Home

Analysis and Design of a Redundant X-by

1. 1 1 Project Background 0000 12 PU POSE e oii se Ohare Goan Se GG AMO aha ja ae oR Hare eB 121 Objectives s ge seed a ee ee a a OOK 1 222 limitations s cca a dee a SER se sh r 1 3 Report Structure 2 20 20 002 es ses 2 Inventory 2 1 Car Overview eas ae Rh eee ea 2 2 Network and Connections 0 0 00000 ora 2 21 Node CH cece wat a a de ere ee Be eS 2 22 N deCR i cose ea dad r l Aw a e dr dr dr da a wo 2 2 3 Wheel Nodes 2 0 0 00 rr ee ee 2 3 Mechanical Components 0 0000002 eee 2 3 1 Act ators s s e s gos a saa od RR dh db de ana a OR RR 2 32 Sensors e ceiu d r a Se Ee ew a Oe EE 2 4 Power and Electronics 2 2 ser rr sr ee eee ee 2 4 1 Electric Power Supply 0 2 4 2 Motor Control 0 0 a 2 4 3 Signal Adapting Electronics 2 css ccs cc coc cc 3 Modifications 3 1 Steer Actuator Joint 2 ee 3 2 Front Wheel Encoders 2 2 2 ee 3 3 Wheel Actuator Control Boxes 2 0 2 2 000002 ee 3 4 Wheel Nodes Circuit Boards 2 208 4 Synthesis 4 1 Possibilities and Drawbacks 040 4 2 Controller Structures oso rr rr rr rs sr ss ss sr 4 2 1 Wheel Angle Controller
2. mm Figure E 5 Front control box attachment the attachment base 114 Manufacture Drawings mm Figure E 6 Front control box attachment the attachment clamp mm Figure E 7 Front control box attachment the clamp support 115 116 Manufacture Drawings mm Figure E 8 Front control box attachment the short bracket fixes the attachment base in the car 117 ao mm Figure E 9 Front control box attachment the long bracket fixes the attachment base in the car 118 Manufacture Drawings rom Figure E 10 Rear control box attachment the attachment base A A SECTION mm Figure E 11 Rear control box attachment the attachment top 120 Manufacture Drawings oy ri kT LINK PING UNIVERSITY oo ELECTRONIC PRESS te S Pa svenska Detta dokument halls tillg ngligt pa Internet eller dess framtida ers ttare under en l ngre tid fr n publiceringsdatum under f ruts ttning att inga extra ordin ra omst ndigheter uppstar Tillg ng till dokumentet inneb r tillst nd f r var och en att l sa ladda ner skriva ut enstaka kopior f r enskilt bruk och att anv nda det of r ndrat f r icke kommersiell forskning och f r undervisning verf ring av up
3. Balls crew actuator The balls crew is completely rigid so the time constant is zero Linear sensor The resistance changes instantly when the length of the sensor changes The PowerNode then samples this value which adds a delay The data sheet for the Motorola PowerPC processor states that sampling takes at least 14 QCLK and the QCLK is 2 MHz gt 7 us 5 1 Wheel Angle Controller 47 Encoder The data sheet states that the codes coming from the encoder can change with a frequency of 100 kHz gt yy 1 107 10 us Software algorithms This will take no longer than 5 ms Network The delay of messages on the network depends on the TDMA round see Chapter 6 which was chosen to be 10 ms This delay will only affect the messages that need to be transferred over the network In conclusion the system that needs to be controlled from PWM value to wheel angle has a delay of roughly 4 1074 45s 10 7 47 10 410 10 45 10 lt 20 ms plus an additional 10 ms for values that has to travel the network So an integrator with a time delay of 20 ms would be a reasonably good approx imation for this system The delay was chosen high since the practical resolution in the finished system is 10 ms since the controller is updated every second TDMA round refer to Chapter 6 for more information F 1 1 Dyin out gt gt gt 5 motor_speed wheel_angle Variable Discrete Time Unit Delay Unit Delay1 sa
4. 50 10 201583 50 5 14530 oao 2 74 A 9 To be able to change the throttle valve opening in the carburettors a servo20 has been mounted underneath the intake manifold Due to the lack of space the servo is not directly attached to the throttle valve instead a flexible steel wire has been installed between the servo output wheel and the valve arm the wheel is seen in Figure 2 13 The carburettors also have a cold start enrichment device commonly known as a carburettor choke not to be confused with DC motor chokes mentioned earlier which is controlled by a car door lock servo mounted behind Node CL see Figure 2 3 20HiTEC HS 805BB 4 11 12 21VDO IMP 6880 no specification available 18 Inventory Figure 2 13 The throttle servo no 2 viewed from above the carbu rettors Just below the centre of the figure the wheel no 3 attached to the servo output is seen Slightly to the left of that connected to the wheel is the flexible steel wire no 1 going up to the throttle valve Last but not least there are two MR brakes installed in the car They are both used to apply some sense of force feedback to the driver by increasing the friction The first one22 module see Figure see Figure The MR brake in the steering module is controlled by a special control card 2 15 has been attached directly on main shaft in the movable
5. In addition two other subsystems supervision and calibration have to be de fined to make the car more dependable and easier to maintain and setup These subsystems are also run on all nodes naturally since all nodes have sensors that need calibration and it is important to supervise all nodes Finally a subsystem that is responsible for driver feedback is defined This subsystem is only run on the two center nodes 59 60 Schedule Table 6 1 The subsystems and their responsibilities Subsystem Center nodes Wheel nodes Steering Read steering wheel Read wheel angle SS Oeuiate wool anges Tara wnest Speed controlling Read brake pedal Read brake pressure Calculate brake values Actuate brake Read clutch pedal only CL Read wheel velocity Read throttle pedal only CR Supervision Read switches Create status Control indicators Create status Driver feedback Calculate feedback Not implemented Actuate feedback A summary of all of the subsystems can be found in Table 6 1 6 2 1 Steering For the two center nodes the steering subsystem includes reading the steering wheel angle and calculating the four commanded wheel angles one for each wheel This subsystem is completely replicated since the two centre nodes make exactly the same calculations so even if one of the nodes should fail the system will still work However if both of the center nodes are functional the values of the commanded wheel angles fr
6. 70 mA 24 VDC 600 mA sumption As mentioned earlier the system is equipped with several analog sensors On top of the four wheel actuators linear position sensorg25 have been mounted see 27Leine amp Linde 670 670066350 17 16 28 BEIDuncan 610 22 Inventory Figure 2 16 and 2 17 In principle these sensors are doing the same job as the digital encoders mounted on the steering spindle i e they measure the wheel angle In the movable pedal module four identical analog angular sensors29 are fitted which measures the angular displacement of the shafts when any of the pedals are pressed see Figure 2 15 The accelerator pedal and the clutch pedal have one sensor each and the braking pedal is equipped with two sensors The four independent brake actuators are equipped with one pressure sensor and one micro switch each Both are mounted on the cylinder block the pressure sensor is seen at the bottom in Figure 2 12 The pressure sensor is of course used for measuring the oil pressure in the system and the micro switch is used to indicate when the piston is in its rear end position Finally the car is equipped with two inductive speed sensors 2 mounted inside the rear steering spindles see Figure 2 17 These sensors are used to measure the wheel angular velocity i e the speed of the vehicle since the normal speedometer is a pure mechanical construction The sensors register the change in the
7. Electronics The adaptive electronics for the two center nodes need a remake And all of the boards should be redesigned to better protect the PowerNodes i e use opto couplers and other protective electronics Sensors A yaw rate sensor would be needed to implement more advanced algo rithms Such a sensor could be connected to one of the PowerNodes via the CAN interface This would make it possible to use a standard yaw rate sensor form any modern car Power supply The charging of the batteries is not optimal One solution is to convert the whole car to use the 24V power supply system another is to convert it to the new 42V system that is believed to be fitted in the cars of tomorrow Stress calculations The critical parts like the universal joints should be tested 8 1 Future work 85 8 1 2 Software Modifications Algorithms More advanced algorithms for both steering and braking could be implemented Both ones that make more use of the existing sensors in the car and those that would require fitting of additional sensors Controllers The implemented controllers need further testing and they can probably be improved in many areas One is to derive an algoritm that dynamically set the value for switching between the two steer angle con trollers Redundant sensors More advanced algorithms to decide if a sensor is failing The algorithms could probably also be refined to deliver safe values even if fail situations Failsafe software
8. Node CR has a circuit board see Figure 2 21 quite similar to the one just described Similar components are used to adapt the second encoder signal the second brake pedal sensor and the throttle pedal sensor This is also partly true for the rest of the switches connected to this node i e the signals are passed through with no adaptive components in between The difference is in the way the switches have been connected to the PowerNode Instead of using the I O pins or the PWM channels as been done in Node CL the analog ports have been used Again this is not as expected but possible way for connecting the switches since also the analog ports can be re configured to function as I O pins Figure 2 21 The circuit board with adaptive electronics for Node CR Note the special control card for the steering wheel MR brake actuator to the right no 6 Other components shown are the operational am plifier no 1 the resistor bridge no 2 and protection circuit no 3 for the encoder signal the DC DC converter no 4 and lastly the 12 V no 5 and 5 V no 7 voltage regulators The connection of the dash board diodes are not that different compared to the switches They are also directly passed through to the PowerNode but three of them are connected to the PWM channels and one to an I O pin 28 Inventory Two actuators are controlled by this node the throttle servo and the steering wheel MR brake The control signal to t
9. Parking brake dash board switch Clutch actuator reference input Carburettor choke forward Carburettor choke retract PowerNode power supply 40 pin PowerNode connector Brake pedal sensor Clutch pedal sensor 12 V brown 24 V red 12 V out n a GND black Board power supply 10ns t Board Connecti ircui C 104 st SE IEEE TR Arn Rd srt ta Rd Peer ee eerenre fortfor tr FEK er EF EEC a 4 Fre ret Ci L Phe eah baad Pheer hanes lt gt f a ME ELUIFIANOD 30 30 si VEO eee en ob VOR eee yee hak ve UOC eres tba ek et Nt gig ie sed eben eens Petey terry Oper PPC eee ey a a wa tw Figure D 2 Circuit board wire connections for Node CL The num bers are specified in Table D 2 105 Table D 2 Node CR circuit board connections Below is the description to the numbers in Figure Number CO NI Ol OY AY Co DO Fe siol OU BY Co DOP ejl ole a oo D 2 The connectors are specified when appropriate from left to right Description Steering wheel MR brake MR brake controller card Carburettor throttle valve servo 2 way dash board switch 50 pin PowerNode connector 2 way dash board switches three in total Housing fan power supply 2 way switches GND Steering wheel encoder signal output Steering wheel encoder power supply Dash board LED s GND
10. cations bus or star Each node in the system has one CPU for the host application and another for the communications system The nodes in the cluster considered in this master thesis are six TTTech PowerNodes The host CPU is Motorola MPC 555 and the communications processor is TTTech s own TTPC C1 chip The host CPU is where all the computations are made and the communications system is responsible for all data exchange between the nodes The interface between these are called the CNI or Communication Network Interface A 1 1 Communications Subsystem Communication between the nodes happen on a broadcast type bus or star topol ogy network using the time triggered protocol TTP C Broadcast type means that a message will not have a receiver but instead all nodes can read it The difference between bus and star topology is that in a star network each node is connected to some sort of hub whereas in a bus topology a single bus connects the nodes together Since the communications are time triggered all messages must be defined in advance both in the temporal and the value domain This also guarantees that the nodes all have one synchronized clock and it makes it easy to detect errors 1 When the message is sent and what it contains 91 92 Programming and Software Tool User guide To access the network the nodes follow Time Division Multiple Access or TDMA rounds This means that a node is only allowed to transmit in its o
11. It describes the step from the overall demands to something that can be used in practice Chapter 6 covers the steps to be taken to get the system up and running This involves defining tasks messages and a schedule Chapter 7 includes results from different tests performed on the system It gives feedback on how well the practical results relate to the simulated ones Chapter 8 summarizes the work done and presents conclusions made Also com ments on future work are found here Appendix A gives an introduction to the software tools used to program the network Appendix B presents a detailed diagram of the power distribution to the in stalled hardware Appendix C includes circuit diagrams for all adapting electronics used in the car Appendix D gives an detailed explanation on how to connect the wires to the circuit boards used Appendix E lists the manufacture drawings made 7 This is a description of when the system should do what Chapter 2 Inventory Before any work can be done detailed knowledge about the car and installed components is needed In this chapter the car is first examined at a macro level to give an overall picture Later each component in the X by Wire system is covered in more detail starting with the most complex part the network Thereafter mechanical components such as sensors and actuators are analyzed
12. The car should be able to adjust its capabilities and handling to cope with certain faults in the system so called degraded operation modes Safety The car needs a lot of testing to verify the systems and the implemented software It is also possible to implement crash avoidance ABS and various anti skid systems 86 Summary and Conclusions Bibliography 10 11 12 13 BEI Technologies Inc http www beiduncan com Model 610 Filename BEIDuncan 610 linj r givare pdf BEI Technologies Inc http www beiduncan com Throttle position and industrial control sensor modules Filname 9811 9812 pdf BOSCH http www bosch com Pressure Sensor Filnamet pres sure_sensor jpg Burns A and Wellings A Real time systems and programming languages Ada 95 real time Java and real time Posix Addison Wesley 75 Arlington Street Suite 300 Boston MA 02116 USA 2001 T Glad and L Ljung Reglerteknik Grundlaggande teori Studentlitteratur 2nd edition 1989 In Swedish T Glad and L Ljung Reglerteori Flervariabla och olinjara metoder Stu dentlitteratur 1997 In Swedish Fredrik Gustafsson Adaptive Filering and Change Detection Wiley 2000 HENGSTLER http www hengstler de index_e html Absolute Shaft Encoders Filename A DGpage106 182fileD pdf HENGSTLER http www hengstler de index_e html Installation in structions Filename
13. actuate wheel angle Wheel Actuator Speed RL Calibration Mode Calibration Mode Figure 6 9 The tasks messages and how they are related for the steering subsystem in the rear left wheel node 6 3 Tasks and Messages 71 Local Global read_encoder Wheel Encoder_Raw_RR 3 read_actuator_len Wheel_Actuator_Len Raw_RR Wheel Angle RR Wheel Actuator_Speed_RR check_wheel_angle Wheel_Encoder_Raw_RR Wheel Actuator_Len_Raw_RR Wheel_Angle_RR i Calibration Mode jqe Wheel_Angle_ Steer_Value_ calculate actuator speed Wheel_Actuator_Speed_RR actuate_wheel_angle Wheel_Actuator_Speed_RR Calibration Mode Figure 6 10 The tasks messages and how they are related for the steering subsystem in the rear right wheel node calculate actuator speed Calculates the value that should be sent to the motor control box by comparing the commanded wheel angle and the current wheel angle for all four wheels actuate_wheel_angle Sets a PWM value to control the wheel angle ac tuator Messages Wheel_Encoder_Raw_ Raw value from the encoder attached to the wheel Wheel_Actuator_Len_Raw_ Raw value representing the length of the steer angle actuator as measured by the linear sensor Wheel_Angle_ Verified value representing the angle of a wheel This value is adjusted so negative values mean that the wheel is steered to the left and positive that it is steered to the right Steer_Value_ Commanded angle for a wheel
