JOURNAL OF THEORETICAL AND APPLIED MECHANICS 43, 3, pp. 631-653, Warsaw 2005 REAL-TIME CONTROL FOR A MAGNETORHEOLOGICAL SHOCK ABSORBER IN A DRIVER SEAT Bogdan Sapiński Department of Process Control, University of Science and Technology, Cracow e-mail: deep@agh.edu.pl The paper summarizes the author’s research study on real-time control of a magnetorheological shock absorber (MRA) in a driver seat. The performance of theMRA for vibration and shock isolation of the driver seat was investigated experimentally in open loop and feedback system configurations. Real-time controllers forMRAswith on-off and continu- ously variable control schemes were developed in the integrated design and control environment of MATLAB/Simulink. The sensors used in experimentswere tested to see how reconstructed velocity signals should affect output signals of the controllers to theMRA. Key words: magnetorheological shock absorber, driver seat, vibrations, shocks, sensor, real-time controller 1. Introduction Professional drivers spenda great deal of timebehind thewheelwhere they are exposed to vibrations and shocks when their vehicles encounter irregulari- ties of road conditions. The undesired inputs are transmitted to the driver by vehicle sub-systems, when the seat suspension runs out of travel. The truck drivers call that phenomenon ”topping” and ”bottoming”. The topping can injure a driverwhile the bottoming can lead to loss of control of a vehicle. The risk of the occurrence of that phenomena increases particularly for driverswho adjust the seat height away from the center of the travel. To measure the vi- bration exposure, the International StandardOrganization (ISO) recommends a method called the Vibration Dose Value (VDV). In accordance with that method, the vibration exposure involves numerous shock events. It is well known that the human body is most sensitive to seat vibrations in the range of (4,8)Hz, so the seats are designed to effectively isolate the 632 B.Sapiński driver from vibrations in this range. In on-highway trucks, most suspensions of conventional driver seats are passive as they employ an air-ride suspension andapassive damper to vibro-isolate thedriver.Those seat suspensiondesigns always sacrifice certain degree of either vibration or shock isolation. Such seats are too soft to prevent the driver from the topping and bottoming, and even elastomer snubbers or stiffened springs and/or dampers in seat designs do not provide adequate protection for the driver. For this reason, some driver seats are equipped with semi-active or active suspensions. That allows the drivers to feel more comfortable and less fatigued, and besides gives them better protection. A smart semi-active solution for the driver seat suspensions is called the Motion Master R© Ride Management System (MMS) (http://www.lord.com). The MMS is intended for seat suspensions in trucks, buses and tractors and is used in pupil transportation, trucking and transient industries.Main featu- res of the MMS are: the smoothest air ride without topping and bottoming, automatic vibration and shock control (500times/sec), prevention end-stop collisions – regardless of driver weight. Results of experiments using a proto- type of the MMS revealed that the overall performance of the vibration dose decreased by up to 40%, depending of the seat height. TheMMS includes three components: anMRA containing Lord Corpora- tion’s patented RheonticTM MR fluid, controller that continuously monitors seatmotion anddetermines the optimal damping force, ridemode switch, allo- wing one to choose among soft, mediumor firm settings, following the driver’s preference. The system is powered directly from the common 12-volt automo- tive source. The force produced by theMRA in theMMS is controlled by the use of a controller which adjusts the current driver connected to the MRA coil to the current operating conditions. The controller can be programmed in accordance with an assumed control scheme. Control methods in semi-active suspensions are classified into on-off and continuous categories (Ahmadian, 1999). The methods made use of velocity signals from suspension components. The category of on-offmethods involves the switching of the suspension system from the minimal (on) to maximal (off) positions which correspond to the minimal andmaximal damping states for the MRA. One of these methods was patented for an MRA used in seat supports (U.S. Patent 5,712, 783, 1998). This is a modified-skyhook method that enables simple, inexpensive hardware to be used and actually outper- forms skyhook theoretical control. In order to avoid the harsh feel of end stop collision, end stop limits are suggested to be used to increase the damping for- ce when the damper is about the bottom or top position. Continuous control Real-time control for a magnetorheological shock absorber... 