Digital Controlled High Speed Synchronous Motor

Transcription

1 Digital Controlled High Speed Synchronous Motor Zdeněk Čeřovský ), Jaroslav Novák ), Martin Novák ), Marek Čambál ) Czech Technical University in Prague, Prague, Czech Republic ) Faculty of Electrical Engineering, ) Department of Electrical Drives and Traction ) Cerovsky@fel.cvut.cz ) Faculty of Mechanical Engineering, ) Department of Instrumentation and Control Engineering ) Jaroslav.Novak@fs.cvut.cz, Martin.Novak@fs.cvut.cz, Marek.Cambal@fs.cvut.cz Abstract Contribution deals with predictive torque control of the high speed permanent magnet synchronous motor. Such motors are used for electrically driven compressors (E-booster) on turbocharged gasoline or diesel engines. Theory and experimental results from the laboratory are revealed. Keywords Adjustable speed drive, Automotive electronics, Adaptive control, Synchronous motor, Control of Drive I. INTRODUCTION Permanent magnet synchronous machines are broadly used in the propulsion technique. Synchronous motors are very robust and can be used in almost all hard conditions. They are reliable especially in high speed drives. Compressors for turbocharged gasoline or diesel engines are such cases. Turbocharged combustion engines are produced for applications where the mass of the engine is one of many very important parameters. It is so in railway, ship and car transportation. The green house gases production especially the production of carbon oxide and nitrogen oxides is very important parameter recently. Compressors of turbocharged combustion motors are usually driven by gas turbine taking the power from exhausted gas. Compressor blows the air into inlet tube with compression of 4,5 up to 5,5. Turbocharger benefits: Possibility to increase fuel quantity into cylinder Increase engine power with 5% or more Fuel consumption diminishing Fuel burning time in cylinder prolongation Green house gases emission diminishing. Diminishing especially of carbon oxide and nitrogen oxides emissions. Better intake valves cooling Disadvantage of compressors driven by a turbine is the fact that they are unable to deliver needed air amount for the combustion engine in the full range of engine revolutions and power. Following the construction of the turbine and compressor it happens that in low revolutions of the engine the turbine does not give enough power and compressor does not give enough air pressure. In high revolutions the turbine power may be higher then needed on the contrary. In this case it would be possible to drive with the overflowing power an electric generator and to supply the gained power into the internal vehicle network. The turbine driven compressors have relatively low dynamic. The reason lies in the fact that after fuel addition it takes time to raise the engine revolutions to make higher the exhaust of gas, to accelerate the turbocharger, to raise pressure on the compressor and to apply it into inlet tube. The ideal compressor drive should be therefore controlled with respect to the fuel quantity and engine revolutions. Its dynamics should quickly follow the engine fuel control. Turbo compressors are driven mostly by turbine today. It can be added either an electric motor mounted on the shaft between turbine and compressor or the separate additional compressor driven by special electric motor. In the first case the velocity of the motor is the same as the velocity of the turbo compressor. In the second case it is possible to compromise the optimal velocity for the compressor with the optimal velocity for the electric motor. New control manner of the torque of an electric permanent magnet synchronous motor for combustion engine is described in this paper. II. DISCUSSION OF POSIIBLE APROACHES TO HIGH SPEED PERMANENT MAGNET SYNCHRONOUS MOTOR TORQUE CONTROL From the hardware point of view, the standard solution for powering a permanent magnet synchronous motor is a three phase bridge inverter. In the case of automotive industry, the most perspective element is a power FET, considering that the motor is working with a small voltage, maximally 48V, and a switching frequency over khz with minimal losses in the semiconductor element. At our test stand, the motor has a nominal voltage of 4V and we are therefore using an IGBT inverter. The choice of controller structures for high speed permanent magnet synchronous motors can be based on several criteria. In the case of a high speed motor for automotive industry, the goals are maximum efficiency, robustness and minimal price while maintaining reasonable but not extreme control qualities. Also the question of EMC is important. Three control strategies come in question for the torque control of high speed permanent magnet synchronous motor: direct non-linear torque control, simple electronic commutation methods and linear vector control.