14. and calibrating subsystems in the front left wheel node 78 Schedule Chapter 7 Results This chapter answers the question on how well the theoretical controllers and algorithms presented in Chapter 4 and 5 work when implemented in the car It starts out by discussing the wheel angle controller and thereafter the brake controller is evaluated Something is also said about the algorithms that distribute brake forces and steer angles among the wheels Only simple tests were performed on the system since the time frame did not allow more thorough evaluations 7 1 Wheel Angle Controller As a start the wheel angle controller was implemented in one of the rear nodes exactly as the simulated one see Section 5 1 with the same parameters on PI part of the controller A first test showed however that these parameters could not be used at all since the gain of the model did not correspond to the real system in the model the loop was closed with the DC motor angle as the measured signal and in the real system the wheel angle is used To adapt the Pl controller parameters the Ziegeler Nichols method was used on the real system this time giving Ko 10 and Tr 0 23 s The corresponding parameters for the controller were PI K 0 45 Ko 4 5 Tr 2 0 19 1 2 A second test was performed with a step reference and the results showed a significant overshoot when the Pl controller was in actio
15. and last but not least the surrounding electronics needed to adapt signals between components and the network are examined Please note that the hardware listing only apply to the car s present state during the end of this project First however a clarification regarding the use of expressions should be noted In the following sections and in the rest of the report the expression Node will be used frequently Node CL for example It should NOT be confused with PowerNode as Node refers to a collection of equipment fitted inside a protec tive housing or the housing itself The expression PowerNode refers to a special piece of hardware and is part of the equipment inside the housing 2 1 Car Overview As mentioned briefly in Chapter 1 the car is a Tiger Cat El a replica of the old Lotus Super 7 racing car but modified during a final year project by students at Lulea university of technology into a complete X by Wire vehicle In co operation with Volvo Cars in G teborg the project design task was to deploy solutions for the steering and braking systems in order to allow the car to turn around its own axle be moved in parallel and have both the left and right hand steering Another design task was to modify the engine suspension in order to reduce vibrations during idle running At the end of the project the students had indeed succeeded in their task and were able to demonstrate a functional prototype To
16. by Katja Tasala Numerous of incidents have happened over the years which all are related to hard or software faults in the computer based controller system In a computer controlled radiation treatment equipment the Therac 25 six accidents happened where three people died of radiation injuries during the years 1985 87 The reasons were faults in the Dynamic Stability and Traction Control 4 2 Controller Structures 39 user interface a 6 bit counter where set to 0 and a hardware lock had been removed 40 In 1992 a Boeing 767 belonging to Lauda Air suddenly changed the jet drag direction during a fight 40 Although it is not possible to guarantee a flawless system the chance of stability and correct functionality can be increased by using redundant components Doing that though presents new problems like choosing which component to trust in a fault situation and which action to take if one or several of the components fail Again it is the engineer who has to construct algorithms to handle that 4 2 Controller Structures The car presents opportunities and drawbacks not found in any of the commercially available cars today The wheel angles and brake pressures can be controlled individually at each wheel without the need of driver input In fact with some small additions one could actually remove the driver entirely and steer the car remotely The same contingencies also apply to the throttle valve
17. called Ackermann steering and is often measured in percentage of true i e 100 Ackermann An explanatory sketch can be found in Figure Using 100 Ackermann the four wheel angles can be calculated as o tana ZE r 3 tan Joe an 3 re tany gt S To tan r 3 when using the notation given in Figure 5 8 Traditional cars never have true Ackermann because the wheel angles are al ways a compromise between stability and tire wear In the car at hand all wheel angles can be set independently of the others which make true Ackermann an obvious choice 5 2 2 Steer Angle Distribution For a traditional two wheel steered car the center of rotation is always fixed to the rear axle This is not the case for a four wheel steered car where the center of rotation can be put anywhere on a line going through the middle of the car It is also possible to control how the car is orientated during the maneuver If the steering angles on the front and rear axle are equal the car can move side ways without rotating This can be good for lane changing or obstacle avoidance since the car avoids rotating The maneuver can be seen in Figure 5 9 Another special case is when the center of rotation is set in the middle of the car This lets the car turn with the shortest radius and this is especially good for parking and maneuvering in tight places However under normal driving conditions none of these two special cases are usable sin
18. differs from the other control boxes as it is pre configured to function together with the actuator as a position servo i e the input voltage corresponds to a length on the actuator 33maxon ADS 50 10 201583 34maxon ADS 50 5 145391 26 35SKF CAED 9 24R PO 24 Inventory Figure 2 18 The actuator control boxes for the rear wheel actuators are located just behind the batteries The two boxes no 1 4 closest to the batteries control the steering wheel actuators and the other two no 2 5 the brake actuators Also seen at the top of the figure is the fuse box no 3 containing power supply fuses for the rear nodes and actuators Figure 2 19 The clutch actuator control box located inside Node CL 2 4 Power and Electronics 25 2 4 3 Signal Adapting Electronics Although the PowerNode is very advanced there are some limitations to what it can do The maximum voltage allowed on the input pins both on the digital I O pins and the analog ports are 5 V and the output PWM 2 channels can provide signals between 0 and 5 V 41 To handle these limitations adapting electronics have to be used between the PowerNode and its surrounding equipment The electronics contribute with essentially three functions e Supply the sensors with power e Adapt and filter sensor signals to the PowerNode s I O ports e Amplify and filter control signals coming out from the PowerNode As the connected eq
19. each other using the bus topology The installed sensors and actuators are in turn connected to these PowerNodes see Figure 2 2 where each PowerNode is listed with its surrounding components 2 MPC555 Black Oak 30 3 All nodes participating in the cluster receives every message sent on the network 4Although the star topology introduces a higher degree of safety in the system it has not been used since the old version of the software tools used at Lulea did not support it 8 Inventory Brake limit switch Brake limit switch Brake pressure sensor Brake pressure sensor Brake actuator Brake actuator Wheel angle actuator Node FL lt Node FR Wheel angle actuator Encoder Encoder Linear potentiometer Linear potentiometer Carburettor choke Throttle servo Clutch actuator Steering wheel MR brake Brake pedal MR brake Node CL e gt Node CR Indicators Park brake switch Switches Brake pedal Steering wheel Throttle pedal Clutch pedal Figure 2 2 An overview of the network and how the different components are connected to each other CL Central Left CR Central Right FL Front Left FR Front Right RL Rear Left RR Rear Right 2 2 Network and Connections 9 2 2 1 Node CL Node CU is found on the left side just in front of the dash board the grey box to the left in Figure 2 3 where it controls the clutch actua
20. gt controlled wheel SET out gt motor_speed wheel ange ej P Lp other wheel controller2 wheel2 Figure 5 5 Simulink model which represents two wheels In Figure 5 7 the same commanded values as before were used but one of the wheels was disturbed by a sinusoidal signal instead of white noise 52 Implementation 25 2L ste 4 I c oO oO 2 5 o a ib ee SER 0 5 oe J yi d oa 2 o i i 1 fi fi 1 1 1 fi 1 1 2 1 4 1 8 1 8 2 2 2 2 4 2 6 2 8 3 Time s Figure 5 6 First test run of the wheel angle controller 25 2L sl 4 D c oO oO 5 o a ib E ea a 0 5 T 4 Lo Oh fi fi fi fi fi fi fi fi 1 1 2 14 1 6 18 2 2 2 24 2 6 2 8 3 Time s Figure 5 7 Second test run of the wheel angle controller 5 2 Steer Algorithms 53 5 2 Steer Algorithms This section discusses the problem of translating a given steering wheel angle into commanded wheel angles The algorithm plays a significant role in how the car is behaving which make it important that this is a well designed algorithm Everything that is said below assume that the wheels all roll without slipping This is of course far from the truth but the results are nevertheless useful 5 2 1 Ackermann Steering When a car performs a turn the inner and outer wheels roll on different radii To prevent the tires from skidding the inner wheels must turn with a greater angle than the outer wheels This is
21. hand it requires a quite large engagement in the car to manufacture the linkage and the encoder mounting and this was beyond the scope of this project Figure 3 3 The front wheel encoders The old 10 bit encoder with the gasoline tube connected to the spindle is seen to the left and to the right is the new 14 bit encoder with bellow coupling Instead the second solution was chosen and several different couplings have been investigated It were found that quite a few couplings allowed the required degrees of freedom and among these were the curved tooth gear coupling and the bellow coupling but none offered the extent of movement needed On the other hand the springs in the wheel suspension have been tightened to a maximum probably done when this problem was first encountered In practice this implies that the suspension will not move at all during normal driving conditions and therefore release the demands on coupling movement this also implies the car will not be very comfortable when driven The conclusion was to go for another coupling despite that the requirement of extensive movement was not fulfilled Here the curved tooth gear coupling should have been chosen but instead a pair of bellow couplings were fitted as they were found among some spare parts to car 3 3 Wheel Actuator Control Boxes The housing for the wheel nodes were originally pretty crammed Not only did they house the PowerNodes and
22. i e the delay in the system before full brake pressure was reached Both tests point to the fact that the open loop controller is not suitable for the brake actuators installed in the car T T T Pressure Brake Pedal pee bH A oe tp ey Park Brake Pressure Pedal Park brake Figure 7 3 Two simple tests performed on the open loop brake con troller 7 3 Algorithms The steer and brake distribution algorithms have not been implemented or tested due to lack of time The car can not be driven on normal roads since it is not licensed and to evaluate the algorithms the car has to be driven with differ ent speeds Safety is also an important factor since the authors insurance would probably not be valid if an accident happened while test driving the car Chapter 8 Summary and Conclusions This project has shown that the Sirius 2001 Concept Car is an ideal platform for research in safety redundant sensor handling and vehicle dynamics just to name a few Even though the car still have a couple of technical flaws that somewhat limit its capabilities the freedom of the application programmer is enormous The flaws are also described in this report so they can be modelled easily The software and scheduling that are implemented in the car make it especially easy to target one area and concentrate on that without the need to learn other aspects of the system This report is also useful when starting new projects
23. limit is due to the top speed of the motors and the gear ratio of the ball screws In a panic situation tests have shown that the driver can turn the steering wheel up to 800 s which corresponds to 80 s at the wheels This is more then four times faster then the maximum speed possible with the actuators fitted in this car and the only way to remedy this is to replace the actuators with faster ones 7 2 Brake Controller Only an open loop brake controller was implemented due to the limited time frame However some simple tests were performed to verify the impact of some of the drawbacks discussed in Section 5 3 with a controller of this type The controller structure was made very simple The PWM signal to the ac tuator was made directly proportional to the brake pedal position i e full pedal reading represented 100 PWM duty cycle Only when the pedal was released completely the brake piston was explicitly retracted 82 Results In Figure 7 3 the results of two simple tests are seen First the pedal was pressed as fast and as far as possible and then released The result shows a delay of almost a second before the pressure was built up in the system Also which is more serious the pressure did not decrease until the pedal was released fully Second the parking brake switch was turned on which sets the PWM duty cycle to 100 instantly representing a step input The result confirms the first result stated above
24. magnetic field when a tooth gap in the gear wheel fitted on the axle shaft passes by 30 2 4 Power and Electronics Not only does all equipment need some kind of power source in order to function the PowerNodes also need a number of surrounding electronic components to help them interact with the rest of the parts in the system These components are classified into two categories motor control boxes which amplify control signals to the actuator motors and electronics to adapt and filter signals coming in to and going out of the nodes 2 4 1 Electric Power Supply As in all systems which include any type of electrical devices a power supply is needed This car has two separate supply systems one 24 V system and one 12 V system The 12 V system is used for all the standard equipment in the car among these are the ignition system head lights etc It will not be described in further detail in this report since it does not interact with the by Wire part of the car The 24 V system on the other hand is indeed interesting since it supplies power to all the added X by Wire equipment i e network nodes actuator control boxes and sensors The base of the 24 V system consists of two 12 V 100 Ah VARTA batteries connected in series These are located at the back of the car see Figure 2 18 and are charged with a 24 V 70 A Motorola generator mounted on the right side
25. of the engine One should note that although the 12 V system is separated from the 24 V system it is still connected to one of the batteries making the load on the two 29BEIDuncan MOD 9811 2 30Bosch 0 265 005303 31Saia Burgess VANCSK2 34 32VOLVO 6849311 0988 10 0711 1146 1 no specification available 2 4 Power and Electronics 23 batteries differ and the charging difficult Because of this the batteries should be shifted from time to time To protect the equipment fuses have been put in between each of the connected component and the power supply i e each node and actuator control box has its own fuse These fuses are found inside two fuse boxes located on just behind the engine see Figure 2 3 and above the batteries see Figure 2 18 There are also two battery master switches one for each system located be tween Node RL and Node RR just behind the seats in the car the 24 V master switch is the one closest to the nodes in Figure 2 3 In Appendix Bla schematic overview over the power distribution can be found 2 4 2 Motor Control There are three different types of motor control boxes in the car one type for the steering actuators one for the brake actuators and one for the clutch actuator The boxes make it easy to control the motors and remove the need for analyzing the specific properties of the motor since the boxes are specially constructe
