Delay Time Compensation in the Current Control Loop of Servo Drives – Higher Bandwidth at no Trade-off Heiko Schmirgel Cologne University of Applied Sciences Betzdorfer Str. 2, 50679 Köln, Germany Heiko.Schmirgel@FH-Koeln.de, Tel.: +49 221/8275-2480, Fax.: +49 221/8275-2477 Jens Onno Krah Cologne University of Applied Sciences Betzdorfer Str. 2, 50679 Köln, Germany Jens_Onno.Krah@FH-Koeln.de, Tel.: +49 221/8275-2439, Fax.: +49 221/8275-2477 Reiner Berger Danaher Motion GmbH Wacholderstr. 40-42, 40489 Düsseldorf, Germany R.Berger@DanaherMotion.net, Tel.:+49 203/9979-191, Fax.: +49 203/9979-200 Abstract: Even after compensation of nonlinear phenomena in the power stage, delay and sampling times remain in the current control loop that can not be eliminated. These time lags limit the speed and possible performance of the current controller and thus of all superposed control algorithms. An effective method to compensate the time lag in the current control loop of digital servo amplifiers is presented. Based on the Smith Predictor concept, a model predictive control algorithm is implemented. Usability is ensured by reducing the algorithm to be dependent on very few parameters. The presented method provides a fast and simple solution implemented on the current control loop level of the cascaded drive control system. 1. Introduction High performance servo drives are still a fast growing market segment. The brushless technology offers significant advantages in terms of reliability and motor size. Higher drive complexity was balanced with additional logic and sophisticated power electronics. Due to the innovation cycles of the semiconductor manufacturers the size and cost for more and more complex drive systems did not increase. Each new drive generation offers more performance, more flexibility and higher integration density. The challenge is to provide the best possible performance/cost ratio with the given setup. Better performance will allow more sophisticated processes and lower cost makes these technologies attractive to new applications. While the hardware setup remains essentially the same throughout a single generation, better algorithms have to be developed and constantly improved in order to make best use of the limited available computing power. In a cost optimized servo control environment, the crucial requirement is always to provide real-time performance. This demands simple and fast algorithms, still providing the maximum possible accuracy and dynamics. Phase lags caused by dead times inside of control loops reduce the stability margin and limit possible gains to lower values and thus the dynamics of the complete system. To reduce the effect of time delays in the current loop of the drive system the implementation of an algorithm known as Smith Predictor [1] inside the current control loop is proposed. The Smith Predictor offers the possibility to make use of known system properties in order to predict the reaction of the controlled system before it can be measured, which substantially improves performance of the drive system. It can be entirely implemented only by changing the current controller software without the need for additional hardware. These improvements offer the possibility of designing a drive system with faster response while avoiding the well known trade-offs of higher switching losses and increased EMI (electromagnetic interference), that require the system to be bigger and more expensive. The algorithm is implemented on a drive system with 16kHz switching frequency and 32kHz current loop PCIM EUROPE 2006 • PROCEEDINGS • 541 take an infinitely small amount of time, almost PWM every element inside the current control loop Velocity Feedback i Current Feedback contributes to the time d/dt delays. Some elements are even only used at Position Feedback Motor R certain time instants which means that they “wait” until it is “their Fig. 1: General cascaded drive system turn” (Fig. 2). update, and it is shown that a closed loop current Usually the acquisition of the current feed back signal control loop bandwidth of 5kHz is achievable. is such a process. It takes place during one or both of the zero-sequences (all outlets on low or all outlets on high potential) that are generated by the PWM2. Control Structure algorithm. At these instants of time when no (Circuit, Scheme) switching takes place noise is low and the current assumes the average value without additional Commonly, the control architecture of drive systems filtering that would otherwise cause additional