EVALUATION OF A NOVEL ANALOG BASED CLOSED-LOOP SENSORLESS CONTROLLER FOR SWITCHED RELUCTANCE MOTOR DRIVE S K Mondal1,2, S N Saxena2, S N Bhadra3 and B P Muni2 1 Dept. of Electrical Engineering University of Tennessee Knoxville, TN 37996-2100 Tel: (865) 974 9886 e-mail: smondal@utk.edu 2 Bharat Heavy Elecricals Ltd. Corporate R&D Division Hyderabad - 500 093 India 3 Dept. of Electrical Engineering Indian Institute of Technology Kharagpur – 721 302 India region of the phase at some angle before the instant at which the positive-dL/dθ starts. The IGBT in series with the stator phase must be turned-off in the positive-dL/dθ region, at some angle ahead of the instant where the inductance becomes maximum (Lmax). Abstract - A novel analog closed-loop controller for a 4phase Switched Reluctance Motor (SRM) in variable speed drive applications, dispensing with position and speed sensors, has been proposed in this paper. The scheme is based on comparing the active current of a phase with a closed-loop switching reference current (Ioref) with the intention of switching-on and switchingoff the other phase in the pair, which comprises of two stator phases, whose Lmin-zones do not overlap. This technique requires no parameter measurement and/or estimation and without prior knowledge of stator fluxlinkage variation. Paper also includes an analog controller with position sensors to evaluate the developed sensorless controller. A detailed simulation study of both the drive systems, supported by experimental steady-state and transients results, are presented to validate the proposed technique. A double polynomial-fit is used in the simulation study to describe the non-linear magnetization characteristics of the SRM. The performance of the sensorless drive is excellent and is in good agreement with the drive with position sensors. To overcome the drawbacks of using direct rotor position sensors, various methods of sensorless rotor position detection are available [2]. The general principle behind most sensorless drives is that, if at a given instant, fluxlinkage (ψ) (or inductance (L)) and phase current (i) are obtained, rotor position (θ) can be obtained from preknowledge of the SRM magnetic characteristics. A number of papers have been published on the analysis, design and control of SRM taking saturation into account [3]. Authors propose a simplified mathematical model, representing flux-linkage characteristics, for use in the simulation study of a saturated SRM. A double polynomial-fit is used to describe the non-linear relationships between ψ, i and θ. This model does not require mechanical details of the machine, but requires only the experimental determination of the magnetization characteristics. 1. INTRODUCTION The Switched Reluctance Motor (SRM) is now a recognised machine on account of its several advantages [1]. It is a doubly-salient electric machine, with single excitation to the stator winding. When a stator pole-pair is energized, a flux-path is established. The torque is generated by the tendency of the rotor pole-pair to align with the stator pole-pair. However, SRM needs real time rotor position detection, either directly or indirectly, for energization and de-energization of its stator phases in sequence with relation to its rotor position to obtain the desired control of speed and torque and to maximize the efficiency. The controller can be implemented using analog, microprocessor or personal computer based techniques. An inverter is a must with the SRM. The IGBT in series with a stator phase must be turned on in the Lmin- 2. PHILOSOPHY AND CONCEPT OF DEVELOPED SENSORLESS CLOSED-LOOP SRM DRIVE The basic principle of a sensorless control scheme, without prior knowledge of stator inductance variation to extract the instantaneous rotor position information, is introduced in this section. The proposed control approach requires no parameter measurement and/or estimation. The controller block diagram including its power circuit is shown in Fig. 1 for the most widely used 8/6-pole, 4-phase, 4 kW SRM. The principle of the proposed analog closed-loop sensorless SRM drive, dispensing with speed and position sensors, is based on comparing the active current of a phase with a closed-loop switching reference current (Ioref) 0-7803-7116-X/01/$10.00 (C) 2001 IEEE 2073 with the intention of switching-on and switching-off the other phase in the pair, which comprises of two stator phases, whose Lmin-zones do not overlap [4]. The value of Ioref, which is determined by the closed-loop speed control of the SRM drive, depends on the state of the motor (starting or running) and/or the load demand. This approach