633 methods allows us to enhance the number of switching levels of damping for the suspension as a continuously variable damping coefficientmay be achieved (Sapiński, 2005). In this study, we present examination of the problem how theMRA could protect the driver from vibrations and shocks in a systemwith the open loop configuration (for constant levels of current applied to theMRAcoil) andwith the feedback configuration (current levels adjusted by real-time controllers). The real-time operation means here an ability of an MRA control system to follow up the changes in damping of vibration and shock. The experiments were run in a laboratory setup equipped with an adopted driver seat equip- ped with a commercial MRA of RD-1005-3 series (http://www.lord.com), a current driver engineered for the purpose of the author’s research program and data acquisition and control system based on a PC with a multipurpose I/O board, operating in MS-Windows through the MATLAB/Simulink envi- ronment. For the purpose of measurement, we used two transformer linear displacement transmitters. Experiments for the driver seat were proceeded by testing the sensors that could be used in the measurement system of the la- boratory setup for the driver seat. The aim of testing of the sensors was to see how reconstructed velocity signals should affect the output signals of the developed controllers to theMRA (Sapiński and Rosół, 2004). 2. Model of a driver seat with controllable damping The model of a driver seat is shown in Fig.1. This is a two-degree of freedom (2DOF) system with the following designations: k1 – stiffness of the suspension system; cr – damping constant of the suspension system, m1 – suspension mass (mass of the seat with the cushion); k2 – stiffness of the cushion; c2 – damping constant of the cushion; m2 – bodymass (mass of the driver); x0 –displacement inputexcitation; x1 – seatdisplacement; x2 –driver displacement. Let us assume that the seat cushion is removed (k2 = 0, c2 = 0) and denote the total mass of the driver and the seat with no cushion by m. That yields a model of the driver seat with a 1DOF system. Taking into account that x10 is the static position of the seat, l1 – initial length of the suspension spring, we obtain the following equation of motion for the seat-driver system (dry friction is neglected) m1ẍ1+ crẋ1+k1x1 = crẋ0+k1x0 (2.1) 634 B.Sapiński Fig. 1. The model of a driver seat with the initial condition x10 = x1(t) = l1 −mg/k1, (g – acceleration of gravity). Equation (2.1) corresponds to the transfer function G(s)= X1(s) X0(s) = crs+k1 m1s2+ crs+k1 (2.2) Assuming that k1 = 36861N/m, m = 112kg, the undamped natural fre- quency of the system f0 is about 2.9Hz and the cross-over frequency is fc = √ 2f0 = 4.1Hz (fc is such a frequency that displacement transmissi- bility is X1(s)/X2(s)= 1). In Fig.2 we show the acceleration transmissibility of a 1DOF system obtained for the above values of k1 and m and values of cr corresponding to the following current levels in the MRA coil (I): 0.00A, 0.05A, 0.10A, 0.15A. It is readily seen that the resonance frequency of the systemwith no current is about 3.1Hz and it increases with the current level. Simultaneously, it is apparent that rapiddampingof free vibrations is provided by the value of current in the range (0.05,0.15)A. Fig. 2. Acceleration transmissibility for the driver seat Real-time control for a magnetorheological shock absorber... 635 3. Magnetorheological shock absorber The RD-1005-3 is a small and compact MRA with simple electronics, low voltage and current demands that enables real-time damping adjustment (Fig.3). ThisMRAhas ±25mm stroke. The input voltage is 12VDCand in- put current can be varied in the range (0,2)A. The response time (dependent on an amplifier and power supply) is less than 25ms (time to reach 90% of maximun level during a 0A to 1A step input at the velocity of 51mm/s). Fig. 3. The RD-1005-3 – a general view The advantage of theMRA is associated with the capability of continuous modificationofdampingcharacteristics over the controllable range.This iswell seen when we consider the RD-1005-3 family performance curves determined experimentally for the assumed current and velocity ranges (see Fig.4). These performance curves illustrate a relationship between the force producedby the MRA (output) and shaft velocity (disturbance) that can be adjusted by the magnetic field (control) induced by the applied current. Fig. 4. Performance curves for the RD-1005-3 4. Sensors A schematic depiction of the driver seat-MRA system