Modern electric motors are preferably fitted with an additional system for precise position control. One way to achieve this is use of an external position sensor (optical encoder, magnetic resolver, and like). Another method, which is often more attractive, is sensorless position control.

Many types of sensorless monitoring of rotor position in synchronous electric motors are already known. Motors with such control have no need of optical encoders, or resolvers, or some other sensors of rotor position. In such sensorless control, the rotor position is usually determined from the emf voltage in the motor winding. One part of this voltage is the Ohm's voltage and is proportional to the ohmic resistance of the coils and to the electric current in these coils. The other part is the voltage (which is usually the greater part) produced by magnetic induction in the motor coils. This induced voltage is a complicated mathematical function of several variables, including the rotor position. With modern Digital Signal Processor (DSP) analysis, it is possible to determine the rotor position from known values of other variables. These variables are the electric currents in the coils belonging to all the motor phases (e. g. , three currents in three-phase motors) and the voltages in these coils (in three-phase motors, again three variables).

However, this method has at least one major drawback : The rotor position cannot be determined at zero speed of the rotor, since at zero speed there is no induced voltage. So usually motors with sensorless control according to the state of the art make several small blind jerks at the motor start (say, jerks by 60 electrical degrees) to determine the starting position. But these jerks are not compatible with the desired smooth start. They are undesirable especially in applications where high precision is indispensable (like in robotics, FA, etc.).

On the other hand, there are sensor controls. However, traditional resolver methods introduce massive parts besides the motor, and magentic resolvers and motors can also magnetically interfere which can also be a problem of many designs.

The problem of the present invention resides therefore in finding such a constructional change of the hybrid synchronous motor that the motor and the magnetic resolver are integrated together. Further, this change should not introduce any massive, expensive or complicated parts into the motor. Therefore, we are looking especially at such solutions that are without considerable constructional modifications.

In comparison to sensorless methods, this invention introduces a motor with the same low weight, and further, someone can get precise information about the rotor position even at zero speed.

This problem has been successfully solved by a small constructional addition in the hybrid synchronous motor, namely a special measuring coil which is added to conventional power coils (in the following, this special coil is called the measuring coil). Further, another part of solution is a special method of generation of pulse-width modulated (PWM) pulses in the polyphase motor coils. So in comparison to classical sensorless hybrid synchronous motors, the motor according to the present invention has an additional measuring coil, special PWM generators and decoder, which calculates the rotor position from the emf induced in the measuring coil. The Digital Signal Processor (DSP) performs regulation of the synchronous phase in such a way that it determines the precise width of PWM pulses in each one of the polyphase coils from the rotor position.

This new method of position control is the subject of the present invention. The principle and the functioning of this control will be expounded with regard to the hybrid synchronous motors of the type explained in the patent No. PC/JP01/00070 (hybrid synchronous motor with toroidal coils). Of course it can be used also with other members of the class of hybrid synchronous electric motors. In further text, the motor, explained in the above mentioned patent shall be named Mukade motor.

The motion control accordingly to the present invention functions with undiminished high precision also at zero speed but preserves full advantage of sensorless controls. Only a small and very simple measuring coil is introduced into the motor interior. There are no other additional parts in the motor.

The invention will now be further explained by referring to the drawings illustrating examples of the machine and electronic functional blocks according to the present invention, showing: Fig. 1 is an axonometric view of a three-phase hybrid synchronous electric machine with the measuring coil according to the invention, in partial cross-section, Fig. 2 is the same as Fig. 1, but in an expanded view, so that the rotor and the stator are shown separately, Fig. 3 shows temporal patterns of PWM pulses in the three coils (belonging to the phases U, V, W) of a three-phase synchronous motor according to the invention, Fig. 4 shows the block scheme of the special PWM generator for the three phase system, according to the invention, Fig. 5 shows the block scheme of the decoder of the rotor position, according to the invention.

The invention will be explained with regard to the example of a three-phase synchronous motor with toroidal coils (this is the motor of the type Mukade), but the same theory is valid also for other polyphase hybrid synchronous motors.

In Fig. 1 an embodiment of a three-phase hybrid synchronous electric machine with the toroidal coil and with the measuring coil according to the invention is shown. The same embodiment is shown also in Fig.

2, with proviso that in Fig. 2, the rotor (15) and the stator (14) with a ball-bearing (8) are shown separately. The active parts of the rotor are four cogged iron rings (9,10, 11, 12). The active parts of the stator are two cogged iron rings (2,3), the ring-shaped permanent magnet (4) inserted between the. se two rings (2,3), and the toroidal coil (6) wound onto the assembly of the stator rings (2,3, 4). Accordingly to the present invention, a simple measuring coil (1) which is coaxial with the motor axis (7) is wound onto the ring- shaped permanent magnet (4). The measuring coil (1) can be made of a very thin wire so it does not occupy a lot of place and does not considerably change the geometry of other parts of the motor. The two ends (la, lb) of the measuring coil (1) are led out of the motor through the stator housing (5), where they are connected to the electronic control unit, more precisely to the decoder of rotor position.

