respectively connected to the dynamo and to the 'bus bars, or by connecting a single volt meter first to one and then to the other, which avoids the error due to a difference between two instruments. A still better plan is to connect the dynamo to the 'bus bars through a high resistance and a galvanometer which deflects one way or the other according to whether the dynamo voltage is higher or lower than that of the circuit. For this purpose it is very convenient to use a volt meter having a scale on both sides of the zero point. After the pressure of the dynamo has been properly regulated, the three switches, D, E, and F, are closed. When this is done simultaneously with a three-pole switch, a considerable current will flow through the series coil B, which tends to still further increase the voltage of this dynamo, at the same time taking current away from the series coils of the other machines, and thereby reducing their potential. The shifting of load thus produced may be so sudden and so great as to be objectionable. To avoid this difficulty the two switches E and F are sometimes combined to form a double-pole switch, the other one, D, being operated independently. With this arrangement the double-pole switch EF is closed first, allowing the current to flow through the series coil B, and the regulation of voltage is made under these conditions. The switch D is then closed, and the E.M.F. of that machine will not change materially. The current which it generates will also be small, provided its voltage was adjusted to be only slightly greater than that of the 'bus bars.

When the three switches, D, E, F, are simultaneously closed, it is found in practice that armature reaction, etc., tend to lower the potential of the generator about as much as the current in the series coil tends to raise it, hence the effects counteract each other. But it is merely an accident if such is the case, and it can only be determined by trial. It often happens that the two actions do not balance each other, the rise of E.M.F. being greater than the fall. In these cases, which are common in electric railway stations, the attendants learn by experience that the pressure of a dynamo should be regulated a certain number of volts below that of the 'bus bars before it is connected to them, in order that it shall act properly when the three-pole switch is closed. This is certainly a crude method of working, and increases the chance of having a back current flow through the series coil, which would tend to

demagnetize the field if the E.M.F. of the dynamo is considerably less than the pressure at the 'bus bars, particularly when the equalizer is somewhat long or is too small in cross section.

The use

It would seem to be generally desirable to separate the switch D in order to have independent control of the equalizer. Another advantage secured by this arrangement is the field excitation that is positively produced when the switch E F is closed, avoiding the delay and uncertainty which are always involved when selfexcitation alone is depended upon. In fact, self-exciting dynamos often fail to generate, or become reversed in polarity.* of separate switches also enables the series coil to be left in circuit when a dynamo is not working, for the reasons explained on page 53. The three switches D, E, and F might all be made independent; but there would then be a chance for D and F to be closed, and the equalizer switch E left open, which is likely to cause serious trouble, due to an excessive or reversed current in the series coil B; or the switches D and E might happen to be closed with F open, in which event the series coil would not be in circuit, and the dynamo could not generate sufficient voltage when the load increased.

Compound or over-compound generators are generally used in isolated plants and smaller central stations, and are almost universally employed in electric railway power-houses; but in large electric-lighting stations plain shunt dynamos are often employed in order to give greater flexibility of regulation. In such systems the lamps and other devices are supplied through a number of feeders, which are fed with different pressures at the station according to their length and the load upon them. The methods employed will be described later under the head of "Feeder Regulation." It should also be noted in this connection that the business of large stations warrants the constant employment of one or more men to regulate the voltage, while in small plants the regulation should be automatic as far as possible, in order to reduce the required attendance to a mimimum. It is not unusual for such plants to be left to take care of themselves for considerable periods of time. In most cases automatic regulation has to be supplemented more or less by hand adjustment of the field rheostat to make up for change in speed due to variations in * Vol. i., p. 362.

steam pressure or water pressure in the case of hydraulic power. The heating of the field coils and resistances in shunt and compound dynamos, as well as hysteresis in the magnets, also cause variations in the voltage which usually have to be overcome by hand regulation.