inwhich the control method for a controllable damper patented in (U.S. Patent 5,712, 783, 1998) 636 B.Sapiński can be implemented, reveals that the system requires one accelerometer and one displacement sensor. It was mentioned that the controllers for the MRA made use of velocity signals from the suspension components of the driver se- at. Thatmeans that the velocity signals have to be reconstructed from output signals of the sensors above. In order to see the reconstructed velocity signals, the investigations for the driver seat-MRA systemwere realised by experimen- tal testing for available sensors which can be employed in the measurement system of the laboratory setup for the driver seat. In the tests, we used a linear displacement transmitter of PSz20 series (http://www.peltron.home.pl, 2004) and an accelerometer of ADxL210 se- ries (Analog Devices, 1999). The PSz20 is based on a differential transfor- mer, placed in a cylindrical housing together with an electronic system that can be employed in static and dynamic measurements of the displacements. The ADxL210 is a high-performance 2-axis integrated accelerometer that can measure both dynamic acceleration (e.g., vibration) and static acceleration (e.g., gravity). It produces digital outputs whose duty cycles (ratio of pulse width to period) are proportional to the acceleration in each of the 2 axes. Basic technical specifications for the PSz20 and ADxL210 are provided in Table 1. Table 1.Technical specifications for the PSz20 and ADxL210 Parameter Value PSz20 ADxL210 Measurement ±10 ·10−3m ±10g range Power supply ±15VDC +3, +5.25VDC Output signal ±5V DC PWM (frequency depends on an external resistor) Pass band 3dB, 50Hz 3dB, 500Hz (PWM out) 3dB, 5000Hz (analog out) Non-linearity ¬ 0.5% 0.2% Operating −20 . . .+70◦C 0 . . .+70◦C temperature Shock survival 100g, 11 ·10−3ms 1000g ThePSz20 andADxL210 sensors (see Fig.5)were tested in the experimen- tal setup shown schematically in Fig.6. The setup comprises: electro-dynamic shaker, power amplifier (for shaker control), PC with a multi I/O board of RT-DAC4 series (Inteco Ltd., 2002) operating in the system ofWindows 2000 Real-time control for a magnetorheological shock absorber... 637 through theMATLAB/Simulink (for acquisition and analysis ofmeasurement data). Fig. 5. The PSz20 and ADxL210 in the experimental setup – ready for tests Fig. 6. A diagram of the experimental setup for testing of the PSz20 and ADxL210 The tests were performed for sine excitations with the frequency (1,10)Hz and amplitude 3 ·10−3m.Basing on a dulymeasured output voltage signal of the PSz20, velocity and acceleration signals were reconstructed. Similarly, the velocity and displacement signals were reconstructed from the output voltage signal of the ADxL210. Themethods used for the signals reconstruction were discussed in (Sapiński and Rosół, 2004). Some selected results are presented below to see the reconstructed velocity signals. In Fig.7 we show velocity signals which were reconstructed from displa- cement and acceleration signals measured by the PSz20 and ADxL210 at the frequency 5Hz. It is readily apparent that the reconstruction was quite cor- rect andnophase shifts or transientswere observed.The appearingdistortions 638 B.Sapiński come as a result of numerical differentiating of the displacement signal, a pro- cedure available in theMATLAB/Simulink. Note that the methods involving the reconstruction of velocity signals from theADxL210 require a certain time for transient states to stabilize. This time period gets longer with increased excitation frequency. The length of this time period is affected by parameters of a low-pass filter. Fig. 7. Reconstructed velocity signals from the PSz20 and ADxL210 When a displacement signal was reconstructed from theADxL210 output, the integrating circuits cause that the signal stabilization takes longer. That is illustrated for a sine excitation of 5Hz in Fig.8. It is worthwhile tomention that the reconstruction of the displacement signal from the ADxL210 output brings about the loss of vital information about the constant component of the displacement. Fig. 8. A reconstructed displacement signal from the ADxL210 compared with the PSz20 When a acceleration signal was reconstructed from the PSz20, major di- stortions appeared as a result of the use of a double differentiation procedure (available in theMATLAB/Simulink) and a low-pass filter. That is shown for a sine excitation of 5Hz in Fig.9. The observed signal distortions might be Real-time control for a magnetorheological shock absorber... 