offers a very cost-effective, simple and rugged controller for the SRM drive. Table-1.2: Switch-on logic Phase of the falling current 1 4 3 2 Phase to be made ON 3 2 1 4 With the above logic, the phase-1 remains OFF as long as i3 is above iref value; and the phase-1 remains ON as long Fig. 1: Block Diagram of Sensorless Controller The ON/OFF logic of the phases is selected in a way so as to ensure that an incoming phase is switched-on in its Lminzone and the outgoing phase is switched-off near its Lmaxzone. For this purpose, the rising current in an incoming phase is compared with Ioref; and when this phase current exceeds the Ioref, a logic is used to switch-off the IGBT of the other pair-phase which is nearing or has crossed the Lmax-zone. This sequence of switching on and switching off logic is illustrated in Fig. 2 with reference to phases 1 and 3, whose Lmin-zones do not overlap. Phase-3 is turnedon in Lmin-zone; and when the rising current i3 exceeds the Ioref, this HIGH/LOW logic is used to switch-off the IGBT1 of phase-1. Similar switch-off logic is decided for the other three phases also, as detailed in Table-1.1. Next, when the same phase current i3 in its falling mode becomes lower than the switching reference current Ioref, this rotor position is detected and is used to switch-on the IGBT of the other pair-phase (i.e. i1), which was turned-off by the phase-3 during its rising current mode. The switch-on logic of the four phases has been furnished in Table-1.2. Fig. 2: Generation of Switch-on and Switch-off Signals as i3 is below iref value. The principle is applicable for even number of phases. The strategy, however, becomes operative only after the motor starts. A starting logic circuit has been incorporated in the scheme. Phase to be made OFF The actual phase currents of the 4 phases (i1 to i4) are sensed through four respective Hall-effect current sensors (HC1 to HC4). These phase currents are used for position detection, as well as for speed estimation and for balancing the split-link dc voltages. The phase currents are compared with a switching reference current (Ioref) in a comparator block CR (block-(1) in Fig. 1) to decide the switch-on and switch-off logic for the appropriate phases. The particular phase(s), which is(are) in Lmin-zone at that instant, will only give HIGH output at the output of the comparator (as di/dt=(v-Ri)/Lmin). This HIGH status will remain HIGH as long as the value of the particular phase current is greater than Ioref. 3 2 1 4 Ioref is obtained as the algebraic difference of the reference current (Iref) and the offset current (Ioff) (block-(7) in Fig. 1). Iref is the output from the PI-block, limited by the current limiter block. Ioff is a fraction of the reference Table-1.1: Switch-off logic Phase of the rising current 1 4 3 2 2074 current (k.Iref). If there is any change in speed and/or load, the value of Iref is changed dynamically, which, in turn, changes the value of Ioref dynamically with the changes in the value of Iref. Switching reference current (Ioref) has profound influence on dwell angle. The value of Ioref has not been analytically established. However, the best course has been found to be experimental tuning of Ioref. Extensive tests carried out suggests that Ioref should be between 65 to 85 per cent of the reference current (Iref). Lower value of Ioref suits the increase value of load torque. generation of actual speed signal, using a frequency-tovoltage converter IC (block-(5) in Fig. 1), which gives a voltage signal (ωact) proportional to the actual speed. Speed error (ωerr) is processed through a PI block to generate the reference current (Iref). The HC block (block-(2) in Fig. 1) is a hysteresis type current regulator, which attains OFF or ON state according to whether the phase current is greater or lower than the reference current. This block is used for balancing the phase currents. By introducing this block, it has been ensured that all the four phases carry balanced currents. This also takes care of the voltage balance in split-link dc supply. The HIGH/LOW comparator outputs of the four phases from the CR block are fed to the LOGIC block (block-(3) in Fig. 1). The LOGIC block consists of three sub-blocks. To start the motor in a particular direction from any starting position, initially all the phases are fired in a sequence one after another. For example, for starting in the forward direction, the starting sequence is 1, 4, 3 and 2. Each output of the HC block is ANDed with the corresponding ON/OFF signal output from the LOGIC block to obtain the final signals (block-(4) in Fig. 1) for the gate-drive circuits of the IGBTs, controlling