26. on the bit is 1 otherwise the bit is 0 Calibration Mode This message is used to inform the user if the system is in calibration mode Node_Status_ The system uses these to inform the driver of any faults in the system Node_Status_C This message contains the status of the two center nodes It is created using the status information from the messages Swheel_Angle Pedal Clutch Value Pedal Throttle Value and Pedal Brake Value 68 Schedule Calibration mode is chosen with the dashboard switches and the procedure is then stepped through by operating another switch Calibration is made by storing raw sensor readings in the PowerNodes memory The calibration task on the center nodes must also create the calibration mode message to inform the other nodes Finally the driver_feedback subsystem consists of two tasks and two messages Tasks calculate_feedback Calculates values for the feedback by looking on the steering wheel angle and the brake pedal value actuate_feedback Uses the value calculated by calculate_feedback to control the magneto reological brakes that are fitted to the steering wheel and the brake pedal Messages Swheel_Feedback The amount of feedback that should be applied to the steering wheel It is created using the Swheel Angle message Brake_Feedback Amount of feedback that should be applied to the brake pedal created using the Pedal Brake Value message 6 3 Tasks and Messages 69 Local Global Wheel_An
27. opening and the clutch The only thing you still need to do manually in the car is shifting the gears In order to control all these parts and make use of the performance built into the car controllers are needed These controllers must be stable in every situation even if a sensor fails To guarantee stability in a situation where a sensor gives incorrect readings an open loop controller is the safest alternative since it stays unaffected It can however be hard or even impossible to construct such a controller if the system is non linear or otherwise difficult to model Ost 4 2 1 Wheel Angle Controller The wheel angle controllers job is to keep the wheel angles synchronized with a commanded value i e work as an angle servo In addition it also have to keep track of where the other wheels are when moving A very important factor for the cars stability is the aforementioned toe angles This makes it imperative for the controller to at all times keep the wheels moving correctly with respect to each other The main demand on a wheel angle controller is naturally stability It has to be in every situation stable since an oscillating wheel would surely result in a violent crash In this particular application it is also important for the controller to be tolerant towards model errors since the four wheels all have different parameters The parameters of a wheel can also change substantially when driving if for instance the road chan
28. property law the author has the right to be mentioned when his her work is accessed as described above and to be protected against infringement For additional information about the Link ping University Electronic Press and its procedures for publication and for assurance of document integrity please refer to its WWW home page http www ep liu se P r Degerman amp Niclas Wiker
29. should have been if using only one sensor e If the readings from a redundant sensor differ which is correct More on this subject can be found in 7 4 4 1 Wheel Angle Sensors The two sensors that measure the wheel angle on one wheel do not have a linear relationship to each other depending on the geometry of the suspension This means that the values have to be adapted to each other One way to do that is to look at the geometry of the wheel suspension and construct an algorithm that takes either value and translates it to the other This is however cumbersome since it involves a whole lot of complex algebra due to the complicated geometry of the suspension A more practical solution is to create a lookup table to translate between the values The dynamics of the steering including actuator motor and motor control box make it easy to model the behavior of the wheel Since the input to the motor control box is a reference value for the motor speed a simple integrator with delays taken from the data sheets of the motor control box and motor 25 27 is a fairly correct model see Section 5 1 for a more detailed description of the model The model is used when comparing the two sensor values In reality three values for the wheel angle can easily be found two measured and one calculated using the output of the model when it is fed all earlier control signals This makes it possible to spot single errors by invalidating the s
30. small modifications were done in order to adapt the added equipment and take care of the prior faults The same board layout is used in all wheel nodes although all components are just utilized in the rear nodes This was made to simplify for a prospective addition of sensor in the front nodes like speed sensors for example The circuit boards were manufactured in the University workshop in Figure 3 7 the milling of the connection wires is seen For the complete circuit diagram of the board please refer to Appendix C and in Appendix D a description on how to connect the actuator and sensor wires is found For a review of the board and its components see Section 2 4 Just before the printing of this report a fault in the circuit board design was discovered The 2 2 kQ resistor used for the pressure sensor had been misplaced instead of being connected in parallel between the output signal and the 5 V power input it was put in series with the power input However it is possible to connect the sensor to the board despite this error see Appendix D for details Figure 3 7 The milling of wire connections on the wheel node circuit boards Chapter 4 Synthesis Until now only the hardware installed in the car has been discussed and nothing has been said about how to make everything work together as a whole The glue to tie everything together is the algorithms used in the distr
31. steering 2 14 The second ong acting on the brake pedal has been installed in the pedal module 24 located inside Node CR the card is seen in Figure 2 21 The other MR brake however also needs a controller card similar to the other but although prepara tions have been made to add one the actual card is yet to be found 2LORD RD 2028 18X ol 19 23LORD MRB 2107 3 18 24 ORD RD 3002 03 20 2 3 Mechanical Components 19 Figure 2 14 The movable steer module The steering wheel MR brake no 1 is seen in the middle of the module and the two absolute encoders no 2 are found further to the right in the figure Figure 2 15 The pedal module Note the duplicated sensors no 1 4 and the MR brake no 3 attached on the brake middle pedal The clutch pedal sensor no 2 and the sensor for the accelerator pedal no 5 are seen at the outer ends of the figure 20 Inventory 2 3 2 Sensors The car is equipped with both digital and analog sensors At each wheel a 14 bit digital absolute shaft encoder has been installed to measure the wheel angle The encoders at the front have been mounted on top of the upper lever arm and measure the angle of the spindle see Figure 2 16 Figure 2 16 The front right wheel angle sensors The absolute shaft encoder no 1 is seen mounted on the upper lever arm and towards the bottom right co
32. the sign of the angle error to make the wheel turn in the right direction The model can easily be extended to handle four wheels by replacing the other wheel signal with the largest angle error of the other wheels 5 1 4 The PI Controller To remedy the bang bang controllers poor stability it is deactivated for small wheel angle errors instead a Pl controller is used Ziegler Nichols method 5 was used to tune the parameters for this controller When tuning a controller using Ziegler Nichols the steps below generate the initial values 1 Create a closed loop system with only proprotional control 2 Increase the proportional gain until the system starts oscillating with a con stant amplitude 3 Note the gain and period of oscillation When this was used on the wheel model presented above in Figure 5 2 the oscil lations began with a period of Ty 0 1 s when the gain was Ko 62 With these values the Ziegler Nichols rules give the controller parameters listed 50 Implementation below P K 31 PI K 045 Ko 27 9 Tr amp PID K 0 6 Ko 37 2 Tr 2 Tp 2 when using a controller on the form ult Kle t f e dr To Felt to where e t denotes the wheel angle error as a function of the time t Out Kp T z 1 Discrete Time Vii Integrator Figure 5 4 Simulink model of the PI controller From these values the PI controller was chosen and the amplification was low ere
33. using the car However the main objective of the project stated in Section 1 2 has not been reached as the car is at this time not fully functional Above all this is due to lack of time during the implementation and testing phase This is a direct consequence of the numerous problems encountered throughout the whole project like malfunctioning hardware and problems with the software tool licensing to name a few Despite this all of the goals stated in Section 1 2 has fully or partly been reached e The different parts included in the control system should be identified and well documented A complete listing of the hardware is found in Chapter Chapter 6 as well as Appendix Aj includes a comprehensive guide on how to master the realtime controller network e The implemented controllers and algorithms should be modified so the car behaves in a consistent manner when driven A wheel angle controller has been implemented with satisfying results Section 7 1 and drawbacks with an open loop brake controller has also been verified However the brake controller should have been modified to fulfill the demand of consistent behavior of the car Also algoritms for generating reference values for the controllers discussed in Section 5 2 and 5 4 has not been implemented at all e The system should be able to handle redundant components in order to 83 84 Summary and Co
34. 4 V Generator Front fuse box A CL A Clutch A FL steer A FL L Automatic fuse A FR brake A FR brake B ae A FR Starter motoj A FR steer 3 A CR An ie D oO MR brake 5 ot me lo Appendix C Circuit Board Diagrams This appendix shows the diagrams for the adapting electronics between the TT Tech PowerNodes and the sensors and actuators Figure C 1 show the boards in the wheel nodes Figure C 2 in Node CL and Figure C 3 in Node CR 97 98 240 Circuit Board Diagrams N F VUIN UQUT com i UIN COM UPPLY Hi PRES_SENS 1 VOUT HRR L78s12cu ENGA DC DC CONVERTER TEM2 1221 P 0 33a 3u a LIN_POT 1 Qu x to poe aT cT 12 z pres_sens 2 SY Ci c2 L78sa scu LIMIT SWITCH 1r 3 PARK BRAKE 2 T LIN POT 2 LIMIT SWITCH 2 eve Pt STEER_BOX 3 4 ENC 2 BRAKE _BOX 3 4 rt sus t ETRIGL ETRIG2 BANSS BANS8 BANSZ BANSS iea N gt i BANSS BANS4 3 E BANS3 BANS2 a 8 I T PRES_SENS 3 AGND5 AGND4 2 il gt BANSI BaNsa CAZ41E BAN49 BAN48 BANS BAN2 l BANI BANO AGCND3 AGND2 4 AANSS ANSE AANSZ ANSE DEN AANSS AANS4 G5 AAN53 AANS2 j ae LIN_POT 3 AGNDL AGND 4 JE J AANSL AAN5 1e7 Pe AANSS AAN48 3 NS AN 2 SPEED_SENS 1 AANI A
35. 5 Local Global read_throttle_pedal Pedal_Throttle Raw read_brake_pedal Pedal_Brake Raw je Pedal Throttle Raw Pedal Throttle Value check throttle pedal i Pedal Throttle Value je Pedal Brake Raw Pedal Brake Value check brake pedal Pedal Brake Value Pedal Throttle Value calculate throttle value C L Throttle Value a Pedal_Brake_Value Switches calculate_brake_values ace Value Park Brake actuate_thottle Throttle Value Calibration Mode Pedal Brake Value Calibration Mode Figure 6 4 The tasks messages and how they are related for the speed controlling subsystem in the CR node read_throttle_pedal Gets a raw reading from the throttle pedal this is only done on the CR node check_throttle_pedal Sees if the value that read throttle pedal reads is sane calculate_throttle_value Generates a value that can be used to con trol the throttle servo only on CR node actuate_throttle Generates PWM signal from throttle servo only CR read_brake_pedal Reads a raw value from the brake pedal This task is run on both nodes check_brake_pedal Checks the values from read_brake_pedal to see if they are correct This task reads values from both nodes and com pares them calculate_brake_values Calculates commanded brake forces for all four wheels Messages Pedal Clutch Raw Raw value from the clutch pedal Pedal Throttle Raw Raw value from the throttle pedal Pedal Bra
36. 65 F rfattare Author Analys och design av ett redundant x by wire kontrollsystem till Volvos konceptbil Sirius 2001 Analysis and Design of a Redundant X by Wire Control System Implemented on the Volvo Sirius 2001 Concept Car P r Degerman amp Niclas Wiker Sammanfattning Abstract Nyckelord Keywords The purpose of this master thesis project has been to analyze and document the Sirius 2001 Concept Car In addition it has also been a goal to get the car in a usable state by implementing new software on the on board computers The car is a Tiger Cat E1 that is modified with four wheel steering and an advanced X by Wire system The computers in the X by Wire system consist of six TTP PowerNodes that communicate with each other over a redundant fault tolerant TTP C communications bus The computers are connected to a number of sensors and actuators to be able to control the car This project has contributed to the car in several ways A complete documen tation of the systems implemented in the car is one Another is a programmers manual which significantly lowers the threshold when working with the car Last but not least is the modifications in hardware and software which have made the car usable and show some of the possibilities with the system The results show that the Sirius 2001 Concept Car is a suitable platform for research in car dynamics and fault tolerant systems The work has also shown that the TTP C communica
37. ANG HCPL2280 b 5 Pa J z p igati 8 OS SCK OS MOSI RIS 2 as MISD QS PCS3 OKT QS PCS2 QS PCSI e as_Pcsa GND4 Rut Oe eae aay oe GND3 QS_TXD2 ak RS 8_RXD2 OS ECK GND2 DDH ik it MPIOL5S MPIO14 M TESAET ERE MPIOL3 MP1012 ENC 15 Ba 1 MPIOLL MEIOLO N ENC 14 ENC 13 N ENC 12 MPIOS MPID8 ENC 11 Ria MPIOZ MPIOS 3 CAZ4LE ENC 18 EE ENC 9 k ENC 8 ENC 7 BRAKE _BOX 1 ENC 6 aot vE2 ENC 5 ENC 4 tl vaa ENC 3 MOALL MOAI2 N MDAL3 MDA14 MDALS MDA27 l 5 R6 MDA28 MDA29 8 a i MDA32 MDA31 N 4 GNDL GND l MPM ME WMI i MPUMZ MPWM3 4 4 2 CAZ41E MPUMLE MEWMIZ 6 y TROT PEAME 4z k a STEER_BOX 1 R5 J Ics Il RELAY NR R3 KF X 2 7 R mi 7 7 PARK _BRAKE 1 P I CAZHE STEER_BOX 2 n GND Figure C 1 Schematic for the adapting cicuit board for the wheel nodes 24U 12 24 12 out GND upply node x 2 REDBLACK2 ETRIGL BANSS BANS BANSS BANS3 AGNDS BANSL BAN4S BAN3 BANL AGND3 AANSS AANSZ AANSS AANS3 AGNDL AANSL AANSS AANS AANL Qs_SCK OS MISO QS_PCS2 as_pcsa GND3 OS RXD2 GND2 MPIO1S MPI013 MPIOLL MPIOS MPIOZ MPIOS UFL VFL MDALL MDAL3 MDALS MDA28 MDA38 GNDL MPWM MPWM2 MPWM16 MPWM18 WHITE2A Figure C DC1 UIN UQUT eno ral un con TC z L78S12CU 1 wut 2 DC DC CONVERTER TEM2 1221 24 1 IN QUT GREEN2A supply encoder 1 2 GREEN2B ETRIG2 BANS8 BAN56 BANS4 BANS2 AGND4 BAN58
38. BAN48 BAN2 BANG AGND2 AANSS AANSS AANS4 AANS2 AGNDB AANS AAN48 AANZ AANO 3 18 11 12 13 14 15 16 encoder bits X X 1 SP72 AP 3 4 QS_MOSI OS _PCS3 OS PCS1 GND4 encoder bits X X 9S_TXD2 OS ECK UDDH MPIOL4 MPIOL2 MPIOLO MPIO8 MPLOS UFLI UF2 VEG ea MDAI2 uA741CP MDA14 6 MDA27 MDA29 MDA31 GND MPUMI park brake su MPWM3 1 MPUMIZ 2 MPUMIS GREEN4 RED2 BLUE2 reobr ake uAZ41CP 6 RELAY2 regbpake suppl 1 2 WHITE2B WHITE2_TOP Schematic for the adapting cicuit board for Node CL 99 100 Circuit Board Diagrams x ae A N DCL 24 1 urn uouT 14 com 18 13 12 uy peak 22 IC1 UDUT 23 12 24 12 out GND L 8S12CU 1 DC DC CONVERTER 2 a z TEM2 1221 3 e S WHITE2 A ols ole SUPPLY F supply encoder 1 2 supply node 1 BLACK2 2 HI TEBLACK2 4 cz la pedal 1 GND ETRIGL ETRIG2 WHITE4B WHITE4C AEA 2 4 2 suithianso BAN58 4 4 3 1 BANSZ BANS6 3 3 2 BANSS BAN54 2 2 3 BANS3 BANS2 1 L WHITE4A AGNDS AGND4 BLUE3 BANSI BANSAdicators indicators gnd theo adi 1 2 4k7 3 4 BLUE4 F
39. Dash board LED s four in total 40 pin PowerNode connector PowerNode power supply 3 way dash board switch Accelerator pedal sensor Brake pedal senor 12 V brown 24 V red 12 V out n a GND black Board power supply 106 Circuit Board Connections a eeeeeeeooeeoeoeeoeeeeee 0 eeeeeeeeoeooeoeooeoooeoeoee Ribbon Cable 4 P waa O O a Figure D 3 Circuit board wire connections for the wheel nodes The numbers are specified in Table D 3 107 Table D 3 The description to the numbers in Figure D 3 Each connector is specified when appropriate together with the corresponding wire colours from the top and down refer to Figure D 3 Note the connection of the pressure sensor The peculiar wiring is due to a mistake when the boards were made Number 10 11 12 Description GND black Wheel angle actuator set value blue set value red signal green Linear position sensor GND white 5 V_ brown bit 13 brown green bit 12 white black bit 11 white red bit 10 white blue bit 9 white pink bit 8 white grey bit 7 white yellow bit 6 white green bit 5 white brown bit 4 violet bit 3 red blue bit 2 brown grey bit 1 brown yellow bit 0 grey pink 14 bi