2 Direct non-linear control promises very good control loop dynamics, but requires high frequency of changes in the controlled system and high computational power of the controller. For the high speed motor, this would mean an increase in the switching frequency of the inverter, meaning higher switching looses and lower efficiency. The demands for high computational power of the controller would also mean its increase in price. It would also mean less favorable conditions from the EMC point of view. For this reasons, we are not considering direct non-linear control in this paper. Only some experiments with on-off control of instantaneous phase current values were made while synchronizing with the rotors position. However, to achieve sufficient control quality, it would be necessary to increase the inverter s switching frequency so this way was abandoned. Simple methods that use electronically commutated motors are easily implemented even in the cheapest control units; there are even application oriented programmable circuits for this area. The methods are based on the production of square waves of the motor s phase currents whose value is changing 6 times per period. These methods allow to use simple and cheap position sensors with only 6 positions per revolution or to implement sensor less approaches based on measuring of the induced voltage. These methods are mainly used for low power high speed motors and can also be used in the replacement of some non electric drives in automotive industry with higher power motors. The disadvantage of these methods is generally the lower quality of control and in some cases also the impossibility to use generator braking. In the future, we would like to focus also on these methods. Linear control methods can be divided into two groups: control of currents in transformed rectangular coordinate system in the d,q axes and control of instantaneous phase current values with linear controllers. In this case, the requested value of phase currents is synchronized with the rotors instantaneous position. The control in the d,q transformed coordinate system can give in principle higher quality of control, especially for high speed motors, but the algorithm is more complex and demands higher computational power from the control system. On the contrary, the control of instantaneous current values is simpler, but it has much higher demands on the controller operation as it has to work with changing values of the controlled value. In our work we want to focus on a comparison between the properties of both methods of linear control of high speed motors. The comparison will be done from the point of view of control quality, algorithm complexity demands, necessary PWM switching frequency and required number of measured rotor positions. The work was initiated by implementing methods based on instantaneous current values control. III. TORQUE CONTROL DESCRIPTION Torque control uses the basic formula for torque calculation: T =.5 p p ( Fd iq Fq id ) F d is the d-axis magnetic linkage component, F q is the q-axis magnetic linkage component, i d is the d-axis stator () current component, i q is the q-axis stator current component, p p is pole pair number. F d, F q, i d, i q are instantaneous values. The coordinate system is firmly coupled on the magnetic rotor flux in no load and that means that it is firmly coupled on the space rotor position. Fig. Vector diagram for torque controlled synchronous motor with permanent magnets excitation and equivalent scheme. At the torque control of the motor with permanent magnets mounted on the rotor the motor does not need magnetic excitation. We do not want to change the magnetic flux with rotor current and therefore the i d current component is not required. The remaining i q current component is controlled to lead 9 the space rotor position. Therefore it is in phase with the induced voltage U i. Under this presumption it holds for the torque: T =. 5 p F I () p d On Fig. the U is terminal stator voltage, and U i is the induced stator voltage component. Magnetic flux weakening is not assumed so far. Therefore demagnetizing stator current component i d is zero. IV. FEEDBACK TORQUE CONTROL STRUCTURE Designed new torque control structure rises from previously published papers but was enhanced with respect to recent experiments performed in Research Center of Combustion Motors and Automobiles at Czech Technical University in Prague, [4], [5], [6], [7]. Aim of the paper is to improve features of phase current control of high speed synchronous motors with permanent magnets on the rotor. To use on-off controllers [4] is disadvantageous because of the high frequency of needed calculations in the controller and high switching frequency of power electronic devices in the converter which can cause electromagnetic compatibility difficulties and lower converter efficiency. Linear control has simple algorithms [4], [5] and [8]. Therefore it is often used on simple, cheap universal microcontrollers. Its computing demanding is low. This feature is advantageous in high revolution drives. Linear control with controllers in all three phases, with the outgoing signals u Ar u Br u Cr proportional to the phase