the currents in the four phases of SRM. As soon as the motor starts rotating in a particular direction, the starting sequence logic becomes ineffective. Therefore, just immediately after the start, all the four phases are ON. As can be seen from Fig. 3, there is always at least one, or maximum two Lmin-zone (this overlapping zone is small), at any rotor position. Thus, only that particular phase current, which is switched-on in Lminzone, will cross the reference value (Ioref) first, and switchoff the IGBT of a pair-phase unambiguously as per Table1.1. Also in the Lmin-zone, since the back emf of rotor is zero, the current-rise is independent of the motor speed. Thereafter, depending on the logic (as given in Tables-1.1 and 1.2), the phases are switched-off and switched-on sequentially. 3. PHILOSOPHY AND CONCEPT OF DEVELOPED CLOSED-LOOP SRM DRIVE WITH POSITION SENSORS This section presents a direct control strategy based on rotor position information obtained from a physical position sensor. One rotor replica is mounted on the rotor shaft of SRM at the non-drive end of the motor, and four real time position sensors (which are proximity switches) are mounted through the holes at 450 apart on an annular circular frame mounted at the same end on the stator in the plane of the rotor replica and concentric with it. Rotor replica moves within this fixed circular frame. Fig. 3 shows the idealised variation of inductances (L1 to L4) of the four stator phases with rotor angle (θ) over one rotor pole pitch of the SRM. Four sensors for the four phases occupy the positions, each of which is at 30 after the front edge of the respective stator pole (Fig. 3). Each sensor gives a HIGH signal whenever the leading edge of a rotor pole replica passes under the sensor. The duration of this HIGH signal is equal to the rotor pole-arc. Thus, four signals (S1 to S4 in Fig. 3) are obtained, one each at the position of 30 from the start of the positive-dL/dθ region of one phase over the rotor pole arc. This signal would become LOW as soon as the trailing edge of the rotor pole replica crosses that proximity switch. At low speed (<250 rpm), since the back emf is small, the current-fall may be slower in its falling mode. Due to this, phase current may not come down below the switching reference current. This may lead to the continuous OFF of a phase in a pair; and the motor may stop, as the logic in Table-1.2 fails. To avoid this, external ON and OFF logic signals are generated from variable frequency oscillator; and these logic signals are ORed with the main logic signals (which are LOW) to make the phases ON and OFF sequentially. This low speed logic below 250 rpm operates in open-loop control. Running in closed-loop operation below 250 rpm requires the phase voltage to be made a control variable. For this, a feedback loop to control the dc-link voltage at low speed is under progress. Depending upon the motor characteristics, it is possible to have different ON/OFF signal sets for different speed zones by changing the initial location of the position sensors before starting. It is seen from the Fig. 3 that three different ON/OFF signal sets for the three different speed zones are generated from the existing fixed four ON/OFF signals. In the low speed zone (0<w≤500 rpm), switch-on angle is -5.50, and switch-off angle is 180. In the medium At high speed (>250 rpm), the phases are switched off and switched on sequentially, as given in Tables-1.1 and 1.2, respectively. The speed control is done through a closed-loop PI control (block-(6) in Fig. 1). The HIGH/LOW signal output of any one phase current of the CR block is used for the 2075 speed (500<w<1200 rpm) zone, switch-on angle is -120, and switch-off angle is 180. In the high speed zone (w≥1200 rpm), switch-on angle is -20.50 and switch-off angle is 180. where i Wc(i,θ) = ∫ψ(i,θ)di 0 (2) A phase winding of the SRM can be represented as a nonlinear (due to magnetic saturation effects), time-varying (due to changes in rotor position with time) inductance. Magnetization characteristic, neglecting hysteresis, for a given θ between θu and θa may be represented by a polynomial in phase current i with odd powers (n=1,3,5…….) and coefficients that are function of rotor position θ. 4. MODELING OF SRM – POLYNOMIAL-FIT The simulation of electric drives helps in the pre-study of the desired performance and in designing both power and control circuits. Detailed simulation study to foresee the performance of two drive systems has been done. ψ(i,θ) = ∑an(θ)in n (3) Fig. 3: Switching Logic Diagram for Controller with Position Sensors with Idealized variation of L Fig. 4: Typical