GB2205450A - Rotary electro-dynamic machine - Google Patents

Abstract

A salient pole dynamo-electric machine operates as a reluctance motor energized by an alternating voltage having a frequency that is a multiple of that corresponding to normal synchronous operation of such a motor, a commutator admitting excitation by complete cycles of the alternating voltage limited to periods during which the poles are coming into register so as to obtain zero winding current when the stator and rotor poles are in registry. An energizing cycle begins when a rotor pole 13 is at A in relation to stator pole position B. This is represented by the back-EMF shown by the lower waveform in Fig. 3, denoted d/dt(LI) and results in a winding current waveform denoted by I. As the poles separate between B and C excitation is suspended, thus excluding one or more cycles of the alternating voltage until the rotor poles reach the A, C positions. This arrangement minimizes a retarding reluctance torque as the rotor and stator poles separate and limits eddy-current loss. In order to employ efficiently a continuous alternating excitation source, arrangements having multiple offset rotors (Fig. 4) and commutation are disclosed. <IMAGE>

Description

ROTARY ELECTRODYEAXIG POWER GENERATOR This invention relates to a specific form of electrodynamic machine which exploits the recently discovered anomalies in the energy exchange processes associated with the cyclical magnetization of ferromagnetic materials under certain conditions. These conditions arise when there is relative motion of two ferromagnetic core elements allowing mechanical power to be extracted by virtue of their direct electromagnetic interaction and where the magnetic condition of the elements is changed for motion in the opposite sense.

It has been traditional practice in electrical engineering science to calculate the driving torques or linear forces in electrodynamic machines on the express assumption that the ferromagnetic substance makes no sustained contribution to the energy input. The energy balance has been restricted to the electrical power fed to windings, any mechanical energy involved and heat, as from eddy currents, hysteresis and friction. This has been the standard textbook practice for continuously operating machines, and as a result a technological field of vital significance has been neglected.

The particular circumstances in which this unfortunate situation arises are not predominant in the more important machines used at present as our primary power sources. The conventional synchronous alternator used in power house equipment develops a rotating magnetic field which is not seen as a changing field by the rotor, owing to its rotation in synchronism in the same sense. However, there is, even here, some measure of what is termed 'reluctance torque', which is the province of this invention.

In order to understand the nature of the effects under consideration one should imagine a magnetic core comprising a movable core element, so arranged that there is a variable gap in the magnetic circuit. A conventional textbook calculation (H. H. Woodson and J. R. Xelcher, "Electromechanical Dynamics", Part I, John Wiley, London, p. 85; 1968) of the force acting between the pole faces defining this gap, as based on the energy balance argument, tells us that the force is the rate of decrease of the inductive energy of a winding exciting the magnetic core with respect to the increment of the gap.

The whole calculation is based on electrical energy fed to the winding, energy stored inductively by that winding and energy stored in the air gap, where there is no ferromagnetic substance. The ferromagnetic properties of the magnetic core are only evident in that they cause the inductance to be higher than it would otherwise be. Yet, one would think that somehow the force acting across the gap would be partly due to interaction forces between the microcurrents in the ferromagnetic substance of the fixed and movable parts of the core.

The role of the ferromagnetic material is recognized as acting, in effect, to concentrate the effects of the magnetizing winding as if they are coextensive with the length of the air gap, so that the forces across the poles are those associated with a very strong field effect in the air gap itself. Usually, textbook analysis supposes that the core material has a constant but infinite magnetic permeability and zero magnetic reluctance, which is an idealization not supported in practice.

However, accepting the textbook theory, there is no basis for supposing that the energy balance is at all affected by energy transfer drawing on the intrinsic microcurrent sources which account for ferromagnetism in the magnetic core.

Now, it is a fact of experiment that this theory poses some problems when one considers how the air gap variation affects the measured inductance of the magnetizing winding. For constant alternating voltage excitation the increase in the air gap from zero should result in a linearly-related increase in magnetizing current, corresponding to a direct inverse relationship between inductance and length of air gap.