3 voltage from controllers to control the PWM modulator has following disadvantages: First disadvantage is, that when the integral component of the current controller is used, then the zero mean values of the per unit voltages u AR, u BR, u CR set as input on the PWM modulator, are not guaranteed. The time curves of u AR, u BR, u CR, can be shifted to the control zone margin. This disadvantage can be eliminated by using only two per unit voltages of u AR, u BR, u CR and by calculating the third from the condition u AR + u BR +u CR =. Important improvement can be reached by modification of the control structure and calculation of each per unit voltages u AR, u BR, u CR as sum of the controller output u PI and induced p.u. phase voltage u i as gives Eq (3): u + R = u PI u i (3) u i represents the compensation of induced stator voltage U i shown on Fig.. The is denoted in the scheme on Fig. as u comp. The use of this method improves markedly the dynamics of the response by which the controller reaches the requested current value. The controller controls then the motor terminal voltage for which it holds u = R i + L di / dt + A u i (4) R and L are resistance and inductance of the stator winding (L d =L q =L ), u i is the induced voltage and i is the motor current. Performed experiments [5] proved the improvement. Nevertheless the control system had at stable proportional and integral controller constants at different revolutions different features. Therefore the adaptation of constants for different instantaneous current derivation and for instantaneous induced voltage was implemented. This adaptation depends on motor revolutions and was implemented by using following equations. K K P I = K = K P I P I i i set set / t / t Iu Pu u u K P is actual controller proportional constant, K I is actual controller integral constant, K P is actual controller proportional constant corresponding to optimal proportional constant at low revolutions, K I is actual controller integral constant corresponding to optimal integral constant at low revolutions, C P and C I are constants corresponding to the weight of current derivation in the proportional respectively integral constant calculation, i set is the changed requested current value during period of controller calculation, t is period of controller calculation C Pu and C Iu are constants corresponding to the weight of instantaneous induced phase voltage u i calculated in controller units. Experiments were performed with this arrangement on synchronous motor 4kW, 5min-, 5Hz. Sufficient control quality was reached on high speed motor with revolutions 5min-. To develop the control system for higher motor revolutions a new testing working place was established. V. NEW TESTING PLACE New testing place for testing static and dynamic features of high speed synchronous motors with permanent magnets on the rotor was built in Research Center of Combustion Motors and Automobile Technique Josef Božek at the Czech Technical University in Prague. Working place consists of: Frequency controlled asynchronous dynamometer,3kw, 35V, 7 min -,,3Nm. (Fig.3) i i (5) Fig. Block scheme of the control system

4 Fig.3 High speed dynamometer with tested Synchronous motor Two pole high speed synchronous motor,9kw, 4V, 4 min -,,7Nm with permanent magnet rotor (Fig.3). IGBT converter with microelectronic controller. Motor was supplied from its separate IGBT converter. Chokes in series with stator winding and inductance of,4mh were implemented to diminish stator current pulsation. Synchronous motor has two pole resolver integrated to determine the space rotor position. Special electronic unit was developed to supply the resover with exciting signal khz and evaluates the signals from the resolver. The unit has bites resolution and discriminates 496 positions in one motor revolution. The information on rotor absolute space position is transferred via parallel bus after reset. Information on relative rotor position is transferred in sensor signal form IRC. More information is in [3]. System DSP TMS3F4 for torque control was used in experiments. Switching frequency was 5 khz and discrimination of 496 positions in one motor revolution was used. The calculation power of the used DSP has limited the motor