Flux-linkage (i, ) and Inductance L(i, ) characteristics for a SRM Phase Winding The performance of SRM is influenced by its magnetic (flux-linkage) characteristics and the control strategy. For the SRM, ψ or L, is a non-linear function of i and θ, as can be seen in Fig. 4; where, θu represents the unaligned rotor position of minimum inductance and θa represents the aligned position of maximum inductance. If iron losses are neglected, the static magnetization characteristics can be applied to evaluate the developed torque under dynamic conditions when i and θ change continuously. Odd powers of i have been taken since the same flux level is set-up by both the positive and the negative values of i. Coefficient an(θ) is again a non-linear function of θ, and should exhibit a monotonous change in its value with the increase in θ up to the maximum value of θ (i.e., aligned position). Secondly, it should also exhibit periodicity as the rotor poles move past the stator poles. Another experimental evidence, viz torque is zero at aligned and un-aligned positions, may be considered while giving mathematical expression to an(θ). Considering these aspects, an(θ) may be expressed as a polynomial in sinkθ with even powers. Like in any electromechanical energy converter [5], the instantaneous electromagnetic torque Te(i,θ) produced by one phase at any θ can be calculated from the winding coenergy Wc(i,θ) by the relationship : Te(i,θ) = (∂Wc(i,θ)/∂θ) (1) 2076 5. RESULTS (n) an(θ) = ∑dmsinmkθ m (4) This section presents experimental waveforms obtained for the same prototype SRM drive as well as the results for the simulation study. The parameters of the SRM are given in Appendix-II. The SRM was coupled to a dc generator. The load on SRM was varied by varying the load on the dc generator. The drive system was tested extensively under both steady-state and transient conditions; but, only a few pertinent results are furnished in this section. for m=0,2,4,6……., and k is (Π/2θa). Substituting the value of ψ from equation (3) to equation (2) gives the following expression for co-energy Wc : Wc = ∑an(θ)(in+1/n+1) n=1,3,5 (5) From equations (1), (4) and (5), the motor torque produced by a phase current is: (n) Te(i,θ)=(Π/2θa)coskθ∑(∑mdmsinm-1kθ)(in+1/n+1) n m (6) for m=0,2,4…….; n=1,3,5…… ; and θ expressed in radians. For each of the four phases of SRM, the electromagnetic torque component can be found out using the corresponding values of current and rotor angle; and the total electromagnetic torque developed by the SRM would be the algebraic sum of the torque produced by each phase at an instant. Measured points [6] and the best fit polynomial (equation 3) of the machine flux-linkage are shown in Fig. 5. Reasonable agreement can be noted. Appendix-1 gives the numerical values of the coefficients in equations (3) and (4). Predicted static torque, based on equation (6) and some measured values [6] show reasonable agreement. Fig. 6: Phase Currents for Sensorless Scheme at 500 rpm; i1 Presented for Comparison and Reference (Scale: 4A/Div.) Fig. 6 presents typical steady-state phase current waveforms for system without sensors at 500 rpm. The corresponding simulated current waveforms are shown in Fig. 7. The waveforms indicate that the machine phase currents in all the four phases are well balanced with the desired phase shifts of 450, 300, 150. It is observed that the simulated results match closely with the experimental results. The developed ON/OFF switching logics for the sensorless drive has been confirmed both by experimental as well as by simulated waveforms, as can be seen from Fig. 8 for the simulated conditions and from Fig. 9 for the experimental conditions at a speed of 1000 rpm. The sequence of switching on and off logic is illustrated in Fig. 5: Magnetization Characteristics with Pole Alignments as Parameters 2077 these figures with reference to pair phases-1 and 3, whose Lmin zones do not overlap. Figs. 10 through 12 show the building-up of actual speed, reference current and phase-1 current upon sudden application in the set speed of 550 rpm from zero value for both the schemes. Overall agreement in the response between the two schemes is noted, inspite of starting the drive at any arbitary rotor position with sensorless scheme. Responses for both the schemes are almost identical. Similarly, Fig. 13 shows the response for sudden change in load torque for the set speed. Results are matching for both the schemes. Therefore, the developed controller gives steady-state and transient performances comparable to those of SRM drive with position sensors whose controller has also been developed. Fig. 8: Phase-1 