The above-described constructional solution of the motor is already known, with the exception of the measuring coil (1). Therefore here we shall describe only the functioning of the measuring coil (1), which is the subject of the present invention.

The measuring coil (1) is wound around the ring-shaped permanent magnet (4) therefore it measures the changes of the magnetic flux through this ring-shaped permanent magnet (4). During motor operation, the magnetic flux through the magnet (4) suffers a continuous change, which is small but measurable, for instance by means of the measuring coil (1). This flux change is approximately zero if the working currents in polyphase motor coils (6) are well synchronised with the rotor position; and increases as soon as the synchronisation goes away from the ideal value of the electric angle.

Accordingly to the mathematical theory of the Mukade motor and similar hybrid synchronous motors, this flux change is Where: AO is a small change of the magnetic flux through the permanent magnet (4) (change from the average value), K is some multiplication constant and is approximately is the average magnetic flux through the permanent magnet (4), at the working point of the magnet, Io is the electric current amplitude in each phase of the working coils (6) of the motor, IOmax is this amplitude at peak torque of the motor, and y is the phase shift (electric angle) of the electric current in the main coils (6) of the motor, from that phase shift at which the working current is best synchronised with the rotor position.

The formula above is valid for the working current in the motor coils (6) (namely, the current that should be synchronised with the rotor position) ; but just as well also for the current of any other frequency (therefore, the current which is not synchronised with the rotor position). The only condition is that this current is also a three-phase current like the working current in the motor coils (6).

If frequency of electric current is not the synchronous frequency of the motor then y is a variable which changes in time.

According to the invention, the generator of the three-phase power signal (electronic inverter) produces such a signal that the current in the three-phase system of motor coils (6) is a sum of two different three-phase currents: - The first electric current is the working current (the main current that produces motor torque) so it has a considerable amplitude which is approximately proportional to the motor torque. The electric phase of this three-phase current is ynchronised with the rotor position, therefore its frequency is the synchronous frequency of the motor.

- The second current has much higher frequency but much smaller amplitude. Otherwise, it is also a three-phase current: electric oscillations in the adjacent phases U, V, W are mutually phase- shifted for 120 electrical degrees. Let us call this second current the measuring current.

This second current is not synchronised with the rotor position, therefore the angular shift y of the measuring current is changing all the time. This phase-shift is the difference between the synchronous phase of the working current xt and the phase of the measuring current't. So we have where: o is the synchronous circular frequency of the working current defined as: where ##m is mechanical angle of rotor and Kr number of rotor poles (cogs).

'is the circular frequency of the measuring current, t is time and yo is the phase-shift at time t = 0. If we take the equations (1) and (2) together, we get a small flux change which is the result of the measuring current: This is the magnetic flux of the permanent magnet (4), therefore it goes through the measuring coil (1). The same flux goes through stator poles belonging to all three phases (U, V and W), hence it incorporates information of all three phases. The induced voltage Ui in the measuring coil (1) is, according to the law of magnetic induction and according to the equations (1), (2) and (3): Here we used the following designations: Ui is the voltage induced in the measuring coil, N'is the number of turns of the wire in the measuring coil, Io is the amplitude of the measuring current (in each one of the three phases), all the rest has the same meaning as above.

The electric phase of this induced voltage Ui (in the measuring coil) will be designated'and is from equation (4) found to be The phase shift between the measuring current in the working coils (6) and the induced voltage in the measuring coil (1) is Surely, the phase of the measuring current is dependent upon the fact which of the three phases we choose for comparison. Therefore also the angle yo is dependent upon which phase we make the comparison with. For the sake of theoretical argumentation, let us assume that we have an infinite number of phases, or at least many more than three or five. Then we can always find such a coil segment (belonging to a definite motor phase) in which we have yo = 0. Let us call this particular coil segment the zero coil segment. This is for instance the coil segment belonging to the phase U. So the phase shift between the measuring current in this zero coil and the induced voltage in the measuring coil is: More generally, co is time dependent (m (t) ) therefore the product (cor) should be replaced by the integral: Substituting equation (5) into equation (6) we also get: Now let us remember once again that cot from equation (7) is the synchronous electric phase in the ideal case (that means, when synchronisation is ideal). While the rotor turns forward for one rotor pole division, we get one electric cycle of the working current, so in this time the synchronous phase ot increases exactly for the full electric angle 2fut. But this means that of is also the mechanical angle of the rotor position, multiplied by the number of rotor poles: where Vm stands for the mechanical angle of the rotor. Equation (9) implies linear relation between the mechanical phase Vm and the phase shift Information about the rotor position is therefore given by the phase shift A. During one motor cycle the phase shift A linearly increases from 0 to 27c. This is in the same time the phase of the polyphase working current, which also goes linearly from 0 to 2n, and powers the polyphase coils in the succession U, V, W. The phase shift A is calculated by means of the algorithm according to equation (8).

If we go through all the mathematics of the motors of the type Mukade, we find out also the amplitude of the induced voltage Ui compared to the amplitude of that voltage induced in the working coils by the measuring current: where: Uio is the amplitude of the voltage induced in the measuring coil upo is the amplitude of the voltage of the measuring current (in the working coil) N'is the number of turns in the measuring coil N is the number of turns in one segment of the \emph {Mukade} motor is determined by geometry of the rotor and stator poles and the 8a air-gap width; a typical value is about 0.033 is the circular frequency of the synchronous working current m'is the circular frequency of the measuring current.