revolutions on the testing place. It was possible to control the motor up to revolution min - with good results. But later on experiments on new working place showed that at higher revolutions a new problem occurs. The actual current shapes were delayed with respect to the reference values. A high control error between reference and actual current amplitudes was measured. It was not possible to improve the control quality by no controller constants modification nor by adaptation of C P, C I, C Pu, C Iu constants. This problem could be solved by quicker controller. Such a controller is prepared for the future. Higher control signal frequency for PWM will be also possible when the new controller will be used. In the mean time a new solution was found. The frequency of stator current is changing in broad limits in case of high speed motor. The same holds on stator reactance and additional phase reactance. The voltage drop on these reactance s is changing in broad limits in case of high speed motor too. The voltage drop U=jω(L +L pr )I on the Fig. is very high on high speed motor. Therefore the control structure was completed with the compensation of total inductance voltage drop. The claims on the controller work were notably reduced using this strategy. This idea was implemented by changing the calculation of u comp in the modified structure. Its amplitude is calculated as sooner from the curve of induced voltage but the induced voltage is introduced into the control structure as compensating voltage in such a manner that it leads the induced voltage with the angel φ comp. Practically it is in such case in phase with the motor terminal voltage. The conditions for the work of the controller are enhanced a lot by this procedure because the controller does not need to control the whole motor terminal voltage U but only the difference between it and the compensating voltage u comp. The calculation of the angel φ comp uses following idea: At high revolutions there is possible to omit the influence of the winding resistance in the vector diagram. The induced voltage equals ωψ d, where Ψ d is magnetic stator linkage. Induced voltage is turned with the angle φ comp and used as compensation voltage. We calculate the angle for turning the induced voltage from the Eq. (6): ωl I tg comp = = ω F d LI F ϕ (6) d When we assume constant magnetic stator linkage it holds comp = arctg ( kl I ) ϕ (7) Block scheme of control system is on Fig.. The symmetric asynchronous PWM with pulse modulation frequency of 5kHz was used for converter control. Calculation tact in the control structure was synchronized with the pulse modulation frequency. Sampling period of the controller calculation matches with the period at which the actual values are transferred in the PWM modulator. These periods are µs. In the same period is the actual rotor space position evaluated. VI. RESULTS OF TESTING EXPERIMENTS Many testing experiments were performed on the testing place in steady state and transient conditions. Only tests in motor mode were performed. The dynamometer during testing was in speed control mode and that means that it worked in constant given speed. Tested motor worked in torque mode and that means that its torque remained constant, respectively was changed in a step. Tests were performed up to 8min - revolutions. Detail of measured values is shown in Fig.5. Current reference phase value is () and actual value is (). Compensating phase voltage value is (Ubcomp) at revolutions 8 min- and torque,6nm. Reference current value is violet, actual current value is blue and compensating voltage is red. In curves is leading of compensating voltage good seen. On following figures Fig. 6 and Fig. 7 are dynamic curves of reference and actual current values after step by step torque command seen. Reference current curve is marked (), actual current curve is marked (). Reference current amplitude is marked I zad, actual current amplitude is marked I skut.

5 i/a/, u/-/,5, ,5 - -,5 Ubpwm Ubcomp i/a/,5,5 -, ,5,5 I zad I skut Fig. 4. Curves of actual (blue) and reference (violet) current phase values, voltage u R given in PWM modulator (red) and compensation voltage (light blue) without compensation of inductance influence, n=min -, T=,8Nm i/a/, u/-/,5,5 -, ,5,5 t/s/ Ubcomp Fig.5. Detail curves of synchronous motor reference and measured actual currents and of compensation voltage at n= 8 min - I/A/ 3,5,5 -,5 - -,5, I zad I skut Fig. 7a. Reference current step from,8a to A, (T=,4->,8Nm) at revolutions n=5 min - and actual current response. M/Nm/,6,4,,,8,6,4,,,,3,4,5,6,7,8 Fig. 7b. Step response to the change of requested current amplitude from,8a to,a (M=,4->,8Nm) by n=5min - - torque waveform measured with a tenzometric