639 reduced by providing low-pass filters or by applying themethod of signal sam- ple averaging on the basis of neighbouring samples used in the reconstruction procedure (Sapiński andRosół, 2004). An increase in the excitation frequency improves the quality of the signal reconstructed from the PSz20. Fig. 9. A reconstructed acceleration signal from the PSz20 compared with the ADxL210 It is reasonable to assume that in the experimentswith a driver seat-MRA system with the feedback configuration, two PSz20 sensors would be used. That means that the input signals of the controllers to be developed (e.g. absolute seat velocity and relative velocity – difference between seat velocity and shaker-base velocity) will be reconstructed basing on signal outputs of the PSz20 sensors. In the case of anon-off controller, the output signal (the current in the MRA coil) causes the switching between on and off damping states, dependingon the sign of the velocity product (e.g. product of absolute velocity and relative velocity). For the sake of illustration, we compared the signals of reconstructed seat velocity (Fig.10), velocity product (Fig.11) and current in theMRA coil (Fig.12) obtained by the use of PSZ20 andADxL210 for a sine excitation (frequency 5Hz, amplitude 1.6 ·10−3m). Note that the maximum current level at the on-off controller output was kept 0.10A throughout. When analysing plots in Fig.10, it is readily seen that the signal of seat velocity reconstructed from the ADxL210 output stabilized after about 2s. The transients are responsible for erroneous calculation of the initial velocity product (Fig.11) and current (Fig.12a). For time periods in excess of 2s, currents produced on the basis of reconstructed velocity signals obtained from PSz20 and ADxL210 are similar (Fig.12b). The differences in reconstructed velocity signal patterns from ADxL210 and PSz20 are attributable to the method of hardware processing of signals from the sensors. These differences become more marked as the frequency increases. 640 B.Sapiński Fig. 10. Reconstructed seat velocity Fig. 11. Velocity product Fig. 12. Current in theMRA coil: (a) range (0.0,0.5)s, (b) range (2.0,2.5)s Real-time control for a magnetorheological shock absorber... 641 Note that at the current stage of experiments, the ADxL210 was used on- ly to compare the signals of measured acceleration with those of acceleration reconstructed from displacements measured by the PSz20 (i.e. it was not em- ployed any more in further experiments conducted for the driver seat-MRA feedback system configuration). 5. Controllers Among a variety of design approaches to controllers for the MRA in a driver seat support, we present three real-time controllers developed in the integrated environment for design and control of MATLAB/Simulink. 5.1. Control methods The structure of the developed controllers is shown in Fig.13. The input signals are the seat velocity ẋ1 and relative velocity (difference between seat and shaker base velocities), (ẋ1 − ẋ0), while the output signal is the current in the MRA coil (I). Fig. 13. The structure of controllers Let us assume that the controllers are denoted by CON1, CON2, CON3 and governed by following formulas CON1: I = { c1 for ẋ1(ẋ1− ẋ0)­ 0 0 for ẋ1(ẋ1− ẋ0)< 0 CON2: I = { c2|ẋ1− ẋ0| for ẋ1(ẋ1− ẋ0)­ 0 0 for ẋ1(ẋ1− ẋ0)< 0 (5.1) CON3: I = { c3(t)|ẋ1− ẋ0| for ẋ1(ẋ1− ẋ0)­ 0 0 for ẋ1(ẋ1− ẋ0)< 0 where 642 B.Sapiński ẋ0, ẋ1 – base and seat velocity, respectively ẋ1− ẋ0 – relative velocity c1,c2 – constants c3(t)= c ∗ 3(t)=G3|ẋ1| – continuously variable factor G3 – gain factor. Note that the values of c1, c2 and G3 depend on the maximum level of current applied to the MRA coil. It is readily seen that CON1 is an on-off controller (i.e. it involves the switching between the minimum and maximum damping levels) while CON2 and CON3 are continuous controllers (i.e. the number of damping levels is greater, as continuously variable damping coefficients may be obtained). 5.2. Integrated environment for design and control Real-time controllers for the MRA were implemented in the integrated designandcontrol environment including the following hardwareand software: • PC (Pentium III/1GHz ) with a multi I/O board of RT-DAC4 series • operating system MSWindows 2000 • MATLAB/Simulink (version 6.5) • Real Time Workshop (RTW) toolbox in MATLAB/Simulink with the extension Real TimeWindows Target (RTWT). The MATLAB/Simulink was used to support design of the controllers. The toolbox RTW extends potential applications of the MATLAB/Simulink to control by providing the path of ”rapid prototyping” (Grega, 1999). That allows real time implementation of control algorithms directly from Simulink. Unfortunately, the toolboxRTWisnot capable of generating real-time tasks in the MS Windows environment, that is why the integrated design and control environment is supported by theRTWTsoftware. Communication procedures featured byRTWTallow compilation of anRTWcode and admit its real-time operation inMSWindows on a specified hardware platform. 