40. Institutionen for systemteknik Department of Electrical Engineering Examensarbete Analysis and Design of a Redundant X by Wire Control System Implemented on the Volvo Sirius 2001 Concept Car Examensarbete utfort i Reglerteknik vid Tekniska h gskolan i Link ping av Par Degerman amp Niclas Wiker LiTH ISY EX 3365 2003 Link ping 2003 TEKNISKA HOGSKOLAN LINKOPINGS UNIVERSITET Department of Electrical Engineering Link pings tekniska h gskola Link pings universitet Link pings universitet SE 581 83 Link ping Sweden 581 83 Link ping Analysis and Design of a Redundant X by Wire Control System Implemented on the Volvo Sirius 2001 Concept Car Examensarbete utfort i Reglerteknik vid Tekniska h gskolan i Link ping av P r Degerman amp Niclas Wiker LiTH ISY EX 3365 2003 Handledare David T rnqvist Examinator Svante Gunnarsson Link ping 26 March 2003 Language Avdelning Institution Division Department Automatic Control Department of Electrical Engineering Link pings universitet S 581 83 Link ping Sweden 2003 03 26 Rapporttyp ISBN Report category Svenska Swedish Licentiatavhandling ISRN X Engelska English M Examensarbete LiTH ISY EX 3365 2003 C uppsats Serietitel och serienummer ISSN D uppsats Title of series numbering vrig rapport RL f r elektronisk version http www ep liu se exjobb isy 2003 33
41. Local Global read_swheel Swheel_Enc_Raw a Swheel_Angle Swheel_ Enc Raw check swheelJ og ynrcot Angle lt Swheel Angle Switches calculate_wheel_angles Steer_Value_ Swheel_Angle Figure 6 2 The tasks messages and how they are related for the steering subsystem in the CR node Tasks read_swheel Reads out a value from the ten bit steering wheel encoder check_swheel Checks that the value read by read_swheel is sane calculate_wheel_angles Produces the commanded wheel angles Messages Swheel_Enc_Raw The raw value as read from the 10 bit encoder that is attached to the steering wheel Swheel_Angle An agreed value representing the steering wheel angle with negative values when the steering wheel is turned left and positive when it is turned to the right The message comes repli cated from both center nodes Switches This message is used to select which algorithm the calcu late_wheel_angles task should use Steer_Value_ Commanded wheel angle for a wheel A negative value means that the wheel should be steered left and a positive that it should be steered right 6 Schedule Local Global read_clutch_pedal Pedal Clutch Raw read_brake pedal Pedal Brake Raw je Pedal Clutch Raw Pedal Clutch Value check clutch pedal Pedal Clutch Value je Pedal Brake Raw Pedal Brake Value check brake pedal Pedal Brake Value FF RR Pedal Clutch Value calculate clutch value Clutch Value Pedal Brake Valu
42. Mankan Widholm at the University workshop for helping us with the manufacturing of parts needed vii e Soren Hoff also at the University workshop for helping us with the manu facturing of circuit boards e All the people at TTTech Wienna especially Georg Stoeger Peter Rech and Petra Fierthner for being so professional and supportive e Lars Andresson PhD student at IKP FluMeS for suggestions and valuable input on electronic equipment e Katja Tasala for artistic help In addition Par would also like to thank his wife Mari Stadig Degerman for all support and understanding And Niclas would like to thank Ulf Bengtsson at IKP for tips on handling the Pro ENGINEER software package Abbreviations ABS Anti Blocking System BDM Background Debugger Mode CAN Controller Area Network CL Center Left Node CR Center Right Node DC Direct Current DSTC Dynamic Stability and Traction Control FL Front Left Node FR Front Right Node GND Ground hp Horsepower I O Input Output inc rev Increments per revolution LED Light Emitting Diode MR brake Magneto Rheological brake actuator PCB Printed Circuit Board PWM Pulse Witdh Modulated RL Rear Left Node RR Rear Right Node TDMA Time Division Multiple Access TTCAN Time Triggered CAN TTP C Time Trigged Protocol class C WCET Worst Case Execution Time VDC Volt Direct Current Contents 1 Introduction
43. NS S S encoder Sr X X m 8 i 42 1kI 4 a R3 3 OREORE t 3 4 0S_SCK OS MOSI 1a OS MISO OS PCS3 RNET BLACK4 as_pcs2 QS_PCS1 t 16 1 as_PCs GND4 L 15 2 encoder bits X X _ SND3 QS _TXD2 L 14 3 1 sui tod RXD2 OS ECK t 13 a 2 DDH i 12 a 3 MPIOL4 11 a 4 MPIOL2 18 2 5 OR2B MPIOLL MPIOL 3 a 6 tet BLACKS OR2A 2 L itch d U reobrake MPNM2 2 1 MPWM16 2 MPLIMLS MPMI switches UHITE2 TOP SU2 BLUE4_TOP 4 eobrake card 3 c3 l N oT 2 tka uA741CP L Ri 6 GND co 3 IC3 4 7k WHITE6_TOP al J9 ao ad sf lt x O 4 4 Figure C 3 Schematic for the adapting cicuit board for Node CR Appendix D Circuit Board Connections Below detailed descriptions on how to connect the different sensor and actuator wires found in each of the nodes are given 101 102 Circuit Board Connections iTi T WU MZ 83L89ANO 30 30 om 4 iE Figure D 1 Circuit board wire connections for Node CL The num bers are specified in Table D 1 103 Table D 1 Node CL circuit board connections Below is the description to the numbers D 1 The connectors are specified when appropriate from left to right in Figure Number CO CO NI Ol Ou e Co DO Fe Al col n o ja ot description Brake pedal MR brake MR brake controller card 50 pin PowerNode connector Housing fan power supply Steering wheel encoder power supply Steering wheel encoder signal output
44. a smooth signal between 0 and 10 V but due to the limitations in the PWM channels on the PowerNode the control signal has to be amplified by a factor of 2 Also since the PWM output is a square wave signal i e contains a lot of high frequency components it has to be smoothed out The control signal is therefore passed through a filter and an amplifier stage 40 As mentioned earlier the brake pedal MR brake is not connected due to the absence of a controller card However a filter and an amplifier stage can be found on the card to adapt a future control signal 2 4 Power and Electronics 27 The carburettor choke servo is controlled by two relays one for each direction The relays are supplied by the 12 V power cord and are directly connected to one PWM channel each When the relay is switched on by the PWM signal the 12 V input is transferred to the output where the servo is connected Last but not least there is the parking brake switch Here no adaptive elec tronics is needed i e the switch signal is directly passed through to the PowerNode without any components in between Do note though that the switch signal is not connected to one of the I O pins as would be expected but to one of the PWM channels Although the reason for this has not been explained please refer to 86 35 it is possible to do so since the PowerNode can be programmed to re configure a PWM channel into an I O pin if needed
45. achieve their goals all mechanical connections between the driver controls i e steering wheel and pedals and the rest of the car were removed Instead sensors and actuators were installed and connected via a distributed real time controller network 6 Inventory To be able to switch between right and left hand steering the students con structed movable modules of the steering wheel and the pedals which easily could be fitted to the left or right hand side The modules are seen lying on the ground in front of the car in Figure 1 1 As the car should be equipped with four wheel steering the rear suspension was completely modified and parts in the drive chain had to be replaced with a type that would allow the wheel to change the steering angle Also the engine suspension was modified and an actuator was fitted on one of the engine bearers The actuator created vibrations in opposition on the engine s and in that reducing the overall vibrations in the car Although the actuator still is fitted it was just tested to verify its function and has never been used since Before going on specifying the components which constitutes the X by Wire system there are some parts worth mentioning which have not been modified compared to the original car Originally the Tiger Cat E1 is composed of parts from other car models normally a Ford Sierra This car is no exception Under the glass fibre cap a 2 0 litre 4 cylinder Ford engine is found givin
46. angle error a this means that max a 3 0 a gt Wa Wmar max a 3 d 0 8 gt Wa F Wmaz max a 3 Q 0 o gt Wa 3 Wmaz max a 8 0 9 gt Wa Wmar This eliminates the need to calculate the wheel angle velocity It will also protect the system if a wheel is stuck since the other three wheels will still move although somewhat slower The assumption that the wheels have the same top speed can be made because if a wheel has significantly lower top speed for a short period of time for example 5 1 Wheel Angle Controller 49 D FS controlled wheel out Cu other wheel a Po a inv Figure 5 3 Simulink model of the bang bang controller due to a puddle of mud it is taken care of by the high sample rate If it has lower top speed for a long period of time it is probably something wrong with the wheel and it is more important for the other wheels to reach their destination quickly A Simulink model for the bang bang controller for two wheels can be found in Figure 5 3 In the model the controlled wheel angle error is first compared to the largest of the other wheel angle errors This comparison decides how the wheel angle actuator should move if another wheel has the largest angle error the speed is set according to the rules above otherwise the speed is set to the largest available control signal Wmaz multiplied with
47. ar will turn right 5 4 Braking Algorithms 57 One possible solution to this problem that maintains the open loop structure is to generate the control signal as a linear combination of the commanded force and its derivate So if r t is the commanded force and u t is the control signal that is fed to the brake motor control box as a PWM signal the expression would look like this u t A r t B Lro Where A and B are constants This design will have problems if the brake pedal is depressed or released slowly i e during normal driving conditions So it is not really usable for this system The open loop design must be sacrificed in this case and a controller using the brake pressure as feedback is used This is a bad choice since the brake pressure sensor is non redundant and since it is an analog sensor it is vulnerable to noise but unfortunately it is the only choice 5 4 Braking Algorithms The brake algorithm that is implemented in the car is a very simple one with a static brake distribution that always maintains a fixed ratio between the front and rear axle The possibilities are however almost endless for more advanced algorithms One example is to distribute the brake force differently on the outer and inner wheels when cornering The outer wheels have higher contact forces so they can be braked harder than the inner wheels A car that implements this and is available today is the top of the line Mercedes Benz SL 500 coupe
48. ard equipment Research and development in this area is however constantly increasing and several companies including Volvo Cars have been investigating the possibilities to use this technique to further enhance the car functionality for some time now The expressions Drive by Wire or X by Wire Y are often use to describe one type of enhancement considered These expressions have different meaning depending on the person asked but usually they refer to a replacement of a safety critical mechanical solution the brake system for example with a computer controlled sensor and actuator system 1 1 Project Background During autumn 2000 a final year project for the Master Students in Mechani cal Engineering at Lulea university of technology called Sirius Kreativ pro duktutveckling was initiated On commission of Volvo Cars in Goteborg the students implemented an X by Wire system into a car a Tiger Cat E1 see Fig urel 1 using a new method specially designed to consider reliability during the development process of coupled systems The project ended late May 2001 and the result was a four wheel steered car where all mechanical connections between driver and the rest of the system were replaced with sensors actuators and a dis tributed real time controller network see the Lulea Sirius project report 36 Even though the car at this point was steerable far from all the functionality the equipment al
49. brake could lock up the steering wheel or brake pedal example if the brake pedal is depressed hard and then released the feedback brake could lock the pedal in the depressed position This is naturally not the desired behavior This restriction reduces the possibilities of a realistic driver feedback even more 8Except for the brake pressure sensor but the system was originally designed for traditional open loop brake controlling i e this sensor was never planned to take part in a critical system Stiffness in the brake pedal and the forces from the road that can be felt in the steering wheel 4 7 Summary 43 All of those impairments really only leaves one realizable solution the applied brake friction should be proportional to the angular velocity of the steering wheel respectively the speed which the brake pedal is depressed with This is not really a realistic feedback but it is one solution which guarantees that the steering wheel or brake pedal will not lock up in any position 4 7 Summary This chapter has outlined the most important properties that an X by Wire system needs to have Some of the advantages over conventional cars have also been discussed Some demands on the controller structures have also been defined and these will come to good use in Chapter 5 44 Synthesis Chapter 5 Implementation In the previous chapter the properties of an X by Wire system in general and the Sirius 2001 Concep
50. ce the car behaves very differently from a conventional car and the driver will have a difficult task just controlling the car Instead some algorithm that steers the rear wheels with less extreme angles is needed In such an algorithm the relationship between front and rear steer angles can be dependent on dynamic parameters such as the forward speed or the yaw rate 54 Implementation cb 4 Figure 5 8 Explanation of true Ackermann Steering Note the difference between a versus 3 and y versus 6 5 2 Steer Algorithms 55 Figure 5 9 How a four wheel steered vehicle makes a par allel maneuver Introduce the variables f SIIrTNAPS Forward steer angle Rear steer angle Traction on the front wheels in N rad Traction on the rear wheels in N rad Distance between the mass center and the front axle Distance between the mass center and the rear axle Mass of the car Forward velocity Yaw rate Sridhar and Hatwal 15 discuss five different steering models during a lane change maneuver A Rear steer angles are kept at zero resulting in a behavior like that of a two wheel steered car B The relationship between forward and rear steer angles involving the forward velocity with CrlymV 2CfCrlr lf 1 r K bp with K ee Clm V 20 C 1 Ly lr If the front steer angles are specified this algorithm gives rear steer angles that are well adapted to
51. coder are connected to a special motor control box specified later in this chapter To protect the box from the heavy current draw when 10SKF CARN 32x200x4 38 11 Direct Current 12maxon RE 40 148867 25 13 maxon HEDL 5540 110513 29 2 3 Mechanical Components 13 starting the motor a choke 4 have been inserted between the power cables a choke consist of two iron cored coils connected in parallel and are used to increase a DC motor s terminal inductance The chokes are fitted in the car between the rear motor control boxes as shown in Figure 2 8 and on top of the front control box mounting the small gray box seen to the upper right in Figure 2 5 Figure 2 8 The rear chokes inside its protective housing The left choke no 1 is connected to the left steer actuator DC motor and the right no 2 to the right motor The chokes prevent the DC motors from damaging the control boxes by increasing the motors terminal inductance As this car is a complete X by Wire vehicle the clutch wire has been discon nected from the pedals and instead a ball screw actuator 15 has been installed to pull the wire It is mounted just above the gearbox behind the engine in the centre of Figure 2 9 the circular shaped DC motor of the actuator is shown and Figure 2 10 shows the actuator mounting above the gearbox This actuator has a senso
52. d for the motor they are connected to There are a total of four steering actuator motor control boxes 3 where two are mounted in front of the radiator the two lower ones in Figure 2 5 and control the front wheel actuators The other two are mounted behind the batteries the two closest to the batteries in Figure 2 18 and of course control the rear wheel actuators As seen in Figure 2 5 as well as in Figure 2 18 there are another four motor control boxes 4 not accounted for yet These control the brake actuator motors All of these eight control boxes can be configured to operate in different modes which specifies how the box should interpret the input or command signal The steer actuator control boxes have been set to encoder mode refer to 27 for details on how to configure the control box since that will make the control box interpret the input signal as a speed reference The box will then automatically adjust the output power so the angular velocity of the motor matches the specified input voltage The brake actuator control boxes have on the other hand been set to current mode refer to 26 for details The control boxes will in this mode interpret the input signal as a current reference and adjust the output current so the torque on the motor axle matches the input voltage The clutch actuator control box is fitted inside Node CL the black box in Figure 2 19 and