delay consists of three cascaded control loops. The as well. Together with the time consumed by Analoginnermost loop is the fastest, controlling the current. to-Digital Conversion this makes up the time delay Directly superposed to the current loop is the velocity caused by the feed back system (TFB). Faster A/D controller. The outer loop is the slowest controlling hardware can reduce the latter contribution but far the position (Fig. 1). more dominant is the sampling process here. In each loop the command signal is not processed The time lost due to current control algorithms and instantaneously, but handed over to the next loop or calculation of the PWM signals from the voltage element inside the same loop with a certain time command values (TPI) can be limited by using delay. Computation times and other time lags (e.g. efficient programming techniques. More important is due to feedback) can only be reduced to a certain the time delay acquired during the active phase, when extent. Ultimately a delay time always remains. the desired current is actually generated by the power This has a limiting effect on the overall system stage (TPWM and TPS). This takes naturally a complete performance, where the most adverse effect on system interrupt cycle. The voltage drop across the dynamics has the time lag in the current loop. Since semiconductor switches is here assumed to be this is the fastest loop, which is executed with a compensated by an algorithm equivalent to [2] or higher frequency than all the other loops. small enough to be neglected. Reducing the effect of the time lag in the current loop In order to reduce any one of the time delays they and thus increasing its dynamic range will have a have to be looked at individually giving attention to positive effect on all superposed loops. the respective nature and source of the effect. The compensation with a Smith Predictor is done on a system level. Here all delay times can be replaced by 3. Sources of Delay Times one single delay time Td and then be treated together (Fig. 3). Since the performance of any computation operation The sample and hold process contributes a dead time with a digital processor or microcontroller does not of one half of a sampling period (Ta/2). As long as the individual delays of the computation process are Td PWM Td PS Ui PI-Controller Td PI L, R small, which is usually Power the case, they can be PWM Stage treated together with the delay of the active phase, which adds up to a Td FB complete sampling A/D period (T ). a Sampling Position Command Position Loop Velocity i* Loop Current Loop u* Power Stage Conversion Fig. 2: Delay times inside the current loop PCIM EUROPE 2006 • PROCEEDINGS • 542 required to be able to predict its reaction. In this case this is the LR-model of the machine windings and the applicable dead time of Td the complete system Td. With this information, it is Sample Power ADC PWM Controller possible to make the predictable + Hold Stage current signal (the effect of the voltage command on the L-RTa model) available to the feed back loop before the delay time. This is 2 Ta done by adding the predicted current to the feed back signal Fig. 3: All delay times can be represented by one single dead time without a delay. In the predictor branch of the current control loop, the L-R first order lag and the PI-Controller Td L, R Ui delay time have to be swapped a) (Fig. 4b). The feed back loop then “sees” the predicted value of the current at the same time, when it is actually generated. In digital motion control systems this is Td L, R easily implemented. b) The predicted signal has to be subtracted from the feed back ipre signal again after the delay time has elapsed, because at that time Fig. 4: a) Current control loop without Smith Predictor and b) Basic the regular feed back signal Smith Predictor concept contains the information of the With only one remaining dead time, the real effect of the voltage command. representation of the current control loop including all This leads to the complete current control loop shown the time delays simplifies to (Fig. 4a). in (Fig. 5) where the Smith Predictor branch is The machine windings are modeled as a simple L-R implemented according to the transfer function (1). first order lag. 1 (1) (1 − e − sT ) 1 + sTLR The second part of (1) is the pulse function, described 4. The Prediction Algorithm in [1]. Active Phase Information Processing d The idea behind the prediction algorithm is to create a system that can be controlled as if it was its