sensor on dynamometer In following Fig. 8 curves of actual current (blue), reference current amplitude (red), actual current amplitude (light blue) and revolutions violet are shown at no load accelerating with torque,4nm. The diagram ends at 768 min - revolutions. Actual curves of phase current (blue), reference curves of phase current (violet), voltage u r that is transferred on the PWM modulator (red) and compensation voltage (light blue) on steady state operation 5 min - and torque,4nm are shown in Fig.9. With respect to low revolutions and low load the leading of the compensation voltage before the current is not visible. t/s/ Fig. 6a. Reference current step from zero to,8 A (T=,4Nm) at revolutions n= min - and actual current response. M/Nm/,,5,,5 i/a/, n/hz//,5,5 -, ,5,5 n I zad I skut,,,3,4,5,6,7,8 -,5 t/s/ Fig. 8. No load speeding up with torque,4nm up to 768 min - revolutions. Fig. 6b. Step response to the change of requested current amplitude from to,8a (,4Nm) by n= min - torque waveform measured with a tenzometric sensor on dynamometer

6 i/a/, u/-/,5,5 -, ,5,5 Ubpwm Ubcomp Fig. 9 Reference and actual current curves, voltage ur on the PWM modulator input and compensating voltage on steady state operation at 5 min - and,4nm. VII. CONCLUSIONS As it is clear from the measured waveforms, the torque control quality is good in both steady states and transition states while maintaining a simple algorithm. In transition states a new stable value is reached in some milliseconds. An increased ripple of the current (current is proportional to internal motor torque) for increased speed is evident from figure 8. It is a principal property given by a decreasing ratio between the period of fundamental harmonic of the inverter output voltage and PWM period. The controller structure is in principle based on the prediction of instantaneous phase voltage values calculated from rotor position, speed and requested torque. The current controllers in two phases of the circuit are then only working with a small control deviation corrections. The described method has a simple algorithm and good control quality, but can be used only in areas where generator braking is not necessary. For our area of high speed non-electric motor replacement this is however not an issue. At the present time, we are implementing a new, more powerful controller based on TMS3F8 (Fig.,) which should allow us to increase the PWM switching frequency to khz and to test the motor for speeds up to 4 min -. We will also focus on experimentally comparing the properties of the described simple controller structure with a d,q rectangular co-ordinate system control for high speed synchronous motor torque control. Next area will be to optimize the rotor position measurement system with regard to reduce the number of measured positions per one revolution while maintaining good torque control quality or to use a sensor less approach. Fig.. Drive controller with TMS3F8 and position measurement add-on ACKNOWLEDGMENT Authors would like to thank the Josef Božek Research Center for the financial support of the work project M568. REFERENCES [] Novotny D.W., Lipo T.A.:Vector Control and Dynamics of AC Drives, Oxford Science Publications Nr 4, 996 [] Čemus J., Hamata V.: Transient Stability Analysis of Synchronous Motors. Czechoslovak Academy of Sciences, Academia/Prague 99 [3] Čambál, M. - Novák, M. - Novák, J.: Study of Synchronous Motor Rotor Position Measuring Methods. In 3th International Conference on Electrical Drivers and Power Electronics. Zagreb, Croatia: KoREMA, 5, p ISBN [4] Novák, M. - Čambál, M. - Novák, J.: Application of Sinusoidal Phase Current Control for Synchronous Drives. In ISIE 6 International Symposium on Industrial Electronic [CD-ROM]. Montreal, Canada: IEEE Industrial Electronic Society, 6, ISBN [5] Čambál M. - Novák M. - Novák J.: Possibilities to Increase the Quality of Phase Current Control for Synchronous Motors, The 5th Mediterranean conference on Control and Automation - MED , Athina, Grece [6] Uhlíř, I. - Čambál, M. - Novák, M. - Novák, J.: Synchronous Motor Current Controler Quality Augmentation with Adaptive Control, The 33rd Annual Conference of the IEEE Industrial Electronics Society. Taipei, Taiwan: IEEE Industrial Electronics Society, 7 [7] Čambál, M. - Novák, M. - Novák, J.: Synchronous Motors Phase Current Adaptive Control, Proceedings of the 8th International Carpathian Control Conference. Košice: Technical University, BERG Faculty, 7 [8] Šimánek, J. Doleček, R., - Černý, O.: PMSM Drive Control based on Sinusoidal Commutation Control of Brushless Motor, Proceedings of the International Carpathian Control Conference ICCC 8, Sinaia, Romania, May 8 Fig.. Experimental inverter for high speed motor