the external and internal currents and electromotive forces.

The capacity of the dynamo in question was about 10 arc lamps of 1,200 C.P.

In order to carry out this work it was necessary to build a special commutating device of a character somewhat similar to a number of others before used in the Cornell laboratories for taking instantaneous values of current and potential. Prof. Ryan was the first to introduce this method at Cornell, which is a modification of that described by Joubert in 1881, and the device he used was described at length in a paper read before the Institutet on December 17th, 1889. The device consisted essentially of a metal. tongue mounted upon the shaft of the dynamo and revolving with it, and making contact once each revolution with a small metal spring mounted upon an arm which was so arranged that it could be revolved about an axis coincident with that of the armature of the dynamo and secured in any desired position. By properly adjusting the position of the arm carrying this spring, contact could be made to take place between the spring and the tongue at any desired point of the entire revolution of the armature.

Three sets of curves of exterior current and potential were taken with three different loads upon the machine, and three sets of curves of current and potential through a single coil were taken with the same loads. In order to keep the electromotive force as nearly constant as possible, and prevent variations while a set of observations were being taken, the machine was loaded with a German silver resistance. The loads consisted of about 20, 50 and 80 ohms, respectively, and were equivalent to about 3, 6 and 10 arc lamps. The current for all loads was supposed to be practically constant, and an ammeter placed in the circuit showed but little variation from the normal current of 6-8 ampères. A condenser and ballistic galvanometer were used for making the measurements, and fig. 1 shows a diagram of the connection.

of the dynamo and also of the measuring apparatus. Referring to the diagram, A is the armature, F, F, the field coils, м the regulator magnet, c the controller magnet, and r the carbon resistance used to prevent excessive sparking at the contacts K; P is the positive terminal of the machine, and N the negative, R is the German silver resistance which constituted the load for the dynamo, and i is a resistance of about one ohm placed in the circuit to serve as a basis for the current measurements. L, L, and s, s, are the brushes of the dynamo, which bear against the three segments of the commutator, as shown in the diagram; D is a condenser, K a discharging key, and J a small spring switch; G is a ballistic galvanometer, through which the condenser was dis

Joubert; Etudes sur les Machines Magneto-Electriques; Annales de l'Ecole Normale-Superieur, 10, 46 pp., 1881. Transactions, vol. vii., p. 3.

[JUNE 19, 1891.

charged, and T is a shunt box for regulating the throw of the galvanometer and bringing it within proper limits. From the negative terminal of the machine, x, runs a wire to the commutating device, c, s, and from that to the switch, J. From the positive terminal, P, and the binding post, B, run wires to ƒ and g of the switch f, g, h, as shown. From ཚྭ of this switch runs a wire to the discharging key, K.

Before describing the methods of taking the measurements, it may be well to review briefly the operation of the dynamo and its regulation so far as is apparent from an outside view. The current, leaving the armature through the brushes, Land s, passes through the left field coil, and thence through the regulator magnet and controller magnets to the positive terminal, P, of the machine. Passing around through the external circuit, in this case the resistance, R, the current returns to the machine by the negative terminal, x, passes through the right field coil, F, and through the brushes, Land s, to the armature, thus completing the circuit. The regula tion of the machine is effected as follows:-The controller magnet, C, is so adjusted, by means of a spring, that a current of 6.8 ampères will just raise the cords of the magnets and separate the contacts, k. This permits the current to flow through the regulator magnet, M, and raises the regulator arm, thereby shifting the leading brushes, Land L', forward, and the following brushes, s and s', backward, and thus reducing the potential furnished by the machine. The current is, of course, thereby diminished, and the cores of the controller magnet drop, closing the contacts, k, and thus short-circuiting and cutting out the regulator magne.. The regulator arm then falls, under the action of gravity, and the leading brushes being moved backward, and the fol lowing brushes forward, the potential is raised until the cur rent increases sufficiently to once more open the contacts, k and again cut in the magnet of the regulator. The con troller magnet acts simply as a relay, and is continually cutting the regulator in and out of the circuit, and in thi way the current is maintained constant, regardless of the varying conditions of the external circuit.