own minimum phase equivalent (i.e. the same system without any dead times). Some knowledge about the controlled system is Td PI-Controller - Ui ipre For the following considerations the current controller is assumed to be a general regular PI-controller with a proportional gain factor Kp and L, R integration time constant TN(2). GPI ( s ) = Td L, R 5. Effect on System Performance - K p (1 + sTN ) (2) sTN The system model that consists of the dead time Td and the L-R first order lag with its time constant TLR is added and the control loop is closed. This leads to an overall system transfer function before Fig. 5: Current control loop with Smith Predictor PCIM EUROPE 2006 • PROCEEDINGS • 543 the Smith Predictor was added (Fig. 4 a)) as in (3) 1 sTN 1 + sTLR GS = K (1 + sTN ) − sTd 1 e 1 + p sTN 1 + sTLR (3) GSm = sTN e − sTd -6 dB 6 dB -10 dB 10 dB 1 1 + sTLR ( -1 and (4) can be simplified to (6) (6) 1 sTN e − sTd K p (1 + sTN ) 1 1 + sTLR 1 + 1 + sTLR sTN (6) represents a system where the dead time is completely outside of the feed back loop (Fig. 6) [3] and thus does not affect system dynamics at all. The relevant part of the system for the PI-controller is in fact the minimum phase equivalent as desired. The Nyquist plot of this system illustrates the effect of the added prediction algorithm. The well tuned regular PI-controlled system with dead time (Fig. 7, red curve) shows the typical characteristics of a process with dead time. The Smith Predictor with exactly matching parameters (dead time and motor time constant) eliminates the dead time completely, in accordance with (6) only the first order time lag characteristic of the machine is left that affects the system dynamics (Fig. 7 green curve). PI-Controller L, R -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 Real Axis (5) K p (1 + sTN ) -20 dB -1 ) Td = TdS 20 dB 0 -0.5 (4) If the model is sufficiently accurate, (5) holds GSm = -4 dB 0.5 K (1 + sTN ) − sTd 1 1 1 − e − sTdS 1 + p + e sT sT sT 1 1 + + N LR S TLR = TS -2 dB 4 dB The addition of the Smith Predictor (Fig. 5) with its own approximations of the dead time TdS and the time constant of the machine TLRS, changes only the feed back which leads to (4) K p (1 + sTN ) 0 dB 1 e − sTd Imaginary Axis K p (1 + sTN ) Nyquist Diagram 2 dB Fig. 7: Nyquist plot of the system without Smith Predictor (red), with Smith Predictor and ideal parameters and the same gain (green) and increased gain (blue). The gain of the PI-controller can now be increased to much higher levels than before, without impairing the stability of the system (Fig.7 blue curve). For a true first order time lag and correctly estimated parameters the gain could theoretically be infinite. Since the first order time lag is only an approximation and small estimation errors in the parameters are not avoidable, there are certain limits to the increase of the controller gain. The effect of estimation errors of 15% for the dead time and 15% for the time constant of the Smith Predictor is shown in Fig. 8 (red curve) compared to the ideally tuned Smith Predictor (blue curve). This system has far less stability margin than the ideally tuned Smith Predictor but it is still superior to the system without prediction algorithm since it has a greatly increased proportional gain and thus higher dynamics. Additionally the system will also be robust enough to cope with parameters that are not precisely tuned which will most probably happen once a machine heats up during operation. Td L, R Fig. 6: Resulting system PCIM EUROPE 2006 • PROCEEDINGS • 544 Nyquist Diagram 7. Automatic Parameter Acquisition 0 dB 2.5 -2 dB 2 1.5 1 Imaginary Axis 0.5 2 dB -4 dB 4 dB -6 dB 6 dB -10 dB 10 dB 20 dB -20 dB 0 -0.5 -1 -1.5 -2 -2.5 -1 -0.5 0 0.5 1 Real Axis Fig. 8: Nyquist plot of the system with Smith Predictor and ideal parameters (blue) and 15% higher dead time and 15% longer time constant (red). The data on which the parameterization of the Smith Predictor is based is taken from machine data bases or measured manually. Ultimately only few parameters remain to be tuned that only depend on the connected motor. Additionally, all the measurements needed are already integrated within a normal drive system, and tuning effort can be limited by introducing certain bounds to the possible results. Necessary security to instability problems during the tuning process is provided by the Smith Predictor gain that can be used as a limiter as well. An automated process that acquires the parameters of the correction function by itself during startup will be implemented on the same system, to enable the drive controller to quickly adjust to a number of different motors. 