Experiment shows that with increasing air gap the current does not rise linearly but falls off from the linear relationship, showing that the inductance is progressively higher than it should be as the gap length and the current increase. A very detailed textbook discussion of this is given by E. B. Xoullin in "The Principles of Electromagnetism", Clarendn Press, Oxford, pp. 168 - 175 (1955). The extra inductance is attributed to leakage flux, but, as Professor Moullin explains, the leakage appears to be enormous if the facts of experiment arising from a very small air gap are to make sense. Evenso, he struggles with this theme to conclude that leakage flux cannot be calculated and that one must build a magnet and test it to measure its value.He also concludes with the remark that he is not able to forecast any relation between leakage inductance and the size of the magnet.

the alternative explanation, not seemingly presented in the textbooks, is that the theory for the magnetic flux in the air gap has underestimated the true flux, because it has ignored a contribution from the coupling effects between the fixed and movable parts of the iron core. This would increase the force acting across the air gap and cause work to be done by the ferromagnetic substance, work not sourced in the energy supplied to the magnetizing winding, even though the current in that winding has acted as the controlling trigger allowing this release of energy. All this really means is that the ferromagnetic action, seated as it is in its own solenoidal microcurrent system, makes its own independent contribution to the flux traversing the air gap in addition to its role of concentrating the applied magnetic field into that zone.

In the experimental data provided by Xoullin it is clear that two 7 mm gaps in an iron core magnetic circuit of 1 metre could cause the inductance to be twice that expected from the idealized theory. If this were all attributed to increased flux it would mean a four-fold increase in the mechanical force acting across the air gap, compared with that expected from the idealized theory. Unfortunately, Noullin gave no data to suggest that he checked this by measuring that force in relation to the inductance. Hence, the assumption that leakage accounts for all of the enhanced inductance was adopted and the pathways which have now led to this invention were missed.

Now imagine that we extract energy from the mechanical displacement as the gap closes with the magnetizing winding energized by current. Some of this energy will not come from the electric circuit feeding the winding. It is sourced in whatever microscopic electron currents account for the ordering that gives the polarization property to the ferromagnetic substance. Whatever sustains the quantized state of motion of electrons in atoms is caused by the natural ferromagnetic tendency of the substance to feed out some energy to assert the mechanical forces that have been harnessed to extract energy.

Naturally, one expects this amount of energy to be drawn back, should one try to open the gap by force with the magnetizing current still present. In effect, this oscillation of energy between the ferromagnetic substance and the rotor is occurring continuously in the operation of salient pole machines. Therefore, the conventional design criteria hold valid for practical purposes. However, suppose instead that the current is switched off whilst the gap is closed. Then there can be no sustained restraint on the resetting of the system as the gap is reopened. The energy balance now indicates that a significant amount of energy has been extracted from the quantum sources in the ferromagnet and applied in a useful way. In effect, it is as if the inductive energy fed to the magnetizing winding has merely served to power the catalytic action of the magnetic system, which operates as a kind of magnetic 'heat pump' allowing energy to transfer from one state to another.

This invention exploits this energy source in a rotary power generator and in such a way that the attendant magnetization losses which accompany the rapid flux transitions are reduced substantially.

Active research involving motor-generator devices which may involve the above operating principles concerns, in the main, devices using permanent magnets, either in the stator or rotor or both. There have, in fact, been several claims that machines have been designed which can operate by drawing indefinitely on the energy available from the ferromagnetic ordering within the permanent magnet. The Howard R.

Johnson U.S. Patent Specification 4,151,431 dated April 24, 1979 discloses a linear motor and a related rotary configuration using permanent magnets on the relatively moving elements. It is understood that the linear version of this machine was demonstrated to a U.S.

Patent Appeals Board to support the case for patent grant. However, the patent as published shows that in the linear version the stator permanent magnets are located progressively closer together in the direction of movement of the interacting magnet. Such a configuration, if essential to the operation of the device, could not be applied in a machine intended for continuous rotation. However, the inventor asserts in the patent specification that the action arises from the effective superconducting characteristics of the permanent magnets, presumably meaning the action that sustains the polarizing microcurrents at the atomic level. The patent specification is, therefore, relevant in that it implies the expectation of extracting energy from this source.From the earlier analysis it has to be expected that, by manipulating permanent magnets so as to effectively control the magnetic flux within other permanent magnets, one should be able to extract such energy, provided, of course there is some asymmetry feature in the operation to assure that the energy extracted is not returned on the reverse part of a cycle of operation.

This invention is firmly based upon the clear appreciation of the essential nature of this asymmetry in the magnetic excitation.

Consideration then shows that the successful design of robust and powerful generators exploiting these principles, favours the control flexibility available from current-excited electromagnets rather than permanent magnets. Furthermore, there are special advantages in matching the fluctuating magnetic reluctance to the variation of the waveform of the excitation signals. Also, there are commutation advantages aimed at eliminating unnecessary heating, both in the magnetic circuit and in the windings used to excite the electromagnets.