53. d inside the spindle Other parts seen in the figure are the brake caliper no 1 the armoured hose no 2 and the brake pressure sensor no 4 The actuator consist of a DC motor and a gearbox mounted on a steel block functioning as a cylinder The motor operates on a piston inside the cylinder via the gearbox and a gear wheel The original brake caliper has been connected to the other end of the cylinder via an armored hose To compensate for the increased oil volume due to brake pad ware a small hydraulic tank with a non return valve have been fitted on the side of the steel block The valve prevents the oil from going backwards through the tank when the piston is pushed forward i e when the pressure in the system is increasing Two sensors have also been installed on the block a pressure sensor has been mounted just beside the hose and a limit switch is found at the other end the 16 maxon RE 35 118777 17maxon GP 32A 110367 23 18Swedish bromsok 16 Inventory block In Figure 2 12 all the different parts of the brake actuator have been laid out note the limit switch cables running out of the cylinder block In addition to the parts mentioned above the rear DC motors have a special brake 9 directly attached on the motor axle These brakes function as a park brake by locking the axle when the power is turned off Of course the pressure must first be increased in the s
54. d somewhat to enhance the stability A Simulink model of the PlI controller can be seen in Figure 5 4 5 1 5 Result A model of the final system for two wheels can be seen in Figure 5 51 Note that the two wheels can be fed different commanded angle values the two steps to the left The two delays shown represents the delay when sending messages across the network The model was tested by choosing different commanded values for the two wheels and setting the disturbances in the variable saturation block differently for the two wheels The tests showed promising results and the wheel angles followed each other nicely even if the commanded values and or disturbances where very different In Figure 5 6 a test run was made where one wheel was fed the commanded angle of 2 solid line while the other one was fed a commanded angle of 1 dashed line The variable saturation block was fed a white noise signal for both wheels in this run 1The controllers speed is lowered due to this but since it is only used for small errors this will not affect the overall performance much 5 1 Wheel Angle Controller 51 a ak controlled wheel an out e motor_speed wheel ange FA ej R P other wheel controllert wheell 1 wheel_angles z To Workspace x y 1 z a
55. d subsystem can contain any number of components In software there are a couple of replicated subsystems e The algorithm that calculates the wheel angles are implemented in software on both center nodes and the algorithms are fed by the same input from the network e Brake values are calculated in both center nodes For more information on the software please refer to Chapters 5 and 6 Hardware wise there are also some replicated subsystem in the form of two sensors measuring the same value These can be found in the next section 4 4 Redundant Sensor Handling Some of the sensors in the car are doubled to make the readings more reliable The replicated sensors are e The steering wheel angle is measured by two independent encoders and each encoder is connected to one of the center nodes e There are two identical sensors that measure the position of the brake pedal and each sensor is connected to one center node e The wheel angles are measured in two ways directly by an absolute digital encoder and indirectly by measuring the length of the wheel angle ball screw 6The force that must be overcome to start moving an object 4 4 Redundant Sensor Handling 41 When using redundant sensors more difficulties arise Merging sensor values is a research area in itself which is called Sensor Fusion Some of the problems in this area are e How can the values be merged so that the result is in all cases better than it
56. d then using this value to generate the PWM signal The speed controlling subsystem in the wheel nodes use the tasks and messages that are described below Tasks read_velocity Reads the speed sensor for this wheel check_velocity Checks the value produced by read _velocity read_brake_pressure Reads the brake pressure sensor read_brake_limit_switch Reads the limit switch in the brake unit check_brake Checks that the brake works by comparing the old value with the new one and the limit switch reading calculate brake force Calculates the value that should be sent to the brake motor control box This task also calculates if the park brake lock should be applied actuate_brake Sets PWM value to control the brake motor and the park brake lock 6 3 Tasks and Messages 73 Local Global read_velocity Wheel_Velocity_Raw_ FR read_brake_pressure Brake_Pressure_Raw_ FR i read_brake_limit_switch Brake Limit Switch FR heel Velocity FR Wheel_Velocity_Raw_FR Wheel_Velocity_FR check_velocity Wheel _Velocity_FR Brake_Pressure_Raw_FR rake Limit Switch FR Brake Pressure_FR check brake l LV Brake_Pressure_FR Brake_Value_FR a Brake Value FR Park_Brake_FR Park_Brake_FR calculate_brake_force _ __ Brake_Force_FR Calibration Mode b k Brake_Force_FR Park Brake FR actuate brake Brake_Limit_Switch_FR Calibration Mode Figure 6 12 The tasks messages and how they are related for the spe
57. dash board with switches The upper left one is the parking brake switch no 2 and is connected to Node CL The rest of the two way switches no 3 6 as well as the single three way switch no 5 are connected to Node CR and their function is controlled by the software implementation in the PowerNode The same is true for the four LED s no 1 4 located just below the gauge indicators 2 2 3 Wheel Nodes The four wheel nodes Node RL RR FL and FR all have similar tasks i e controlling braking and steering for one wheel each The two front nodes are located on either side of the radiator see Figure 2 5 and the rear nodes are located just behind the seats see Figure 2 6 They are connected to the wheel actuators the wheel angle encoders the wheel actuator potentiometer the brake actuators and the brake pressure sensor In addition to this the two rear nodes have a special brake directly attached on the motor axle on the brake actuators see below for details and a speed sensor connected to them Rear Left Rear Right Front Left and Front Right 2 2 Network and Connections 11 Figure 2 5 Node FL no 7 and Node FR no 3 are located on each side of the core fan and the motor control boxes for the front wheel actuators are placed just in front of it The two upper boxes no 1 5 control the brake actuators and the two lower ones no 2 6 the steer actuators Also in the top
58. ds The work is exclusively done in TTPtools First TTPplan is used to define the cluster schedule and after that TTPbuild is used to define the schedule on the node level In TTPplan the properties of the cluster and the nodes are specified Also all global messages and all subsystems are specified here After that it is time to use TTPbuild to specify the tasks and node local messages A 2 3 Application Programming Now it is time to do the actual programming and implement algorithms One function for each task that is run on a node has to be implemented The TTPos provides a set of API calls that can be used to access messages and information about the cluster and the node Here some of the weak spots and immature nature of the TTP products show up Some functions are yet to be implemented and some things are a bit cum bersome to achieve There are for example possible to define several cluster modes in TTPplan but there are no way to change the currently active cluster mode This will probably change in a later release of the operating system A 2 4 Transferring schedule and applications to the cluster The schedule is transferred to the cluster with the help of another TTPtool namely TTPload This application uses a dedicated node a monitoring node or download master node which act as a bridge between the cluster and a PC with a standard Ethernet card TTPload will however only download the schedule and no
59. e Switches calculate brake values i ake Vatue Park_Brake actuate_clutch Clutch Value Calibration Mode Pedal Brake Valuej Calibration Mode Figure 6 3 The tasks messages and how they are related for the speed controlling subsystem in the CL node The speed controlling subsystem which can be found in figures 6 3 and 6 4 contains some replicated parts and some that are not The replicated part is the brake pedal handling the tasks read_brake_pedal check_brake_pedal and calculate brake values This part works just like the steer ing subsystem it first reads a raw sensor value compares this to the old agreed value and agrees on a new value This agreed value is then used to calculate the commanded brake forces The clutch and throttle handling are the un replicated part of the speed con trolling subsystem They work in a similar fashion to the braking part except that no agreement is made and no messages are broadcasted Below is a short description of all the tasks and messages involved Tasks read_clutch_pedal Gets a raw reading of the clutch pedals position this is only done on the CL node check clutch pedal Verifies that the value from read_clutch_pedal is correct only on the CL node calculate_clutch_value Calculates a value suitable for controlling the clutch actuator only on CL actuate_clutch Sets PWM value to control the clutch actuator only CL 6 3 Tasks and Messages 6
60. e if the parking brake should be applied 6 2 3 Supervision The supervision subsystem is responsible for generating a status message for a node The status is made up from the status of all the subsystems that the node runs and provides a simple way to see if a subsystem has any kind of problem In addition this subsystem is responsible for reading the dashboard switches and setting the indicators This is only done in the center nodes 6 2 4 Calibration The wheel nodes need a translation table to be able to match the readings from the wheel angle encoder and the linear sensor measuring the actuator length It also needs information of the mechanical limits for the wheel In addition the brake pressure sensor must be calibrated All these tasks are performed by the calibrating subsystem The system receives information from the center nodes if it should go into calibration mode and if it should perform any calibration action such as store left limit create lookup table etc Both of the two center nodes read the dash switches to decide if it should switch to calibration mode and generates a message to the wheel nodes It also has to store its own calibration settings for the steering wheel and the pedals 6 2 5 Driver feedback This subsystem which is only run on the center nodes controls the MR brakes that are connected to the steering wheel and the brake pedal It is supposed to make the driving experience more comfortable and somewha
61. eck if the sensor value exceeds some boundary values either in the time or in the frequency domain The sensors in the car that fall into this category are e Throttle pedal sensor e Clutch pedal sensor e Brake pressure sensor These can all be considered to be less critical than the redundant sensors discussed above since even if one of these should fail and cease to function the driver would still be able to stop the car without danger 4 6 Driver Feedback Control The purpose of this system is to make the car more comfortable to drive It is supposed to create the sensation that an ordinary car gives the driver When one drives a car one can feel the forces that are put on the front wheels through the steering wheel One can also feel the pressure in the braking system when depressing the brake pedal These are both very important factors when driving a car because it tells how the car is handling and where it is going In an X by Wire car this must be simulated using actuators that brake or induce forces in the steering wheel and the brake pedal In the Sirius 2001 Concept Car the means for driver feedback are two MR brakes This implies that only friction and no forces can be applied which make the possibilities limited If the amount of feedback i e the applied friction should depend on some external value for example the error between the commanded wheel angle and the current wheel angle there could be cases where the
62. ed controlling subsystem in the front right wheel node Messages Wheel_Velocity_Raw_ Raw value representing the period of the signal coming from the speed sensor Brake_Pressure_Raw_ Raw value from the brake pressure sensor Brake_Limit_Switch_ State of the piston in the brake actuator If this message contains a 1 the piston is at its rear limit Wheel_Velocity_ Verified value on the velocity of a wheel Brake_Pressure_ Verified value on the brake pressure Brake_Force_ The value that should be sent to the brake motor control box It is built up using the commanded brake value and whether the park brake should be activated Calibration_Mode This value is used to stop the brake actuator from moving when the system is in calibration mode 74 Schedule Local Global read_velocity Wheel Velocity Raw RL read brake pressure Brake_Pressure Raw RL read_brake_limit_switch Brake Limit Switch RL heel Velocity RLI a je Wheel Velocity RawRL Wheel_Velocity_RL check velocity EN Wheel Velocity RL Brake Pressure RLJ Brake Pressure RawRL Brake Limit Switch RL Brake Pressure RL check brake po CCCL L Brake Pressure RL lt Brake_Value_RL Park Brake RL Park Brake RL calculate brake force lt Brake Force RL Calibration Mode b ki Brake_Force_RL Park_Brake_RL actuate_brake Brake_Limit_Switch RL Calibration Mode Figure 6 13 The tasks messages and how they are related for the speed controlling subsy
63. enough no WCET calculations have been made since this is very hard to do analytically 4 In all the graphs in this section the boxes to the left are the tasks the horizontal arrows are messages and the two vertical arrows to the right are the local and the global message space Messages that end in the local message space stay on the node and those that end in the global space are broadcasted on the network Whenever a message is ended with below it means that there are actually several copies of this message One for each wheelnode replace the with FL FR RL or RR and in some cases additional messages for the center nodes replace the with CL CR 6 3 1 Center nodes In figures 6 1 and 6 2 the steering subsystem for CL and CR respectively are found This subsystem is replicated so the graphs are identical They start out by reading the steering wheel encoder and after that this value is compared to the value that the nodes agreed on in the last round to verify that the encoder is working The agreed value is then broadcasted to the network This value combined with the state of the dashboard switches to select different steering modes is then used to calculate the four commanded wheel angles Below is a more in depth description of the tasks and messages involved in the steering subsystem on the center nodes 2Worst Case Execution Time 6 3 Tasks and Messages 63
64. ensor that deviates too much from the calculated value It is also possible to detect other errors if for example the two sensor readings are about the same but differs a lot from the calculated value there is probably something wrong with the actuator or the motor The traditional way to achieve this kind of sensor fusion is to design a Kalman filter But since the wheel angle subsystem has almost no dynamics it is not needed and this simpler approach can be used instead 4 4 2 Steering Wheel and Brake Pedal Sensors When measuring signals that are not controlled by the system the task of sorting out an erroneous sensor is more difficult A predictor cannot be constructed and only a few properties of the signals can be defined For example a sensor reading that measures the steering wheel angle cannot change too quickly and it cannot exceed some mechanical limits Using this a failing sensor can be detected if it changes rapidly but not if it has a constant or slowly changing offset In addition there is no way in most cases 7When one of the sensors fail 42 Synthesis of detecting which one of the sensors that is failing just that something is wrong This makes these systems suitable for detecting errors but they cannot be used to correct errors 4 5 Non redundant Sensor Handling Without the power of redundant sensors the possibility to detect errors is pretty slim The only possibility left is to ch