During these experiments, the regulator was allowed to ac just as in ordinary practice, but each load was constant, and practically non-inductive. Three sets of curves of externa current and potential were taken, and the measurements wer made in the following manner: The commutating device, c, being secured in a fixed position, ƒ and g were connected The discharging key, K, being then depressed, switch, J. wa then closed, and in a few seconds the condenser becam charged with the difference of potential existing between th terminals of the dynamo at the particular point of the revo lution of the armature represented by the position of th commutating device. Switch J being then opened, the dis charging key, K, was allowed to rise, and the condenser wa thus discharged through the galvanometer, G, causing a de flection, which was duly noted. The connections of th switch, f, g, h, were then changed, so that g and wer joined, and in this way the condenser was charged with th difference of potential existing between the points Band: of the external circuit at the same point of the revolution a the first reading. The condenser was then discharge through the galvanometer, and the deflection noted a before.

After moving the contact spring, s, forward 10 degrees the measurements just described were repeated, and so o until the contact had been moved entirely around the circle

The condenser and galvanometer being properly calibrated it was an easy matter to calculate from the data obtained a above, the difference of potential between the terminals, P an N, and between B and N, at 36 points of the revolution of th armature. The resistance between B and N being known, th current was readily determined from the fall of potential. Th results of these measurements are given in Table I., and ar shown graphically in fig. 3, and will be referred to later.

After taking curves of the exterior potential and current the apparatus was arranged for measuring the potential an current in a single coil of the armature. Coil No. 2 wa selected as likely to present results nearest the average of a three of the coils, and the apparatus was arranged as show in fig. 2. The circuit of the dynamo and its loads wer exactly as for the first measurements, and the connections the measuring instruments were as shown in the diagran Referring to fig. 2, a is a small brush which bears against ring, d, secured to the commutator, but carefully insulato

from it. This ring is connected to the commutator bar of coil No. 2, and thus gives metallic connection with this bar at all points of the revolution. A small brush, b, bears against a similar ring wound round the armature shaft, but insulated from it. This last ring is connected to the three inner ends of the three coils, and thus a continuous contact is made with these ends. The outer end of coil No. 2 was disconnected from the commutator bar, and a small noninductive resistance, e, was inserted between the end of the

represents the difference of potential at the terminals of the dynamo throughout the revolution under a light load, and curve IV. is the corresponding current. Curve II. is the potential for medium load, and curve v. the corresponding current, while curves III. and VI. represent the potential and current at full load. It will be noticed that the curves for medium and full loads are practically alike, while the curve for light load differs slightly in its character and is less symmetrical, and its minimum points occur slightly later than in the other curves. The curves of current and potential are practically alike, as might be expected since the self-induction of the external circuit was little or nothing. The curves are all of them about such as would result from a closed coil armature with a six-part commutator. It may be well to remark that in taking the observations for these curves closer readings than are shown were taken near the minimum points, so as to be sure these points were correctly placed.

These curves show that the external current is continuous and much steadier and more uniform than is usually supposed.

Referring to fig. 4, we find that the state of affairs in a

K

Fig. 2.

coil and the bar. The junction of the coil and the resistance, e, was connected to the tongue, c, which revolved with the armature shaft. Once during each revolution the tongue, e, made contact with the small spring, s, from which a wire ran to the switch, J, and thence to one terminal of the condenser, D. Wires ran from brushes, a and b, to the switch, f, g, h, and from this switch a wire ran to the discharging key, K, as shown on the diagram. When g and h were connected, and keys J and K depressed, the condenser was charged with the difference of potential existing between the ends of coil 2 as the particular point of the revolution represented by the position of the contact spring, s. The condenser being then discharged through the galvanometer, a measure of this difference of potential was thus obtained. Switch f, g, h was then changed so that g and h were connected, and measurements then taken with the condenser and galvanometer gave the difference of potential between the ends of the non-inductive resistance, e, and from this the current in coil 2 was calculated. Readings were taken in this way at distances of five degrees apart all around the revolution, and the results are given in Table II., and are shown plotted out in figs. 4, 5 and 6, three curves at three

100

current curve is given in fig. 6. It is probable, though, from the similarity of the current curves at light load and at medium load, that the current for full load differed very much from the other two.