What is to be described is essentially a new kind of reluctance motor.

A relevant description of the operation of the reluctance motor is given by H. H. Skilling in his book "Electromechanics", John Wiley, London, pp. 76 - 78 (1962). Skilling describes its operation using a rectangular magnetizing current waveform which is switched on midway between-the position in which stator and rotor poles are in register and off again when they are in register. In principle, therefore, such a machine should tap the 'free' energy source of the ferromagnetic substance. However, owing to the very rapid rates of flux change resulting from the rectangular pulsed operation and the high current applied in sustaining the magnetic action the losses in such machine would detract from any such 'free' energy gain.

A more practical version of the reluctance motor is then described by Skilling, using a simple alternating current waveform to drive the motor synchronously. Though this is less subject to eddy-current losses, it does, however, offer only marginal benefits in exploiting the principles on which this invention is founded. The reason is that the torque is a driving torque for half the time and a somewhat weaker restraining torque for the other half of the time, meaning that full advantage of the reluctance motor principle' is traded for the simplicity of the power supply and the reduced loss. Skilling further explains that though such motors are used in clocks, for example, the reluctance torque is a factor in large machines, particularly if they have poles which are in any sense salient.Thus he says that perhaps 10 to 30 per cent of the total power of even very large machines is due to the reluctance torque and that the reluctance power of a large machine may be in the tens of thousands of horsepower.

Skilling's book discusses at length the energy balance in such machines and calculates the reluctance torque using principles based on those discussed above for the simple air gap, but the prospect that some energy might be sourced in the ferromagnetic substance itself is not contemplated.

Skilling does, however, discuss the subject from two aspects. Firstly, on pp. 72 - 74, he discusses the energy balance in a machine operating at constant magnetic flux. This leads two a formula for the~machine torque which involves no change in the inductive electrical energy input. The reason is that the magnetic flux is not changing and so there are no induced EKFs interacting with the magnetizing current. The energy exchange is between the magnetic field energy stored in the air gap and the mechanical work involved. Such a machine, if perfect in design, would not draw on the ferromagnetic energy source. Hence the torque formula is valid. Secondly, Skilling, on pp. 74-76, makes the same analysis for constant magnetizing current excitation, and he obtains the same torque formula.This time, however, the energy stored in the field is greater because the current is sustained and there is an inductive electrical power input from the action of maintaining the current in the magnetizing winding against the effects of back ENF.

This latter analysis fails to take into account the fact that those back ENFs act also against the microcurrents polarizing the ferromagnetic core. There is therefore additional energy of the ferromagnetic core to consider in the balance. This increases the magnitude of the torque, whether driving torque or restraining torque, in relation to that given in the textbook.

On p. 72 of his book Skilling has expressly said that he wondered which method of excitation would be best. The constant flux case applied to a very rapid machine rotation and very low resistance magnetizing circuit, whereas the constant current case applied to slow rotor speed and a high resistance circuit. His analysis of the two cases led him to conclude that the torque formula was the same in both cases, but it seems that he has erroneously overlooked the possibility that in the latter case the torque can be much higher. This is simply because the 'free' energy sourced in the ferromagnetic microcurrent sources has become involved in the action.

It cannot be expected, however, that any significant net gain of 'free' energy from the ferromagnetic substance would occur in the standard electrodynamic machine. Firstly, the effect only arises in machines that involve varying flux conditions, but, secondly, the main reason is that, in spite of the reluctance torque being present, it is not normal in such machines to switch the magnetization off when the poles are in register. Usually, the magnetization is regulated by a sinusoidal waveform which peaks close to such positions, there being some phase lag, but the magnetizing current nevertheless can be sufficient to assure the ferromagnetic ordering of the atomic microcurrents without any significant phase slippage with respect to the poles.This means that the normal reluctance torque will not be accompanied by an action using the principles of this invention, other than one that asserts both drive and retarding torques which tend to balance. It is only where the switching of the magnetizing current assures loss of ordering or reversal of the polarization in the ferromagnetic core in a manner that is asymmetric phase-wise with respect to the positions in which the poles are in register that any 'free' energy release can occur.

This invention addresses this special situation.