8. Test Setup 6. Implementation The proposed algorithm was implemented and tested on a conventional industrial drive system. (Servostar 303, 8kHz switching frequency and 16 kHz current loop update and 16 kHz switching frequency and 32 kHz current loop update with PM machine 6SM47L3000) and tested against the same system without prediction algorithm [4]. Due to the simplicity of the new elements additional calculation time is low enough not to have any effect on system performance. In the implementation special attention was given to usability and simplicity. Since the dead time depends purely on the inverter system and its information processing unit it can be assumed to remain constant, even when the controlled machine changes. Experimental results show that this assumption is valid and no adjustment by the user is necessary. The parameters left to tune are the resistance and the inductance of the machine model. These values should be readily available from the data of the machine or can be easily 20dB open loop measured by the user. These parameters can only be acquired 10 with a certain accuracy and especially 3 closed loop 0dB the resistance changes when the -3 machine temperature increases. 100Hz 1kHz 4kHz Experiments show that according to the Fig. 8 the algorithm is indeed robust 0° against minor inaccuracies of these closed loop parameters. open loop -90° Additionally, a Smith Predictor gain is introduced that can adjust the effect of the Smith Predictor during the tuning -180° process. With these few parameters and the 100Hz 1kHz 4kHz possibility to limit the smith predictor gain during tuning procedure, it is possible to quickly tune the algorithm. Fig. 9: Bode plots of the current loop at 8 kHz switching frequency without (red) and with smith predictor PCIM EUROPE 2006 • PROCEEDINGS • 545 40dB open loop 20 10 6 0dB -6 100Hz closed loop 1kHz 0° 7kHz It can eliminate the time delay for tunability of the PI-controller almost completely, allowing a substantially increased proportional gain. Bode plots show the wider dynamic range of the current loop with Smith Predictor. The design of a drive system with 16 kHz switching frequency and 32 kHz current loop update, that has a closed loop current control loop bandwidth of 5 kHz was demonstrated. closed loop 10. Summary / Conclusion -90° -180° 100Hz 1kHz 7kHz Fig. 10: Bode plots of the current loop at 16 kHz switching frequency without (red) and with smith predictor Fig. 9 shows the open loop and closed loop bode plots of the system at 8kHz switching frequency and 16 kHz current loop update without (red curves) and with the Smith Predictor (blue curves). It can be clearly seen that the Smith Predictor increases the dynamic range of the current loop in this low switching frequency mode by 70% (closed loop bandwidth at -90° phase angle). Fig. 10 shows the open loop and closed loop bode plots of the system at 16 kHz switching frequency and 32kHz current loop update without (red curves) and with the Smith Predictor (blue curves). In the high switching frequency mode, the increase of the dynamic range is still more than 30%. The -90° phase angle bandwidth can now be increased to over 5 kHz. Acknowledgement All implementation and testing of algorithms was done on a Servostar 300. This servo drive system as well as the complete funding of this project was kindly provided by Danaher Motion GmbH, Düsseldorf, Germany. References 1. 2. 9. Discussion of the Results The Smith Predictor, while originally designed for much slower processes, proves to be able to compensate even small dead times in current control loops. These improvements offer the possibility of designing a drive system with much faster response while avoiding the well known trade-offs of higher switching losses and increased EMI (electromagnetic interference), that require the system to be bigger and more expensive. 3. 4. Smith, “A Controller to Overcome Dead Time”, ISA Journal, vol.6. no.2, pp.28-33, Feb. 1959 Schmirgel, Krah, “Compensation of Nonlinearities in the IGBT Power Stage of Servo Amplifiers through Feed Forward Control in the Current Loop” Proceedings of the PCIM Europe, pp. 94-99, Nürnberg, June 2005 Veronesi, “Performance Improvement of Smith Predictor through automatic computation of dead time”, Yokogawa Technical Report English Edition, no. 35, pp.25-30, 2003 SR 300 manual, www.DanaherMotion.net PCIM EUROPE 2006 • PROCEEDINGS • 546