and Phase-3 Currents, Control Signals, Iref and Ioref (Sensorless Scheme (Simulation)) at 1000 rpm (Scale for Control Signal: 7.5 V/Div.) Fig. 9: Phase-1 and Phase-3 Currents and their Control Signals (Sensorless Scheme) at 1000 rpm Fig. 7: Phase Currents for Sensorless Scheme at 500 rpm (Simulation) i1 Presented for Comparison and Reference 6. CONCLUSIONS Fig. 10: Speed Response during Starting Ref. Speed: 550 rpm; Shaft Torque: 2 Nm (Top: Ref. Speed, Bottom: Actual Speed) (Scale: 500 rpm/Div.; 200 ms/Div.) The paper has described a novel position and speed sensorless analog SRM controller without parameter estimation and without using any complex control circuit. The active phase current waveforms of the excited stator phases have been used to sense the rotor position. This avoids the need for pre-programmed ICs. The developed 2078 Fig. 13: Speed Response during Sudden Change in Load Torque; Ref. Speed: 550 rpm (Top: Actual Speed, Bottom: Phase-1 Current) (Scale: 500 rpm/Div., 8 A/Div.; 1 s/Div.) Fig. 11: Ref. Current and Actual Speed during Starting; Ref. Speed: 550 rpm; Shaft Torque: 2 Nm (Top: Ref. Current, Bottom: Actual Speed) (Scale: 8 A/Div., 500 rpm/Div.; 2 s/Div.) The starting of SRM is possible from any arbitrary position and in any particular direction, without following the usual approach of the inductance measurement at standstill by diagnostic pulses. Total control circuit is based only on the conventional analog and digital ICs, making the measurement extremely fast. The power circuit cost is also low due to the use of single device per phase. References Fig. 12: Ref. and Actual Phase-1 Current during Starting Ref. Speed: 550 rpm; Shaft Torque: 2 Nm (Top: Ref. Current, Bottom: Actual Current) (Scale: 8 A/Div.; 2 s/Div.) controller is conceptually simple and gives steady-state and transient performances comparable to those of SRM drive with position sensors whose controller has also been developed. It provides a simple and cost-effective alternative to the complex sensorless controllers described in the literature. The applicability of the developed sensorless technique has been validated by means of extensive simulation and experimental results. 2079 1. T J E Miller, “Switched Reluctance Motors and their Control”, Magna Physics Publishing and Clarendon Press, 1993. 2. B Fahimi, G Suresh and M Ehsani, “Review of Sensorless Control methods in Switched Reluctance Motor Drives”, IEEE Industry Applications Conf., Oct 2000, Vol. 3, pp.1850-1857. 3. G S Buja and M I Valla, “Control Characteristics of the SRM Drives. Part-II: Operation in the Saturated Region”, IEEE Trans. on Indus. Electronics, Vol. 41, No. 3, 1994, pp. 313-321. 4. S K Mondal, S N Saxena and S N Bhadra, “A Novel Analog Method for Sensorless Closed-Loop Control of SRM without Parameter Estimation”, Conf. Records, IEEE Power Elec. Spec. Conf, May 1998, Vol. 2, pp. 2076-2082. 5. A E Fitzgerald, Charles Kingsley, Jr., and Stephen D Umans, “Electric Machinery”, 5th Metric Edition, 1992. 6. S K Panda and G Amaratunga, “Waveform Detection Technique for Indirect Rotor-Position Sensing of Switched-reluctance Motor Drives. Part 1:Analysis”, IEE Proc. B, Vol. 140, No. 1, January 1993, pp. 8088. II. SWITCHED RELUCTANCE MOTOR PARAMETERS Output Power - 4.0 kW Number of Phases - 4 Number of Stator Poles - 8 Number of Rotor Poles - 6 Aligned Inductance, Lmax - 110 mH Un-aligned Inductance, Lmin - 10 mH Stator Pole Arc - 200 Rotor Pole Arc - 21.50 Inertia, J - 0.002 kg/m2 Phase Voltage, Vdc - 280 V Nominal Phase Current - 9 A Phase Resistance - 0.7 ohm Rated Speed - 1500 rpm APPENDIX I. COEFFICIENT VALUES For the level of saturation generally encountered in the normal operating range of the motor, a 7 degree polynomial with 4 odd terms in equation (3) and 6 degree polynomial with 4 even terms in equation (4) are found to be adequate. Thus, the flux-linkage variation of the test motor with current and rotor position is described by the following polynomial equation. ψ(i,θ) = a7(θ)i7 + a5(θ)i5 + a3(θ)i3 + a1(θ)i (7) where, a7(θ) = 10-7(-0.1714sin63θ+0.0350sin43θ-0.0793sin23θ -0.0027) a5(θ) = 10-5(0.5540sin63θ+0.0589sin43θ+0.2840sin23θ +0.0018) a3(θ) = 10-3(-0.6821sin63θ-0.1097sin43θ-0.4909sin23θ +0.0130) a1(θ) = (-0.0401sin63θ+0.1043sin43θ+0.0442sin23θ +0.0099) in which, θ is the rotor position in radian with reference to the un-aligned position. Co-efficient ‘a1’ estimates the linear inductance value for any position. Thus, by (7), linear inductance values for unaligned (00) and aligned (300) positions are, respectively, 9.9 and 118.3 mH, which are in reasonable agreement with 10 and 110 mH, as available from manufacture’s data. 2080