According to the invention, a dynamo-electric machine comprises a ferromagnetic rotor and a ferromagn,etic stator, both having salient poles which come into register cyclically at successive angular positions of the rotor, drive means coupled to the rotor for supplying output power from the machine, means comprising an energizing winding for controlling the magnetization of the poles, a source for an energizing electrical supply providing the magnetizing current, characterized in that its alternating voltage waveform changes polarity more than once during the time taken for the rotor to move between two adjacent positions in which the stator and rotor poles are in register, and commutator means sensitive to the angular position of the rotor for regulating the periods during which the magnetizing current is supplied to the winding, these commutator means being operative to confine the energization of the winding substantially to periods in which there is only one polarity reversal of the alternating voltage waveform and corresponding to limited ranges of angular position of the rotor fully in advance of those in which the rotor poles have come into register with the stator poles.

Features of the invention will become evident from the following description and claims, but a principal feature involves energizing the magnetizing winding with an alternating pulse of one complete cycle duration immediately prior to the poles coming fully into register. The object of this is to magnetize the system only over a range over which the poles are closing and preferably only over a near range of approach, but to secure demagnetization preferably spread over a range just before the poles are in register at their closest position. In this way the machine is energized only for very limited periods during the rotor revolutions. This limits the heating losses due to current supplied to the machine when its inductance is low and assures that the forces between the stator and rotor poles act to develop torque rather than essentially radial stresses.It also reduces the rate of flux change as the magnetizing current diminishes, by spreading it over a range substantially commensurate with the position in which the pole faces are beginning to overlap to the position in which they are fully in register.

It will be appreciated that, in causing the magnetization to be switched off in the short period corresponding to rotor travel through an angle of approximately half that taken up by a pole, the eddy-current losses will be very much smaller than they would be for a more rapid switch-off at the exact position of radial alignment of the poles.

The following description of the invention makes reference to sev eral figures.

Fig. 1 shows schematically a salient pole configuration with the stator and rotor poles in register.

Fig. 2 shows the same configuration with the rotor poles midway between in-register positions with the stator poles.

Fig. 3 shows the electrical waveforms applicable to the excitation of the poles in a preferred embodiment according to the invention.

Fig. 4 depicts schematically the pole configuration on a rotor comprising three sets of four poles axially spaced but angularly staggered.

Fig. 5 is a schematic of a dynamo-electric machine comprising two rotor pole configurations of the form shown in Fig. 4.

Referring to Fig. 1 a dynamo-electric machine comprises a stator 10 having four salient poles disposed around a rotor 11 having also four salient poles. Both stator and rotor comprise ferromagnetic laminations separated by insulation with their planes perpendicular to the axis about which the rotor rotates. Four series-connected windings such as 12 are mounted on the stator poles to provide excitation around four magnetic circuits including the rotor poles, there being one such circuit in each quadrant of the system core configuration shown.

When the rotor has an angular position midway between adjacent positions in which the stator and rotor poles are in register, as depicted in Fig; 2, excitation of the core by energizing the magnetizing windings causes the rotor poles and stator poles to be mutually attracted. This causes the rotor to turn to a position in which the poles are in register, as shown in Fig. 1.

By de-energizing the magnetizing windings when the rotor comes to this position, the inertia of the rotor causes it to move freely to the next position corresponding to Fig. 2. The removal of the magnetizing field has removed the powerful force of attraction between the rotor and stator poles. Upon reaching the new position corresponding to Fig. 2, re-energization drives the rotor through another quarter revolution, and so on. This, essentially, is the principle of operation of the reluctance motor.

Fig. 3 illustrates the advance afforded by this invention. We suppose that the waveforms illustrate how certain electrical quantities vary as the poles move from a position shown in Fig. 2, through the position shown in Fig. 1 and on to the position shown in Fig. 2 again. This motion corresponds to advance of the rotor pole at A, denoted 13, to a position B in register with the stator pole, denoted 14, to a position C. The corresponding positions A, B and C are also shown in Fig. 2.

At time corresponding to position A, the applied voltage waveform of the signal used to energize the windings 12 changes from its steady value of zero to oscillate once only through a complete cycle, as shown in the lowermost part of the figure, so as to complete the cycle at position B.

This waveform is denoted as the time (t) rate of change of the product of L and I, where L denotes the inductance of the four series-connected windings and I is the magnetizing current carried by the windings. For the purpose of illustration the resistance of these windings is deemed to be negligible. Hence the applied voltage will equal the back EMF represented by this rate of change quantity.

Integration of this voltage signal gives the waveform of LI, as shown.