Automatic Constant-Potential Regulators. To entirely avoid the necessity for hand regulation, or to reduce it to a minimum,

automatic devices may be used to maintain constant potential. One of these regulators, illustrated in Fig. 31,* consists essentially of a rheostat in the shunt field circuit, whose moving arm, N, is operated by a solenoid, Q. A relay solenoid, A, is connected across the main conductors through the binding-posts ff', and has contact points, i, that govern the admission of current to the working solenoid Q. The latter is differentially wound with four coils, two of which have a small continuous current flowing through them, and are in opposition to each other, the current being sup

* Electrical World, N.Y., March 20, 1897, p. 395.

plied through the switch G, and binding-posts ff. The other two coils, when the circuit through them is closed by the relay, act to neutralize one of the continuously excited coils. This method of operation avoids the injurious sparking that would occur if the circuit of the main solenoid were actually broken, and thus permits a close adjustment of the relay contact points, and secures a more sensitive regulation. The cores of the solenoid are composed of small, soft iron wires to give quick action and reduce hysteresis effects. Two lamps, F F', constitute a non-inductive resistance for the relay circuit. An ordinary hand rheostat, E, is included in the shunt field circuit in order to adjust the resistance, and also give independent control. The voltage for which the device is set may be altered by shifting the small weight J on the relay lever. In this device the consumption of energy is not large, being only 60 or 70 watts for a large regulator. The construction as described would tend to maintain a constant potential at the points on the circuit to which the binding-posts ƒƒ' are connected. It is evident that these may be in the station or at any desired position on the system of conductors, thus producing the effect of overcompound winding. This would necessitate the running of special pressure wires to some distance, which may be avoided by providing the relay solenoid, A, with a series coil in addition to the shunt coil, or, in short, by compounding the regulator instead of the field magnets themselves. The series coil would have to be connected

in the main circuit, or shunted around a resistance placed in it.

In order to maintain a constant current for charging storage batteries and for other purposes, the relay A is wound with a series. coil only. These devices are also made for alternating current regulation. A later form of the Chapman regulator is described in the Electrical World, April 16, 1898, p. 480.

Methods of Exciting the Shunt Coils of Dynamos. The three principal ways of exciting the shunt field coils of either a plain shunt or compound generator are known as Self excitation, 'Bus excitation, and Separate excitation.

If self-excited, the terminals of the shunt field coils are connected to the brushes, hence the magnetism gradually dies away as the dynamo is slowed down after being disconnected from the circuit; and when it stops the field has disappeared, except a little residual magnetism. This avoids the danger of piercing the insulation of

the field or armature winding, which is likely to occur if the magnetism be suddenly discharged. The disadvantages of this method are the slowness with which the dynamo builds up its own field, and the possibility that it may entirely fail to generate, or become reversed in polarity. In 'bus excitation the shunt field coils are fed from the station mains or 'bus bars, the advantages being the fact that the field magnetism is promptly and positively brought to full strength so that the dynamo may be connected to the others as soon as it attains its speed, and the polarity cannot become reversed. On the other hand, the field may be accidentally left in circuit after the dynamo is stopped, and it is always necessary to discharge it through a bank of lamps or other resistance by means of a field break switch. This method of discharge, although safe, might cause trouble if the bank of lamps were disconnected, or otherwise out of order.

With separate excitation the field circuits of the generators are connected to a special dynamo or other source of current. This plan has all the advantages and disadvantages of 'bus excitation, and has the additional merit that the field strength is not affected by changes in voltage occurring on the main circuit, which would tend to aggravate the variations. It also has the difficulty that the exciting dynamo or battery is comparatively small, and may therefore be weak and unreliable. An accident to it would incapacitate as many generators as were being charged by it. In a three-wire system the shunt field circuits of the dynamos on one side may be supplied with current from the other side of the system, which practically amounts to separate excitation, except that an extra generator is not required. But if any trouble occurs on one side it is likely to affect both sides of the system.

In the method of excitation devised by Mr. W. I. Donshea,* and represented in Fig. 32, one terminal of the shunt field winding S leads to one of the brushes C, while the other end is brought through the regulating rheostat R to the blade of the field switch F. This blade is pivoted on the same axis as the main dynamo switch M, but the two are not electrically connected together. The other main switch, N, and field switch, F, are first closed, M being left open. This completes the field circuit so that it is excited from the 'bus bars B+ and B as shown. When the

*Electrical Engineer, N.Y., Jan. 22, 1896.