65. equency ripple Do note that after the filter the pressure signal is passed through an amplifier stage The amplifier is needed because only 20 of the signal range is used the sensor output is 0 to 5 V which corresponds to a pressure between 0 and 25 MPa but according to 35 the pressure in the braking system will not exceed 5 MPa As mentioned earlier in this chapter the rear nodes have an additional speed sensor fitted which measures the speed by registering the change in a magnetic 2 4 Power and Electronics 29 field The sensor output is a very week sinusoidal signal with a frequency pro portional to the angular velocity of the wheel Since the frequency is the desired property to measure a timer channel on the PowerNode should be used and the preferred input signal is a nice square wave shifting between 0 and 5 V To accomplish that the sensor output is first amplified by a factor of 10 inside an amplifier stage and then passed through an optically coupled logic gate which creates a discrete signal with the same frequency as the input signal In contrast to all other circuits on the board this logic gate is powered by the PowerNode The wheel and brake acutators connected to this node are controlled via their motor control boxes by two PWM channels each This is due to the fact that the control boxes have one input to run the DC motor in one direction and one for the other No different form the other cards the control s
66. er 7 3 Algorithms 8 Summary and Conclusions 8 1 Future work 8 1 1 Hardware Modifications 8 1 2 Software Modifications Bibliography A Programming and Software Tool User guide A 1 The cluster A 1 1 Communications Subsystem A 1 2 Host Subsystem A 2 Using the TTPtools A 2 1 Planning 2 444 545 44448 RAR oh de ae eS 92 A 2 2 Scheduling egos S dead G4 bee we a ROS 93 A 2 3 Application Programming 0 0 000 93 A 2 4 Transferring schedule and applications to the cluster 93 A 2 5 Running and debugging the cluster 94 B System Power Schematic 95 C Circuit Board Diagrams 97 D Circuit Board Connections 101 E Manufacture Drawings 108 Contents Chapter 1 Introduction Today it is getting more and more common to replace or complement mechanical solutions with a computer based control system in order to enhance functionality Also some complex machines would not be possible to construct at all without the aid of this type of systems The Airbus A340 Boeing 777 and JAS 39 Gripen are just a few examples which all completely rely on computer based control 40 Although the vehicle industry has not yet come that far a new car already has several systems of this kind installed as stand
67. ere are essentially three steps involved when creating a TTP cluster 1 Planning 2 Scheduling 3 Application programming After that the finished schedule and application data must be transferred to the hardware to make a working system This is just an overview and a short introduction to the steps involved when creating a TTP cluster Please refer to the manuals 42 44 43 supplied with the tools for further reference A 2 1 Planning The planning is probably the most important and definitely the most time consuming step when defining a TTP cluster Since the complete system is defined A 2 Using the TTPtools 93 beforehand it is very important that this step is well thought through before con tinuing to the next This step should result in a complete description of the system This includes the different subsystems this can be seen as a logical group of functions for example seeing or hearing and all the tasks in the subsystems a task is a single function for example read a sensor or calculate a value In addition to this there should be a complete description of the messages that have to be exchanged between the tasks The result after this step should be somewhat similar to that in Chapter 6 All subsystems tasks messages and how they interact must be clearly defined A 2 2 Scheduling If the details from the planning stage are thoroughly written down this step is as easy as fill in the empty fiel
68. g about 140 hp in this light car that gives enough power to do 0 100 km h in less then 5 seconds Among other parts the Sierra has contributed with are the two DELLORTO DHLA 40 H carburettors the 5 speed gearbox and the ignition system 2 2 Network and Connections The real time controller network plays a central role in an X by Wire system and it has essentially three important tasks to perform Information exchange When turning the steering wheel you would also expect the wheels to turn The network has to provide the connection between these parts so they can communicate with each other Sensor data collection When pressing the brake pedal a sensor registers a change in the pedal angle The network then has to collect the sensor data and translate it into a reference value for the brake pressure for example Actuator control Assume that a throttle valve reference value has been set and it has been correctly transmitted to the part where the throttle valve actuator is connected The network then has to make sure that the actuator is in fact following that reference Some of the examples described above are critical for the function of the car and it is of great importance that the controller network can perform the tasks with a high degree of safety and reliability For a long time now the car industry has been and still is using the CAN protocol to communicate between different systems in a car and is is sufficient f
69. ges from tarmac to gravel 3Two or more components doing the same job 4Swedish m jligheter 5The electrical motors have different top speeds and the actuators have different friction 40 Synthesis 4 2 2 Brake Controller On a traditional car the pressure in the brake system is directly proportional to the force that is applied to the brake pedal It is the brake controllers mission to accomplish this by controlling some sort of actuator When constructing the brake controllers both speed and stability are impor tant factors The speed requirement is important since the latency between pedal force application and pressure buildup must be as short as possible As a rule of thumb this latency can be no more than 100 ms To make the design stable in all situations the rule of thumb has traditionally been to construct an open loop controller In the case of the braking system implemented in the car this is not possible since the brake actuator has very high static friction This makes an open loop controller impossible since an applied force on the piston is not directly proportional to the pressure 4 3 Replicated subsystems A replicated subsystem is defined as two identical systems with identical input It is for example used to minimize errors resulting from power outages which could prevent a processor from completing its task A redundant sensor is a special case of a replicated subsystem with only one component The replicate
70. gle_FL read_encoder Wheel_Encoder_Raw FL j read_actuator_len Wheel Actuator Len Raw_FL Wheel_Angle_FL Wheel_Actuator_Speed_FL heel_Encoder_Raw_FL Wheel Actuator Len RawFL Wheel Angle FL jqe Wheel Angle Steer_Value_ Lo calculate_actuator_speed Wheel_Actuator_Speed_FL Calibration Mode actuate wheel angle Wheel_Actuator_Speed_FL Calibration Mode check_wheel_angle Figure 6 7 The tasks messages and how they are related for the steering subsystem in the front left wheel node 6 3 2 Wheel nodes All of the four wheel nodes run the same tasks but they produce different messages For example the front right node produces a message called Wheel Angle FL while the front right node produces Wheel_Angle_FR Figures 6 7 6 8 6 9 and 6 10 list the tasks and messages for the steering sub system This subsystem starts out by reading the raw sensor values that measure the wheel angle These two values are also broadcasted for debugging purposes After that the values are checked by comparing them to each other and a predictor the check_wheel_angle task is also responsible for updating that predictor The final checked value of the wheel angle is then broadcasted In the task calculate actuator speed all of the current and commanded wheel angles are used to calculate how the steer actuator should be controlled This is where the wheel angle controller see Section 4 2 is implemented The last task s
71. he error from the other wheels This will be further discussed below To construct the two controllers a model of a wheel is needed 5 1 1 Wheel Angle Subsystem Model The system that needs to be controlled when changing the wheel angle consists of e Motor control box e Electrical motor e Ball screw actuator e The linear sensor that measures the length of the ball screw e Wheel angle encoder e Software in the PowerNodes e The network The software in the PowerNodes generates PWM signals that are fed to the motor control boxes This makes the motor turn with an angular velocity that is directly proportional to the PWM value since the control box and the motor are themselves a closed loop system This removes nearly all of the motor dynamics and simplifies the model a lot When the motor turns the length of the ball screw is altered and this in turn makes the wheel twist A fairly correct model for all this is a simple integrator with some delays in it This is reasonably correct since the system from PWM value to angle alteration has nearly no dynamics and the PWM value should be proportional to the wheel angle velocity The amount of delay can be found by looking through the data sheets for the different components 1 8 25 127 30 38 41 i Motor control box In the data sheet the bandwidth is found to be 2 5 kHz gt 3500 4 10 4 0 4 ms Motor The data sheet states that the time constant is 5 ms
72. he throttle servo does not need any adjustment but a 5 V voltage regulator has been fitted to supply the servo with power As with Node CL there are adapting circuits for the MR brake controller card but in contrast to Node CL a card has actually been installed it is mounted directly on the circuit board as seen in Figure 2 21 The wheel node circuit boards see Figure 2 22 have been prepared for five sensors and three actuators each although some of them are not used in the front nodes The simplest sensor to adapt is the wheel angle encoder i e the encoder signal is passed through to the PowerNode directly without any circuits in between The only component needed is the 5 V voltage regulator which supply power to the encoder The analog linear position sensor and the brake pressure sensor are also supplied by the same voltage regulator p Ea coa copa ga gg CE pesesssssscsooeecccci d RON maa Aa TA Figure 2 22 The adaptive electronic circuit board for the wheel nodes The components are as follows operational amplifiers no 1 relay no 2 optically coupled logic gate no 3 5 V no 4 and 12 V no 5 voltage regulators DC DC converter no 6 Both these analog signals are filtered through the same type of filter used everywhere else to reduce high fr
73. ibuted real time controller network The network which not only renders the engineer almost endless possibilities to control how the car should behave and feel when driven it also has drawbacks due to the increased complexity of the system that follows This chapter starts out by looking at some possibilities as well as drawbacks and safety issues that an X by Wire system have compared to a normal mechan ical system In addition demands on the systems involved are specified throughout this chapter These demands will be needed in Chapter 5 4 1 Possibilities and Drawbacks In an X by Wire system the dynamics can be completely altered in software All that is needed in principle is to write an algorithm telling the system what it should do and when In a car this could be Using optimal wheel angles Today all cars have toe angles built into the wheel suspension to control the stability in acceleration or heavy break ma noeuvres In all other cases toe angles are just a drawback since they de crease the expected life time of the tires To be able to dynamically control when to use toe angles and when not to could considerably reduce the owners tire cost Parallel manoeuvres Under certain conditions a lane change on the freeway or parking for instance it could be desirable to move sidewise without having the car turning Stability control during strong side wind When driving a car in strong side wind the dr
74. ics Models B and C both produce extreme under steer However the only realizable four wheel steering model of the ones listed above would be B since the car has no yaw rate sensor The yaw rate could however be calculated by using the speed sensors to mea sure the difference between the speed of the inner and outer wheels This would imply that all wheels must roll without slipping that assumption can only be made for small velocities 31 so the calculations will not produce a reliable value on the cars yaw rate 5 3 Brake Force Controller The demands on the brake controller is mainly stability The function of the brake system must also stay unaffected even if a sensor should fail Therefore it is often common to use an open loop controller However the brake actuators in the car have so high static friction that this is not possible To design a usable system the controller must be able to overcome this friction and at the same time guarantee stability Apart from the static friction the system is very simple The control signal is sent as a PWM signal to the brake motor control box The box is set for current control so the PWM signal is proportional to the current and therefore the torque on the motor axle This torque is transferred to the piston via a gear box When the piston moves the pressure is built up in the braking system and the brake pads are pressed against the brake disc 21f the front wheels turn left the re
75. ignals are passed through a filter and an amplifier stage before leaving the circuit board There is however one exception the signal to retract the brake actuator is not amplified 35 Lastly also just implemented in the rear nodes a relay has been fitted to control the park brake lock on the brake actuator motors one for each actuator The relay is connected similar to the relays used in Node CL The only difference is the power supply 24 V instead of 12 V 41 An I O port which can among other things register the frequency of a signal 42Schitt trigger HCPL2200 30 Inventory Chapter 3 Modifications Although much effort had been put down during the re configuration of the car at Lulea some solutions have been found that need to be reviewed or modified In this chapter some of the more important ones are discussed which all have the common aim to improve the overall system behaviour Both mechanical and electrical modifications have been performed on the car First a description of the problem and how it affects the system is presented This is followed by a presentation of the chosen solution 3 1 Steer Actuator Joint In the original version of the car all of the wheel actuators were joined to the spindles with a ball and socket joint which allowed the outer part of the actuators to twist Since it is a ball screw actuator the length is controlled by a motor acting on a thread inside Turning or twisti
76. implementera ny mjukvara i bilens datorer for att pa sa s tt kunna g ra bilen anv ndbar Bilen r en Tiger Cat E1 som r modifierad s att den r fyrhjulsstyrd och anv nder sig av ett avancerat x by wire system Datorerna som bygger upp x by wire systemet r sex stycken TTP PowerNode som kommunicerar med varandra ver ett feltolerant och redundant TTP C n tverk Datorerna r ocks anslutna till ett antal sensorer och aktuatorer f r att kunna kontrollera bilen Projektet har bidragit till bilen p flera s tt Ett r den kompletta dokument ationen ver de olika systemen i bilen ett annat r en programmeringsmanual som betydligt s nker inl rningstr skeln f r vidare projekt Slutligen har flera f r ndringar i b de h rd och mjukvara f rb ttrat bilens anv ndbarhet och belyser en del av de m jligheter som erbjuds i ett system av den h r typen Resultaten visar att konceptbilen Sirius 2001 har stor potential som en platt form f r ytterligare f rskning inom omr dena fordonsdynamik och feltoleranta system Vidare har ocks TTP C protokollet visat sig motsvara de krav som st lls i X by wire system Preface This Master Thesis project was initiated in the beginning of September 2002 and this report is the result after its almost 24 week duration The project has been carried out at the Department of Mechanical Engineering IKP Link pings universitet However the authors has also registered the project at the Depar
77. iver constantly has to compensate the wind forces that act on 1A small angle on the wheel so that two wheels on an axle are not completely parallel to each other 37 38 Synthesis the car in order to keep the car on the desired path This could instead be done automatically by the system Skid control Some of the new cars today already have skid control systems like the DSTC system in Volvo XC90 but they all try to stabilize the car by braking only The performance could probably be improved by using both steering and braking with an X by Wire system The four wheel steering also makes it possible to maintain control of the vehicle even if the front wheels are locked in a panic brake situation All things presented in the list above cannot be realized in this car in its present state However just small additions in hardware could make these things and more possible The primary drawbacks in computer based controller systems regards reliability and safety These issues make it difficult to guarantee correct operation in all situations Pure mechanical systems have a big advantage in that respect faults are often both easier to spot at an earlier stage in the design process and fewer in quantity Figure 4 1 points out some of the areas where safety issues have to be considered in addition to the pure mechanical ones Communication Figure 4 1 Possible source of faults in the X by Wire system drawing made