Also, the inductance L is shown to vary symmetrically about the central position B, being highest when the poles are in register and dropping off rapidly as the poles move out of register. From these data one cansee that the current I must have a form which rises rapidly as the poles come together but peaks well before the poles are in register and drops to zero at the position B. Essentially, the rate of change of the flux linking the poles, as measured by the back EXF is sinusoidal over the range from A to B. This EMF corresponds to a portion of an applied voltage which has been gated to admit only one complete cycle but which, in its ungated form, reverses polarity more than once between the adjacent in-register positions of the poles.In this particular case it reverses polarity twice over this range, because the source voltage comprises two complete cycles of oscillation in a period during which the rotor makes a quarter of a revolution.

The advantage of this particular method of excitation is that the eddycurrent losses in the stator and rotor are very much smaller than they would be if the flux changes were caused to be more rapid by sudden switching in the in-register position of the poles. Furthermore, though the maximum polar attraction occurs when the poles are near to registration when the current is not reduced as shown in Fig. 3, such attraction tends to develop forces that are more radial than tangential.

As such they enhance the rotor stresses and produce vibration, much of which is avoided by concentrating the torque into a period a little in advance of the in-register position, as is the case in Fig. 3.

A disadvantage of this method of excitation is that the magnetizing windings do not load the alternating voltage source continuously as the rotor rotates. For this reason and in order to make the driving torque more uniform the rotor pole configuration shown In Fig. 4 applies to one embodiment of the invention. This comprises four axially-spaced sets of poles each in a staggered relationship in the angular sense about the rotor axis. The three rotor sections 20, 21, and 22 comprise ferromagnetic laminations and are spaced apart on the rotor shaft 23 by insulating ring spacers 24 and 25.

Such a rotor would have a stator assembly having a corresponding group of three sets of four poles, which are not staggered, though alternative configurations are possible in which-the rotor poles of the different sets are fully aligned and the stator has the staggered pole sets.

Essentially, each set of poles is excited by its own magnetizing winding circuit. Then, with this three-set construction, the alternating voltage waveform used to energize the windings is gated via an electronic commutator, not shown in detail in the figures but utilizing conventional electronic switching and control techniques, so as to admit successive complete cycles of the waveform to different sets of magnetizing windings.

The machine configuration shown in Fig. 5 incorporates two rotor sections of the form outlined in Fig. 4 disposed on either side of a common rotor shaft shared by a centrally-located synchronous alternator.

The alternator is of conventional design but it has two stator windings, each producing an alternating voltage signal having a frequency twelve times that of the rate of rotation of the rotor. Slip rings 29 provide the means for supplying an excitation current to the rotor.

Considering an operating period as one corresponding to a quarter of a revolution of the rotor, there are three complete cycles of alternating voltage generated by the alternator source. As shown in Fig. 5 an electronic commutator 30, has two inputs 31 and 32 from the alternator 33.

The output from the commutator is fed selectively to magnetizing windings on the stators interacting with the two rotors of the design shown in Fig. 4. Only one such winding 12 is illustrated in the schematic layout shown in Fig. 5. Input 31 from one stator winding is subject to cyclic switching by the commutator which admits first complete cycle of the signal to pass to stator magnetizing winding on the stator pole section 34, then the next complete cycle is routed to the magnetizing winding on the adjacent stator pole section 35 and the next complete cycle energizes the next stator pole section 36.

Meanwhile, the signal from the alternator stator winding supplied along input 32 excites the three stator pole sections 37, 38 and 39 in sequence in the same way. This imposes continuous load on the alternator. However, as can be seen from Fig. 3, the current supplied by this alternating source tends to be unidirectional owing to the changing reluctance of the magnetic circuit. To avoid the problems arising from this the inputs 31 and 32 are arranged to be in anti-phase by so connecting the alternator windings via the commutator 30. Then, after each operating period the inputs from these windings are interchanged as between the stator pole sections on the left hand side and on the right hand side of the machine shown in Fig. 5. The effect of this is to invert the polarity of the lower waveform in Fig. 3 for each successive operating period.In its turn this causes the successive magnetizing current pulses supplied to each magnetizing winding to invert polarity. The result is that there is then no buildup of a bias component affecting the current waveform.

It is to be understood that the commutator is merely a sequential switching device and it could be a mechanical commutator mounted on the rotor shaft and having suitably-positioned brushes connected to the respective windings. The commutator segments would then provide the switching sequence in due relationship to the energization pattern specified by the waveforms in Fig. 3. The alternative is to use an electronic commutator which may be regulated by synchronizing signals derived from a sensor located adjacent to the rotor shaft and sensing its angular position. The design of such controls will be evident to those expert in the electrical and electronic engineering art.