5.3. Automatic code generation A block diagram of subsequent stages of executive file development using RTWT is shown inFig.14. It is based on the Simulinkmodel of the controller. Block designations are provided below. The blockmodel.m includes the Simulinkmodel of the controller. It conta- ins input drivers (being the source of input data for the controller) and output drivers providing for actuators control. The Simulink model of a controller Real-time control for a magnetorheological shock absorber... 643 Fig. 14. Use of the RTW toolbox for controller prototyping inMS-Windows was used to automatically generate a C-code, which was then preprocessed and compiled. At the stage of code compilation, an executive file was genera- ted. The file is called up and started as a result of clock interrupts operated by the real-time system kernel. An executive filemight be connected with the Simulink environment as long as it is started in the external mode (TheMath Works, 2003). Itmight be also connected to virtual elements which enable the tuning of task parameters and signal acquisition andmonitoring. Note that the blocks in the diagram coloured bright grey represent real- time tasks, while those coloured dark grey – on-line tasks. The black thick lines connecting the blocks illustrate the flow of information and command signals within the system. Amodel of a Simulink controller is shown inFig.15. It is controller CON1, implementing on-off control in accordance with formula (5.1)1. Integrated Si- mulink blocks are used as controller blocks. Measurement data (shaker base displacements and seat displacements) from the input driver (blockRT-DAC4 Analog inputs) supporting A/C converters on the board RT-DAC4 are trans- formed to the shape required by the controller algorithm. The control signal for theMRA is sent to the output drivers (blockRT-DAC4PWM0), following conversion of the signal from the controller (block I/PWM) and taking into account the constraints upon the maximal current in the MRA coil (block Saturation 1). Besides, the parameters of signals applied to the shaker can 644 B.Sapiński be controlled too, using the output driver (block RT-DAC4 Analog outputs) controlling the C/A converters in the RT-DAC4 board. The maximal signal constraint is taken into account in block Saturation 2. Block Scope is used for data acquisition andmonitoring. Fig. 15. Controller CON1 ready for interaction with the RTW toolbox 6. Experiments The driver seat-MRA system was experimentally tested in open loop and feedback system configurations under harmonic and shock excitations. 6.1. Experimental setup Adiagramof the experimental setupused for testing of real-time controllers is depicted in Fig.16. The electro-hydraulic shaker was supplied via a hydraulic pump and controlled from a control cubicle. Input-output data were acquired using a data acquisition and control system based on a PC (Pentium III/1 Real-time control for a magnetorheological shock absorber... 645 GHz) with the multi PCI I/O board of RT-DAC4 series operating in the software environment of Windows 2000, MATLAB/Simulink and RTW with RTWT. The seat with no cushion (equipped with the MRA of RD-1005-3 series and a designed spring) to be tested is shown in Fig.17. Fig. 16. A diagram of the experimental setup Fig. 17. The driver seat in the experimental setup The total mass of the driver and seat with no cushion was 112kg and the spring constant was 36861N/m. Displacements of the shaker-base and driver seatweremeasuredby twoPSz20 sensors.The input signals for the controllers 646 B.Sapiński were the seat velocity and relative velocity, while the output signal was the current in the MRA coil. 6.2. Performance testing for vibration isolation At the first stage, the open loop system was investigated. Driver seat re- sponses were measured under sine excitations of the shaker base with the amplitude 1.5 ·10−3m in the frequency range (1,10)Hz for the following cur- rent levels in theMRAcoil: 0.00A, 0.05A, 0.10A, 0.15A.Theobtained results are shown in Fig.18 in the form of acceleration transmissibility. Fig. 18. Acceleration transmissibility in the open loop system Fig. 19. Seat acceleration in the open loop system Figure 18 shows that as the level of current increases, the resonance fre- quency of the system will increase too, and the frequency at which vibration control of the seat is most effective ranges from 3Hz to 5Hz. Fig.19 shows time variations of the seat acceleration in response to the sine excitation with the frequency 5Hz (e.g. near-resonance frequency of the investigated system). Note that the current was set as 0.00A. It is clearly seen that the system enhanced the input signal and shifted the phase. Real-time control for a magnetorheological shock absorber... 