Another integral part of this invention is the method, by which the polyphase (typically three-phase) working current, and the polyphase (again typically three-phase) measuring current can be produced simultaneously by one single inverter. Usually the sinusoidal signal of the working current is constructed from PWM pulses with a certain pulse frequency. For example, if this frequency is 17 kHz, then one period of the pulse train is approximately 60 microseconds. The PWM pulses in the coil belonging to the first phase (phase U, coil U) represent the first pulse train. According to the invention, the PWM pulses in the coil belonging to the second phase (coil V) are phase- shifted for one third of the pulse period (for 20 microseconds in our example) relative to the pulses in the U phase, and similarly, the pulses in the coil W are phase-shifted for two thirds of the pulse period (40 microseconds in our example). The temporal patterns of the PWM pulses in all three coils (U, V, W) of a three-phase synchronous motor according to the invention are shown in Fig. 3 (PWM phase modulation).

This method of shifting the centers of the PWM pulses separately in each phase leads to generation of a new rotational magnetics field in the stator coils (6) of the motor, with the circular frequency'.

In each phase (U, V, W) we have a sinusoidal working current with a synchronous circular frequency. This working current is composed of PWM pulses, whose centers are spaced with circular frequency'. In the above mentioned example,'= 2s 17 kHz = 1. 07'105 s-1.

The explanation presented above was given for the three phase system.

More generally, we define this method for any polyphase systems. Let us define n-phase system with n working currents Ij where j=0... (n-1).

Each current Ij is generated by the PWM generator which can shift the centers of the PWM pulses for each phase separately. Let us define a phase shift of the center of each PWM pulse j belonging to phase j, according to some reference position which is arbitrary. According to the invention, it is declared that: where k=O... (n-1). The relation between mechanical phase and phase of the induced voltage in the measuring coil is linear in special case when: as it was the case our above described example of a three-phase motor.

Inside an adequately short time interval, the successive pulses are nearly equal, so the pulse train can be approximated by a periodic function with the circular frequency'. Now this periodic function can be expressed by the Fourier expansion. The circular frequency' is then exactly the circular frequency of the first harmonic component in this expansion. Therefore in the phases U, V and W, this first harmonic components are mutually phase-shifted by one third of the period. In comparison to the first harmonic component in the coil U, the first harmonic component in the coil V is phase-shifted for one third of the pulse period (20 microseconds according to our example), and the first harmonic component in the coil W is phase- shifted for two thirds of the pulse period (40 microseconds). This is exactly in the same way as necessary in a three-phase current system.

In this way, we can produce a three-phase measuring current, in addition to the three-phase working current, by using only one single inverter. The three-phase measuring current is exactly the first harmonic component of the pulse trains in the phases U, V, W.

The block scheme of such inverter for the three phase system is shown in Fig. 4. Clock generator generates PWM carrier circular frequency 'which is equal to 2afp. Three separated PWM generators are clocked with this frequency through clock input and externally synchronised through phase inputs. Variable parameter value determines the width of PWM pulses of, each phase separately, which leads to synchronous working current.

PWM pulses can be represented as a sum of the first harmonic component with circular frequency'plus higher harmonic components, as well as low working circular frequency. Relative values of higher harmonic components can be determined by Fourier analysis.

In practice, these higher harmonic components may influence the precision of the described measuring method.

We have just seen that the first harmonic component obeys the rules for three-phase currents. Further analysis shows that the fourth, the seventh,... harmonic components also yield three-phase currents (naturally, with frequency of these higher harmonic components). The second, the fifth,... harmonic components also yield three-phase currents, but with reversed sequence of the phases, namely W, V, U instead of U, V, W. The third, the sixth,... harmonic components do not yield three-phase currents at all, but simultaneous oscillations in all the three phases (which is even partially cancelled out in the adjoining coil segments of the same phase). The following table shows the behaviour of higher harmonic components. For each harmonic component we can see schematically what kind of oscillation is produced in the three-phase coils U, V, W: Harmonic component Type of oscillation ist) 4th,...(3n + 1)th Three-phase oscillation, sequence U-V-W-U-V-W-... 2nid, 5th,... (3n-) Three-phase oscillation, sequence W-V-U-W-V-U... 3rd, 6th,-(3n) Simultaneous oscillation which is magnetically mainly cancelled One good method which adequately reduces unfavourable influences of the higher harmonic components upon the positioning precision, is the application of a bandpass filter with center of circular frequency approximately co and with filter width of approximately 2.

Fig. 5 shows a block scheme of position decoder. First of all, the induced voltage of the measuring coil (UI) is led through a bandpass filter. Since we have used a narrow-band filter eo and higher harmonic components of'are filtered out and only the desired first harmonic component of'is led to the digital phase detector. The detector compares the phase of first harmonic to a reference phase where j is arbitrary (it may be of the phase U, V or W for three phase systems).