As described, the invention aims to provide a new kind of reluctance motor, which has a self-exciting feature and which, under certain circumstances, can draw on the sustained energy sources feeding the ferromagnetic state in the stator and rotor cores. The object of the invention has been to provide a design which minimizes the eddy-current losses. By incorporating an alternator in the structure shown in Fig.

there is provison for a power output of continuous sinusoidal voltage form. Apart from being suitable for the powering of the specific kind of reluctance motor provided by this invention, this waveform is of the form suitable for delivering any surplus output power. Furthermore, inasmuch as the system shown in Fig. 5 operates both as a motor and as generator, there can be mechanical power supplied or delivered by the rotor shaft to assure energy balance. The invention therefore provides a highly versatile electro-dynamic machine having novel features and suited to the exploitation of the 'free' energy potential of the ferromagnetic materials employed.

Claims (10)

CLAINS

1 A dynamo-electric machine comprising a ferromagnetic rotor and a ferromagnetic stator, both having salient poles which come into register cyclically at successive angular positions of the rotor, drive means coupled to the rotor for supplying output power from the machine, means comprising an energizing winding for controlling the magnetization of the poles, a source for an energizing electrical supply providing the magnetizing current, characterized in that its alternating voltage waveform changes polarity more than once during the time taken for the rotor to move between two adjacent positions in which the stator and rotor poles are in register, and commutator means sensitive to the angular position of the rotor for regulating the periods during which the magnetizing current is supplied to the winding, these commutator means being operative to confine the energization of the winding substantially to periods in which there is only one polarity reversal of the alternating voltage waveform and corresponding to limited ranges of angular position of the rotor fully in advance of those in which the rotor poles have come into register with the stator poles.

2 A dynamo-electric machine according to claim 1, wherein the switching sequence of the commutator means operates to energize the magnetizing winding with an alternating pulse of one complete cycle duration immediately prior to the poles coming fully into register.

3 A dynamo-electric machine according to claim 1, which comprises n sets of stator poles interacting with corresponding n sets of rotor poles, these different sets being axially spaced along the rotor axis, in a relatively staggered configuration so that the poles of different sets come into register at n different equi-spaced angular positions of the rotor, there being also n separate magnetizing windings for energizing the corresponding sets of poles, and wherein the switching sequence of the commutator means operates to cause single complete cycles of the alternating voltage waveform to magnetize each pole set successively in sequence, whereby to provide continuous utilization of the alternating source supplying the magnetizing current.

4 A dynamo-electric machine according to claim 3, wherein the switching sequence of the commutator means operates to reverse the polarity of successive full cycles of the voltage waveform supplied to each magnetizing winding, whereby to cause the direction of the magnetizing current to change for the successive polar energization periods.

5. A dynamo-electric machine according to any preceding claim, wherein the magnetizing windings are mounted on the stator and the commutator means comprise electronic switching means operating under the synchronizing control of position detecting means sensitive to the angular position of the rotor in relation to the stator.

6. A dynamo-electric machine according to any of claims 1 to 4, wherein the magnetizing windings are mounted on the rotor and energized via a mechanical commutator on the rotor shaft.

7. A dynamo-electric machine according to claim 5, wherein the rotor has a laminated ferromagnetic structure with plane of the laminations being perpendicular to the rotor axis.

8. A dynamo-electric machine according to any preceding claim, further comprising additional dynamo-electric means coupled to the rotor and operative to generate the alternating voltage supply source used to energize the magnetizing windings, whereby the machine becomes selfexcited.

9. A dynamo-electric machine according to claim 8, wherein said additional dynamo-electric means has two output windings in which alternating voltages are induced in anti-phase, and wherein the commutator means operate to supply current to a magnetizing winding sequentially from each of the two output windings alternately for successive periods equal in duration to the period between poles of a set coming into regiter.

10. A dynamo-electric machine according to claims 3 and 8, wherein the sets of n stator poles and n rotor poles are duplicated to comprise two units coupled to the same rotor, the relative angular displacement of the pole structure of the two units being staggered when poles in each set are in register, and said two output windings of the additional dynamo-electric means each being connected via the commutator means to energize different said units during one full period equal in duration to the period between poles of a set coming into register and interchanging units for the next such period.