647 At the second stage, feedback system configurations with CON1, CON2 and CON3 were tested under the same sine excitations as for the open loop system. The sampling rate for real time tasks was 0.001s. The constants as- sumed in control schemes were as follows: for CON1 – c1 = 0.10, CON2 – c2 = 60, CON3 – G = 9000. These values would yield the maximum current level which was assumed to be 0.10A. Selected results of the experiments are presented in the frequency and time domain (see Fig.20 and Fig.21). Fig. 20. Acceleration transmissibility in the open loop and feedback system configurations In Fig.20, the acceleration transmissibility for each feedback system confi- guration is plotted and comparison is made to the open loop system. A close survey of the plots reveals that the best performance was achieved in the feedback system with CON2 (Sapiński, 2004). Figure 21 shows time patterns of the current, velocity product and seat velocity for feedback system configurations with CON1, CON2 and CON3 at the frequency 5Hz. These results confirm the operating principle of the developed controllers. In the case of CON1, the current was switched between two values, i.e. 0.00A and 0.10A, while in the case of CON2 and CON3 the current may assume any value from the range (0.00,0.10)A. Note that an undesirable phenomenonwas observed in all tested feedback system configurations. It is known as the chattering effect (currentwas produ- ced in states when the velocity product oscillated around zero value), however it seems to be predominant in the feedback system with CON1. 648 B.Sapiński Fig. 21. Current, velocity product, seat velocity in feedback system configurations To illustrate the effectiveness ofCON1,CON2andCON3,we showzoomed sections of time patterns for driver seat acceleration in open loop and feedback system configurations (see Fig.22). 6.3. Performance testing for shock isolation At this point, we present results of tests on the driver seat-MRA system under roundedpulse shocks.The roundedpulse shock is analytically expressed by the formula x0(t)=X0 e2 4 γωnte −γωnt (6.1) where γ is a parameter expressing time of pulse duration in relation to the Real-time control for a magnetorheological shock absorber... 649 Fig. 22. Zoomed sections of seat acceleration in open loop and feedback system configurations half-period of natural system vibrations. The parameter γ is given by the formula γ= T 2τ = π ωnτ (6.2) where τ – duration of a square impulse with the area equal to that of the rounded pulse ωn – pulsation of natural vibrations of the system X0 – rounded pulse amplitude. The chief advantage of the rounded pulse excitation is that its first and second derivative assume limited values for all time instants t. The desired amplitude of a rounded pulse was set to be X0 = 2.57 · 10−3m (Liu et al., 2002). After rescaling associated with signal passing through aC/A converter of RT-DAC4 and amplifiers of the control cubicle of the shaker, themaximum value of the rounded pulse was 1.75 ·10−3m. InFig.23 plots of pulses applied in experiments for various values of γ are shown. Results of experiments conducted in the open loop system configuration for the rounded pulse excitation applied to the shaker-base are depicted in Fig.24 and Fig.25. It appears that when the value of γ was lower (e.g. time of rounded pulse duration got longer), the maximum value of driver seat acceleration was gre- ater (see Fig.24). Moreover, time required for the system to reach the steady state got shorter. Similarly, when analyzing plots in Fig.25, we see that as the current level increased, the maximal value of driver seat acceleration decre- ased. Throughout the investigated range of the current there were no changes in the time required to reach the steady state. 650 B.Sapiński Fig. 23. Rounded pulse excitation for various values of γ Fig. 24. Seat acceleration in response to the rounded pulse for various values of γ and the current 0.00A Fig. 25. Seat acceleration in response to the rounded pulse for various current levels at γ=1 Selected results of experiments conducted in feedback system configura- tions with CON1 and CON2, and compared with those achieved in the open loop system are shown for two different values of γ in Fig.26. Real-time control for a magnetorheological shock absorber... 