78. ke Raw Raw value from the brake pedal Pedal Clutch Value Verified value on the clutch pedal 66 read_switches Switches Schedule Local Global Switches Calibration Mode actuate_indicators Node Status_ supervise cnode Pedal Clutch Value T _ Node_Status_C Pedal_Brake_Raw Switches calibrate cnode _ lt Calibration Mode Swheel_Angle Pedal_Brake_Value Pedal_Brake_Value Swheel_Enc_Raw Pedal Clutch Raw Swheel Angle Pedal Brake Value calculate feedback Swheel Feedback Brake Feedback actuate feedback K Swheel Feedback Brake Feedback Figure 6 5 The task messages and how they are related for the supervising calibrating and feedback subsystems in the CL node Pedal Throttle Value Verified value on the clutch pedal Pedal Brake Value Agreed value from the brake pedal This message is produced in both center nodes Clutch Value Value that represents the commanded position on the clutch actuator Throttle Value Value that is the commanded position on the throttle servo Brake Value xx Commanded brake force for wheel FL FR RL and RR Park_Brake_ This message contains a 1 if the park brake on wheel FL FR RL or RR should be applied Otherwise it is 0 Calibration_Mode This is used to prevent the different actuators from moving when the system is in calibration mode The supervising calibrating and feedback subsystems can all be found in fig
79. libration Mode Figure 6 16 The task messages and how they are related for the supervising and calibrating subsystems in the front right wheel node The last two subsystems calibrating and supervising are described in figures 6 15 6 16 6 17 and 6 18 They each contain a single task Tasks supervise_wheelnode Checks the nodes status by analyzing messages that this node produces calibrate_wheelnode Listens for calibration commands that are sent out by the center nodes and takes appropriate action one of store limit create a table that translates between actuator lengths and wheel angles store zero angle and store calibration values for the brake pressure sensor 6 3 Tasks and Messages 77 Local Global Brake_Pressure_RL Wheel_Angle_RL A x ig Wheel_Velocity_RL supervise_wheelnode puma T Node_Status _RL Calibration Mode Wheel_Encoder_Raw_RL Brake Pressure RawRL Wheel_Actuator_Len_Raw_RL Calibration Mode calibrate wheelnode Figure 6 17 The task messages and how they are related for the supervising and calibrating subsystems in the rear left wheel node Local Global Brake_Pressure_RR Wheel Angle RR Wheel_Velocity_RR Node_Status_RR Calibration Mode Wheel_Encoder_Raw_RR Brake Pressure RawRR Wheel_Actuator_Len_Raw_RR Calibration Mode supervise_wheelnode calibrate_wheelnode Figure 6 18 The task messages and how they are related for the supervising
80. lowed was implemented in software 1The X in X by Wire could be replaced by Brake Steer or Clutch 2A collection of sub systems which depend on each other in order to function 2 Introduction Figure 1 1 The Tiger Cat E1 X by Wire prototype car figure taken from 36 The car was soon after moved to Volvo Cars in Goteborg where a couple of ad ditional projects were performed before it arrived to the Department of Mechanical Engineering Link pings universitet At this point the car was no longer functioning the rear wheels had been disconnected due do a replacement of sensors which had not been tested and the front wheels had almost a life of their own when driving the car Also the software implementation of algorithms and regulators needed a fair amount of work This Master Thesis project was initiated in order to fix these problems and get the car up an running 1 2 Purpose The purpose of this report is not only to describe the work done during this project it should also serve as a shorthand introduction to the car in order to lower the threshold for future projects As the substance of the report is of a technical character the intended reader is a person with an engineering background On the other hand anyone with an interest in the possible future of the vehicle industry would also benefit from it 1 2 1 Objective The main objective of this project is to modify the car so a
81. n a well known phe nomena when using the Ziegeler Nichols method To reduce that the parameters were adjusted a couple of times resulting in the following values K 2 T 1s The switching between Bang Bang and PI controller was set fixed at 1 75 At this angle error the Bang Bang and PI controller give the same control signal in the wheel with the largest angle error i e the wheel that moves with the maximum 79 80 Results speed This makes the transition between the two controllers smooth if the control signal is close to the maximum value Next the second rear node was programmed with the same controller in order to verify the coordination between the wheels when the bang bang controller was in action In the test two different step references were used in nodes one going form 1000 to 1000 increments and the other from 0 to 1000 The two signals were initiated at the same time and the wheels were unloaded during the test The result seen in Figure 7 1 was satisfactory and agrees quite well with the simulated one in Figure 5 6 The wheel with the largest error was running at maximum speed the whole time while the other had adapted its speed so both wheels reached their destination at the same time 1000 800 600 Relative angle oO 200 400 600 800 1000 0 0 5 1 1 5 2 25 3 3 5 4 4 5 5 Time s Figure 7 1 Step responses for the rear wheels with different
82. n Rheonetic http www lord com http www rheonetic com MR DRUM BRAKE ASSY Filename reo broms_ RD 2028 18X jpg LORD Corporation Rheonetic http www lord com http www rheonetic com MRB 2107 3 Product Bulletin Filenamet MRB_2107_3_2002_20_0 pdf LORD Corporation Rheonetic http www lord com http www rheonetic com Rheonetic Wonder Box Device Controller Kit Filename RD_3002_03_46_0 pdf M Bruce Distribuerad Brake By Wire based on TTP C Master s thesis Department of Automatic Control Lund Institute of Technology Lund Swe den May 2002 maxon gear http www maxonmotor ch index_a cfm maxon gear Plane tary Gearhead GP 32 A Filename GP32 A_110367 pdf maxon motor http www maxonmotor ch index_a cfm maxon DC motor RE 35 Filename RE35_118777 pdf maxon motor http www maxonmotor ch index_a cfm maxon DC motor RE 40 Filename RE40_148867 pdf maxon motor http www maxonmotor ch index_a cfm maxon motor control 4 Q DC Servoamplifier ADS 50 5 Filename ADS 50 5 145391 pdf maxon motor http www maxonmotor ch index_a cfm maxon motor control ADS 50 10 Filename ADS 50 10_201583 pdf Bibliography 89 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 maxon motor http www maxonmotor ch i
83. nclusions detect faults Section 4 4 and 4 5 includes a discussion on this topic but as with the algorithms above it has not been implemented in software 8 1 Future work There are still areas of the car that could be improved further This section will list some of these 8 1 1 Hardware Modifications Brake Actuators The manual parking brake makes the rear brakes stick since brake fluid is drawn through the non return valve which prevents the brake fluid from going out through the fluid container There are two solutions to this problem one simple and one requiring a lot more effort The simple one is to make the actuators resemble an ordinary brake systems i e drill a new hole for the brake fluid container positioned in such a way that the piston blocks it right before pressure is built up in the system but makes the container and the system in contact with each other when the piston is retracted The speed of the brake actuator can also be questioned A solution which would work better with regard to this is to have a constant pressure container with a valve regulating the pressure to the brake caliper The solution requiring more effort involves a complete redesign of the brake actuator and is suitable for a Master Thesis project by its own Dashboard The driver interface is limited to LED s and simple switches A display would be more appropriate to let the driver know what is going on with the car
84. ndex a cfm maxon motor control Choke on plug in Card Filename choke 133350 pdf maxon tacho http www maxonmotor ch index_a cfm Digital Encoder HEDL 55_ with Line Driver RS 422 Filename HEDL 5540_110514 pdf Motorola Inc http www motorola com MPC555 556 User s Manual Filename START pdf Nielsen L and Eriksson L Course material Vehicular Systems Bokakademin Link ping Link pings universitet Sweden 2002 Ostergrens Elmotor AB http www ostergrens se Elektriska bromsar FSB serien Filename Elbroms FSB003 htm P Bjorklund and P Drougge Distribuerad Brake By Wire demonstrator baserad pa TTP C kommunikation Master s thesis Department of Com puter Engineering Chalmers University of Technology Goteborg Sweden January 2000 Saia Burgess http www saia burgess com Switches Filename Mi croswitches pdf Sirius Project group 2001 Sirius 2001 Brake Sytem and Wheel Suspension May 2001 This reoprt contains a more detailed description of the modified brake system and wheel suspension Sirius Project group 2001 Sirius 2001 Main Report May 2001 This report is a summary of the five parts of the project SKF Linear Motion http www skf se CAPR 43 Filename kulskruv CAPR 43Ax100x2A1G3F pdf SKF Linear Motion http www skf se CARR Linear Actuators File name kulskruv CARN 32x200x4 pdf SKF Linear M
85. ng the inner part with respect to the outer without the motor running will produce the same result as when one screws a nut on a threaded bolt This could be seen as disturbance or back lash in the control system The simplest solution for this is to replace the ball and socket mount with a universal joint In Figure 3 1 the old ball and socket joint to the left is seen together with the new one to the right Since no commercially manufactured universal joints could be found that sat isfied the requirements on dimensions the joints were manufactured by the Uni versity workshop In Figure 3 2 two views of the 3D model of the joint made in Pro ENGINEER are seen This model was the base from which manufacturing drawings were made see Appendix E Please note that no stress calculations have been made on the joints Normally this should for course be included es pecially for a safety critical part like this However after a discussion with the experienced personal at the workshop the conclusion was made that the material choice and thickness should be sufficient for the time being 14 CAD Computer Aided Design software package provided by the company PTC http www ptc com 31 32 Modifications Figure 3 1 The modified steer actuator end joints the old ball and socket joint left and the new universal joint right Figure 3 2 An exploded left and a
86. nt state unprotected so finding or designing a protective housing for the control boxes alone would not increase the system endurance by much Since suitable locations had been found means of attachment had to be con structed Using Pro ENGEINEER 3D models and manufacturing drawings see Appendix E of appropriate attachments where produced and handed in to the University workshop for fabrication In Figure 3 5 and 3 6 the 3D models of the attachments are seen Figure 3 5 Two views of the 3D model of the front wheel control boxes one explodes view left and one assembled right Figure 3 6 The 3D model of the rear wheel control boxes one ex plodes view left and one assembled right 3 4 Wheel Nodes Circuit Boards The wheel node circuit boards had to be reviewed due to a number of reasons First as mentioned earlier in this chapter the front wheel encoders had been replaced by another type with higher resolution Second the speed sensors at the rear wheel spindles had never actually been connected into the nodes and to do that new components where needed Third the old rear wheel node boards had 36 Modifications a circuit fault the PWM signal that supposed to retract the piston in the brake actuator was connected directly to ground Instead to continue to add and repair the old circuit boards a new layout was designed To save time the old layout was used as a template Only
87. of the figure the steer actuator choke box no 4 is seen Figure 2 6 The rear nodes are placed just behind the seats Node RL no 5 behind left seat and Node RR no 2 behind the right The rear brake actuators are also located here outside each node no 1 4 The two small holes in the middle are the battery master switches for the 12 V no 3 and 24 V no 6 systems respectively 12 Inventory 2 3 Mechanical Components The car is equipped with a number of sensors and actuators to be able to control the system In the following sections the different components are listed together with a short description of where they are mounted in the car and what function they have in the system Also some important technical specifications are listed in tables 2 3 1 Actuators There are three different types of actuators performing different types of tasks linear ball screw actuators for linear motion servos for rotational motion and MR brakes for force feedback To change the steering angle four identical ball screw actuators are mounted at each wheel connecting the steering spindle to the frame The motions of the actuators are made possible by DC motorg 2 with an encoder 3 mounted on the outer end of the motor axle see Figure 2 7 Figure 2 7 The rear right steer actuator DC motor no 1 The encoder no 2 is seen mounted on the outer end of the motor axle The motor and en
88. om them will be exactly the same Each wheel node then recieves its commanded wheel angle and tries to control the actuator so that the wheel achieves this angle In addition every wheel node has to check that all of the wheels are turning with respect to each other The wheel nodes also have the responsibility to read the two sensors that measure the current wheel angle and report this to the rest of the system 6 2 2 Speed Controlling In this subsystem the two center nodes read out how the pedals are positioned These values are then used to control the clutch actuator for the CL node the throttle servo for the CR node and to calculate brake values for the four wheels The brake values are calculated in both nodes but clutch and throttle are only considered in one node each This design choice was made since if for example the CL node should fail there would be no way of controlling the clutch actuator or read out the clutch pedal value since these are only connected to this node This makes the speed controlling subsystem in the centre nodes partially replicated The wheel nodes use the brake value calculated by the centre nodes to control the brake actuators This subsystem also has to read out the speed and brake 6 3 Tasks and Messages 61 pressure of each wheel Currently the speed can only be read from the two rear wheels since there are no speed sensors on the front ones In addition the dash board switches are monitored to decid
89. or the applications used today However it lacks many requirements especially re garding safety needed in a distributed safety critical real time system like this i e lhorse power 2 2 Network and Connections 7 steer and brake by wire applications To meet these requirements the TTP C protocol was developed some ten years ago by the Vienna University of Technol ogy and Daimler Benz Research Later in 1998 an Austrian company TTTech was formed to develop tools and hardware using the TTP concept For a more comprehensive introduction to the history of the TTP C protocol please refer to The base of this concept is the TT Tech C1 PowerNode 41 see Figure 2 1 which is equipped with TTTech s own TTP C C1 network controller chip for in formation exchange and a Motorola embedded Power PC processor for sensor data collection and actuator control Figure 2 1 The TTTech Cl PowerNode which form the base of the network The two large circuits on the board are the Power PC pro cessor no 1 and the TTP C C1 network controller no 2 A network or a cluster is composed of two or more PowerNodes which are connected to each other via a broadcast data bus using either a bus or a star network topology 42 and communicate using the TTP C protocol The network in this car consists of six PowerNodes located at strategic places see below for detailed locations which are connected to