It will be noticed that at all loads the potential changes sign no less than six times during the revolution, and falls to zero, or almost to it, as often as twelve times, while the current reverses but twice during the revolution, and for the most part goes through very gradual changes. At the bottoms of the figures are lines of spaces showing the positions of the armature during the part of the revolution represented by the curves above them. 81, 82, 83, are the periods of short circuit, and P1, P2, P3, &c., are the periods during which two coils are usually supposed to be in parallel with each other and in series with the third coil. There

[JUNE 19, 1891.

represent the second short circuit, 82, and figs. 13, 17 and 21 show the second periods of parallelism, P, or 220 degrees. In all of these figures the relative positions of the coils and poles of the dynamo are also shown, the coils being nambered 1, 2, 3, just as they are on the dynamo. During the entire period represented by each four figures as above, the armature has turned through 120 degrees, or one-third of the revolution; and if we now consider the position of coil? changed to that of coil 1 in each figure, our figures will then represent the next third of the revolution. If coil 3 is sup

are six periods of each kind in the course of each revolution. Figs. 10, 14, and 18 represent the relative positions of commutator and brushes at the three different loads during the parts of revolution shown at s1. For the light load this period is about 25 degrees, for the medium load about 10 degrees, and for full load not more than 1 degree. The periods shown are not the actual, but the apparent periods of short circuit, as it would seem to be from the positions of the commutator and brushes. The actual periods of short circuit are somewhat longer, owing to the conductivity of the spark, and last until the spark is blown out. The exact point of the revolution represented by figs. 10, 14 and 18 is the point 130 degrees, as marked below the curves. Dios. 11, 15 and 19 represent the positions of the brushes commutator for the parallel periods, P1, or to give the position shown, 160 degrees. Figs. 12, 16 and 20

Fig. 18.

Fig, 19,

the ends of the coil and not the E.M.F. developed in the coil during the revolution. The sudden and marked changes in current which are shown at certain points of the current curves are selected from a number of readings which showed great variation. In some cases the reading had an opposite value from that shown in the curve, and it seems probable that, all things, considered, the average value of these points should be very little, if any, different from the others, and the whole should form a smooth curve without any very sudden changes. In figs. 10 to 21 the arrow shows the direction of the current in each coil during that particular part of the revolution represented by the figure. In each of these figures a neutral line, n n', is shown, in passing which the current in each coil becomes zero and reverses. It will be noticed that the direction of the currents in the various coils for the so-called periods of parallelism and short circuit is quite different from what would be naturally expected and from what has been supposed to be the case in most of the descriptions of the action of this machine. For instance, in fig. 11 we have the current in coil 1 flowing toward the junction of the three wires, showing that a current is flowing from the junction through coil 3 to the left pair of brushes and back through coil 1 to the junction again, or, in other words, that coils 1 and 3 are short circuited upon each other. This state of affairs continues, with the light load, until commutator bar 3 touches the right following brush Thus it will be and short circuit period, s,, commences. seen that there is no instant during the so-called parallel period when coils 1 and 3 are in parallel and that the actual parallelism does not occur until after the beginning of the period of short circuit. In the case of medium and full loads, however, it occurs at from 10° to 20° before the period short circuit. Another point worthy of notice is, that at the first instant of short circuit, when the commutator bar just touches the following brush, there is no sudden increase of current, as would be naturally expected, but the current in the short circuited coils slowly and gradually increases until the short circuit is broken by the blowing out of the spark, at which instant the current is a maximum in the coil whose commutator bar is opposite the sparking point, and in the other two coils it is about the same as the normal external current.

In figs. 7, 8 and 9 are shown curves of potential at the These curves were brushes, at the three different loads.

TABLE II.-INTERNAL CURVES.

777

TABLE I.-EXTERNAL CURVES.

8.3

579

107

60

0

241

13

407

247

12.7

380

199

12:4

285

140

11-7

167

8-22

113

349

2.7

316

516

5.9

110

23

1.61

61

6.7

7.0

6.6

7:0

7.2

561

6.8

obtained from the potential curves of figs. 4, 5 and 6, by adding together the potentials of the three coils according to their relative positions, and thus obtaining the potential at the brushes. It will be noticed that at all loads the curves fall below zero six times during each revolution, and that the difference in area and consequent effective E.M.F. is obtained more by the widening out of the curves than by increase in height. This is what we ought to expect from the change in duration of the short circuit periods. A comparison of these three curves of potential at the brushes with the three curves of potential at the terminals of the dynamo will give one a good idea of the remarkable steadying power which magnets such as the fields of this dynamo No further explanaexert upon an intermittent current. tion is needed of the fact that this dynamo will not work