651 Fig. 26. Seat acceleration in response to the rounded pulse for: (a) γ=1, (b) γ=3 It is readily apparent that those feedback systems did not provide any reduction to the seat accelerationwhencompared to the open loop system, and neither did the system equipped with CON3. That may be explained by the fact that the system response to current changes was prolonged. The reasons for such a state of affairs might be as follows: nonzero time of MRA force stabilisation, magnetic residues in MRA components, delays due to current stabilisation at the output of the current driver. 7. Conclusions The paper is concerned with an experimental study of real-time control of an MRA employed in a driver seat suspension. The driver seat-MRA sys- tem was tested in open loop and feedback configurations for vibration and shock isolation. The designed real-time controllers (CON1, CON2, CON3) for theMRA implement on-off and continuously variable control schemes utilising velocity signals fromthedriver-seat components.For this reason, special atten- tionwas given to sensors used in the experimentswhichwere tested to see how reconstructed velocity signals should affect output signals of the controllers to theMRA. Controllers CON1, CON2, CON3were developed in the integrated design and control environment of the MATLAB/Simulink. The performan- ce of the driver seat-MRA system in open loop and feedback configurations was investigated under harmonic and shock excitations. The analysis of the performance factors for vibration isolation revealed that CON2 had the best features. Tests revealed also the presence of undesirable phenomena during operation of the controllers (e.g. chattering effect). That applies to all con- 652 B.Sapiński trollers, however it seems to be predominant in controller CON1. Similar tests for the driver seat-MRA systemwere conducted under shock (rounded-pulse) excitations. The comparison of system responses in open loop and feedback configurations (with controllers CON1 and CON2) lead us to the conclusion that the action of the developed controllers fails to reduce the acceleration in the system. The reasons for this state of affairs are attributable to the pro- perties and operating principles of the electromagnetic circuit of the MRA employed in the investigated driver seat. Research is now underway to develop real-time controllers for MRAs in driver seat supports on digital microcontrollers. Acknowledgement The research work has been supported by the State Committee for Scientific Research as a part of research programNo. 5T07B02422. References 1. Ahmadian M., 1999, On the isolation properties of semi-active dampers, Jo- urnal of Vibration and Control, 217-232 2. Grega W., 1999, Sterowanie cyfrowe w czasie rzeczywistym, Krakw 3. Liu Y., Mace B., Waters T., 2002, Semi-active dampers for shock and vibration isolation: algorithms and performance, Proc. of the Int. Symp. on Active Control of Sound and Vibration, UK, 1121-1132 4. Sapiński B., 2004, Linear magnetorheological fluid dampers for vibrationmi- tigation: modelling, control and experimental testing, Rozprawy Monografie AGH, 128 5. Sapiński B., 2005, Fuzzy control forMRdamper in a driver’s seat, Journal of Theoretical and Applied Mechanics, 43, 1, 179-201 6. Sapiński B., Rosół M., 2004, Processing of measurement signals for MR damper control in suspension systems,Proc. of the Int. Symp. onActiveControl of Sound and Vibration, USA, 1-9 7. ANALOG DEVICES, 1999, ADxL202/210. Low Cost ±2/± 10g Dual Axis Accelerometer with Digital Output, Data Sheet, USA 8. INTECOLTD., 2002, User’s guide, RT-DAC4multi I/O board 9. LORDCORPORATION, 2003, http://www.lord.com 10. MOTIONMASTER SYTEM, 2004, http://www.lord.com Real-time control for a magnetorheological shock absorber... 653 11. PELTRON LTD., 2004, Transformer linear displacement transmitters PSz series, http://www.peltron.home.pl 12. THEMATHWORKS INC., 2003,Real Time Workshop User’s Guide 13. U.S. PATENT No. 5,652,704, 1997, Controllable Seat Damper and Control Method Therefore Sterowanie w czasie rzeczywistym amortyzatora magnetoreologicznego w fotelu kierowcy Streszczenie Artykuł podsumowuje badania autora dotyczące sterowania w czasie amortyza- tora magnetoreologicznego (MR) w fotelu kierowcy. Wykonano eksperymenty, któ- rych celem było zbadanie skuteczności amortyzatora do tłumienia drgań i wstrząsów w otwartym i zamkniętym układzie sterowania. Regulatory czasu rzeczywistego typu dwupołożeniowego i ciągłego dla amortyzatora zrealizowano w zintegrowanym śro- dowisku projektowania i sterowaniaMATLAB/Simulink. Przeprowadzono testy, ob- razujące wpływ odtworzonych na podstawie sygnałów uzyskanych z czujników prze- mieszczenia i przyspieszenia sygnałów prędkości na sygnały wyjściowe regulatorów. Manuscript received December 16, 2004; accepted for print February 16, 2005