90. otion http www skf se Control unit CAED ANR File name motorkontroller CAED ANR bmp Henrik Thare Sakerhetskritiska realtidssystem 1999 M lardalen Real Time Research Center TTTech Computertechnik AG http www tttech com A TTP Develop ment Board for the Time Triggered Architecture 1 3 00 edition December 2000 Filename TTPPowerNodeV1 3 pdf TTTech Computertechnik AG http www tttech com The Cluste De sign Tool for the Time Triggered Protocol TTP C 3 2 edition July 2002 Filename TTP Plan user man3 2 pdf 90 43 TTTech Computertechnik AG Bibliography http www tttech com The Download Tool for the Time Triggered Protocol TTP C 4 4 0 edition July 2002 File name TTP_Load_user_man4 4 0 pdf 44 TTTech Computertechnik AG http www tttech com The Node Design Tool for the Time Triggered Protocol TTP C 3 2 edition July 2002 File name TTP Build user man3 2 pdf 1 Available on the included CD ROM Appendix A Programming and Software Tool User guide In the following chapter some fundamental issues about the design behind TT Tech s products are discussed as well a short introduction on how to program the cluster using TT Tech s software products TTPtools Some reflections on schedul ing can also be found in this chapter A 1 The cluster The cluster consists of two or more nodes connected to each other by a communi
91. peed This time was then used to calculate the speed by which the current wheel would have to move in order to reach its destination at the same time Introduce the variables a Wheel angle errors for the four wheels Wa WB Wg wW Angular velocities for the four wheels Using this the time that a wheel takes to reach its destination i e when the wheel angle error equals zero can be calculated as a B 0 0 Ta TB To Wa WB We We The goal is that these times should all be equal to the largest time a 6 amp T Max Ta T8 Th To Wa WB Wo We This is accomplished at the wheel with the wheel angle error a like max Ta TB To To Ta Wa Wmax max Ta T6 To TO TB Wa 3 WB max Ta T6 To T0 To Wa F We max Ta T6 To To To Wa 9 Wo And similarly for the other wheels Using this technique all wheels will reach their destination at the same time even if they have to travel different lengths and have different properties It does however involve some calculations that can be eliminated Suppose that all four wheels have the same top speed This implies that the wheel which has the largest angle error also will need the largest time to reach its destination which in turn means that instead of comparing the time it would take for the wheels to reach their destination one could compare the angle errors directly For the wheel with wheel
92. phovsr tten vid en senare tidpunkt kan inte upph va detta tillstand All annan anv ndning av doku mentet kr ver upphovsmannens medgivande F r att garantera ktheten s ker heten och tillg ngligheten finns det l sningar av teknisk och administrativ art Upphovsmannens ideella r tt innefattar r tt att bli n mnd som upphovsman i den omfattning som god sed kr ver vid anv ndning av dokumentet p ovan beskrivna s tt samt skydd mot att dokumentet ndras eller presenteras i sadan form eller i s dant sammanhang som r kr nkande f r upphovsmannens litter ra eller konstn rliga anseende eller egenart F r ytterligare information om Link ping University Electronic Press se f r lagets hemsida http www ep liu se In English The publishers will keep this document online on the Internet or its possi ble replacement for a considerable time from the date of publication barring exceptional circumstances The online availability of the document implies a permanent permission for anyone to read to download to print out single copies for your own use and to use it unchanged for any non commercial research and educational purpose Subsequent transfers of copyright cannot revoke this permission All other uses of the document are conditional on the consent of the copyright owner The publisher has taken technical and administrative measures to assure authenticity security and accessibility According to intellectual
93. purpose is simply to generate the PWM signal for controlling the motor control box The tasks and messages in the steering subsystem on the wheel nodes are described below Tasks read_encoder Reads a raw value from the fourteen bit digital encoder that measures the wheel angle read_actuator_len Reads out a raw value from the linear potentiometer that measures the length of the steer actuator check_wheel_angle Compares the value produced by read_encoder and the value from read_actuator_len to a predicted value of the wheels angle 70 Schedule Local Global read_encoder Wheel_Encoder_Raw_FR read_actuator_len Wheel_Actuator_Len_Raw_FR Wheel Angle FR Wheel_Actuator_Speed_FR check wheel angle lt Wheel_Encoder_Raw_FR Wheel_Actuator_Len Raw_FR i _ _ Aa9 oa_a Wheel Angle FR aaa Wheel_Angle_ Steer_Value_ i calculate_actuator_speed fm Wheel_Actuator_Speed_FR Calibration Mode actuate wheel angle Wheel Actuator Speed FR Calibration Mode Figure 6 8 The tasks messages and how they are related for the steering subsystem in the front right wheel node Local Global read_encoder Wheel_Encoder_Raw_ RL fr read_actuator_len Wheel_Actuator_Len_Raw RL Wheel_Angle RL Wheel Actuator_Speed_RL check_wheel_angle Wheel_Encoder_Raw_RL Wheel Actuator Len Raw RL fm Wheel Angle RL aaa Wheel Angle Steer_Value_ calculate actuator speed fm Wheel Actuator Speed RL
94. r fitted inside which measures the actuator length by detecting the position directly on the moving nut The DC motor is mounted on top of the actuator and together with the sensor it is connected to the clutch actuator control box Some specifications regarding the ball screw actuators mentioned above are listed in Table 2 1 14maxon 133350 28 I5SKF CAPR 43Ax100x2A1G3F D24C 14 Inventory Table 2 1 Specifications for the ball screw actuators Figure 2 9 The clutch actuator seen from above The circular shaped DC motor is seen in the middle of the figure Figure 2 10 The clutch actuator viewed from underneath the car Behind the aluminium gearbox the mounting of the actuator is seen 2 3 Mechanical Components 15 Except for the DC motors mentioned above another type is used in the braking system Here the conventional master cylinder normally mounted just in front of the braking pedal on the separation wall between the engine and the driver compartments has been replaced Instead a system made up of four independent hydraulic pumps has been designed 35 and fitted close to each wheel The front wheel pumps have been mounted inside each steering spindle see Figure 2 11 and the rear ones behind the seats next to the rear nodes see Figure 2 3 Figure 2 11 The front left steer spindle notice the brake actuator pump no 3 in the middle of the figure fitte
95. ra58 s pdf HEWLETT PACKARD http www elfa se Low Input Current Logic Gate Filename HCPL2200 pdf HiTEC RCD USA Inc http www hitecrcd com ANNOUNCED SPEC IFICATION OF HS 805BB MEGA 1 4 SCALE SERVO Filename hs805 pdf HiTEC RCD USA Inc http www hitecrcd com General Servo Informa tion Filename Servomanual pdf Dan Holt Brake by wire Service Tech Magazine pages 18 20 January 2002 87 88 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Bibliography J Bolin and J Hedberg Implementation of a Distributed Control Application Based on the TTT C Architecture Master s thesis Department of Computer Engineering Chalmers University of Technology Goteborg Sweden March 1999 J Sridhar and H Hatwal A Comparative Study of Four Wheel Steering Models Using the Inverse Solution Vehicle System Dynamics 21 1 18 1992 Department of Mechanical Engineering Indian Institute of Technology Kanpur Leine amp Linde http www leinelinde se Mounting instructions File name mount manual_synchro flange pdf Leine amp Linde http www leinelinde se PARALLEL 670 671 File name 670 671parallel_db_eng pdf LORD Corporation Rheonetic http www lord com http www rheonetic com M R BRAKE ASSEMBLY Filenamet MRB2107 3WEB pdf LORD Corporatio
96. reference values To further test the bang bang controller tire friction was added to one of the wheels The test was very simple the wheel with the largest error was prevented from turning by trying to hold the tire by hand in a firm grip Although this is an unscientific testing procedure it gave a quantitative understanding about the behavior of the controller The result seen in Figure 7 2 shows that the controller indeed could adapt to the change in conditions and confirm the simulated results in Figure 5 7 The wheel with the lesser error started to slow down when it came closer to the reference value since the other wheel angle error did not diminish as fast as expected Due to the shifting to the PI controller the time when the references were reached 1One increment is aot degrees 7 2 Brake Controller 81 differ a little A small notch can also be seen in the graph when the controllers where switched This notch could be removed by implementing a more advanced algoritm which decides when the controller switching should be made 1000 800 600 Relative Angle Oo 200 400 600 800 1000 26 27 28 29 30 31 32 Time s Figure 7 2 Step responses with tire friction simulation for the rear wheels The car was elevated so the wheels were not in contact with the ground during all of the tests described above The tests above show that the wheel moves 40 in around 3 5 s This
97. rner in the figure fitted on top of the ball screw actuator is the linear position sensor no 2 In contrast to the front wheel the rear wheel encoders have not been directly attached to the spindles due to lack of space Instead they have been moved towards the centre of the car and measure the wheel angle via a linkage see Figure 2 17 25 Hengstler RA58 S 8 9 26 Swedish lankarm 2 3 Mechanical Components 21 Figure 2 17 The figure shows the rear right encoder no 1 with linkage no 3 to the spindle the linear position sensor no 2 on top of the ball screw actuator and the speed sensor no 4 fitted inside the spindle partly concealed by the brake disc 27 Another type of encoder with a resolution of 10 bit is found mounted in the moveable steering module see Figure 2 14 Here two identical encoders have been mounted next to each other to measure the steering wheel angle The encoders are in turn connected to different nodes one to Node CL and one to Node CR In Table 2 3 some additional specifications regarding the absolute shaft en coders are found Table 2 3 Specifications for absolute shaft encoders Leine amp Linde 670 Hengstler RA58 S Steering wheel angle Wheel angle 10 bit 1024 inc rev 14 bit 16384 inc rev Code switching fre max 50 kHz max 100 kHz quency Supply voltage 9 30 VDC 5 VDC Max current con
98. roadster which has a system called Sensotronic Brake Control This system is reviewed in more detail in 13 58 Implementation Chapter 6 Schedule This chapter outlines the steps involved in scheduling this particular real time system This involves defining subsystems tasks and messages 6 1 The Time Triggered Architecture the TTP C Protocol Before describing the schedule a short introduction to the TTP C is appropriate For a more thorough explanation and the history of TTP C 22 is recommended TTP C is a communication system of class C and must therefore be strongly deterministic This means that the behaviour of the system must be known be forehand and some rules must be set up for it These rules should include how and when both communications and computations are made All the rules put together form the schedule 6 2 Subsystems A subsystem can be seen as a set of tasks or functions that all collaborate to achieve a more complex function In a TTP C cluster a subsystem can be com pletely replicated partially replicated or non replicated A replicated subsystem is run by several nodes in the cluster and this makes the system dependable even if a node should fail In an X by Wire application it is logical to divide the system into steering and speed controlling subsystems In the car all of the nodes must run both these subsystems since they are all involved in controlling both steering and the speed
99. s 1 3 Report Structure The report is structured as a working procedure for the system at hand The chapters describe the different steps involved when working with the system and their order resembles to the order in which the steps should be performed i e before a controller structure can be made demands on system performance are needed and a schedule cannot be made unless the all task are defined which in turn require detailed knowledge about the system This is important to keep in mind specially for anyone who intends working with the system Chapter 2 gives an brief introduction to the car and its history However the focus is on the installed parts and their function in the car Chapter 3 treats mechanical and electrical modifications that have been per formed during the project Chapter 4 considers how to make all parts work together as a whole This involve a discussion on possibilities and drawbacks with a safety critical systems as well as considerations regarding algorithms and controllers 3Time Triggered Protocol class C 4Controller Area Network 5Time Triggered CAN 6The FlexRay protocol was developed by the FlexRay Group which started as a co operation between BMW and DaimlerChrysler in 1998 4 Introduction Chapter 5 treats the actual construction and implementation of the algorithms and controllers used
100. so reses rss sr rss rss 4 2 2 Brake Controller ooo sor rr rr rr rss ss rss sr 4 3 Replicated subsystems oms ss res rse ee 4 4 Redundant Sensor Handling ue ee ee Gee Go a AA ee 4 4 1 Wheel Angle Sensors 0 0 000005 Contents 4 4 2 Steering Wheel and Brake Pedal Sensors 4 5 Non redundant Sensor Handling 4 6 Driver Feedback Control 4 7 Summary 5 Implementation 5 1 Wheel Angle Controller 5 1 1 Wheel Angle Subsystem Model 5 1 2 Wheel Angle Predictor 5 1 3 The Bang Bang Controller 5 1 4 The PI Controller 5 1 5 Result sh cc 5 2 Steer Algorithms 5 2 1 Ackermann Steering 5 2 2 Steer Angle Distribution 5 3 Brake Force Controller 5 4 Braking Algorithms 6 Schedule 6 1 The Time Triggered Architecture the TTP C Protocol 6 2 Subsystems 6 2 1 Steering 6 2 2 Speed Controlling 6 2 3 Supervision 6 2 4 Calibration 6 2 5 Driver feedback 6 3 Tasks and Messages 6 3 1 Center nodes 6 3 2 Wheel nodes 7 Results 7 1 Wheel Angle Controller 7 2 Brake Controll
101. ssembled right view of the manufactured universal joint The 3D model of the joint was made in Pro ENGINEER 3 2 Front Wheel Encoders All wheel encoders had originally a resolution of only 10 bit but it was soon discovered not to be enough So new 14 bit encoders were purchased but for some reason only the rear wheel encoders were actually fitted at that time This left the front encoders unchanged which needed to be corrected Also the encoders are mounted on the upper lever arm of the front wheels and the encoder shaft points downward and meets a rod welded to the spindles see Figure 3 3 When the suspension moves up and down the rod and the encoder shaft moves in respect to each other Earlier the rod and the encoder were connected with a piece of gasoline tubing to allow for this movement However this tube could flex significantly which introduced disturbances in the encoder signal 2Swedish lankarm 3 3 Wheel Actuator Control Boxes 33 One way to correct this problem is to move the encoder below the upper lever arm and connect it to the spindle using some kind of linkage similar to how the rear wheel encoders are fitted see Figure 2 17 Another way is to use some other type of connection The first solution might be better because it can be difficult to find a connection that has all the degrees of freedom i e axial radial and angle displacement to the extent that is needed On the other
102. stem in the rear left wheel node 6 3 Tasks and Messages 75 Local Global read_velocity Wheel_Velocity Raw_RR read brake pressure Brake_Pressure Raw RR read_brake_limit_switch Brake Limit Switch RR heel Velocity RR je Wheel_Velocity_Raw_RR Wheel_Velocity_RR check_velocity Wheel_Velocity_RR Brake Pressure RR Brake Pressure RawRR Brake Limit Switch RR Brake Pressure RR check brake EN Brake Pressure RR Brake Value RR Park_Brake_RR Brake_Value_RR Park_Brake_RR calculate_brake_force Brake_Force_RR Calibration Mode b k Brake_Force_RR Park Brake RR actuate_brake Brake_Limit_Switch_RR Calibration Mode Figure 6 14 The tasks messages and how they are related for the speed controlling subsystem in the rear right wheel node 76 Schedule Local Global Brake_Pressure_FL Wheel_Angle_FL A A supervise wheelnode Wasel Veloeity Fu e Node Status_FL Le Wheel_Encoder_Raw_FL Brake_Pressure_Raw_FL calibr ate_wheelnode Wheel_Actuator_Len_Raw_FL Calibration Mode Calibration Mode Figure 6 15 The task messages and how they are related for the supervising and calibrating subsystems in the front left wheel node Local Global Brake Pressure FR Wheel Angle FR 3 Wheel_Velocity_FR supervise_wheelnode See C _Node Status FR Calibration Mode li h Wheel_Encoder_Raw_FR Brake Pressure Raw FR Ca ibrate_w ee node Wheel_Actuator_Len_Raw_FR Ca
103. t Car in particular were defined In this chapter these properties and demands are used to design controller structures for the individual subsystems 5 1 Wheel Angle Controller The basic design goal behind this controller was to make as much use of the power in the actuators as possible while maintaining the stability of the system Another demand that needed to be fulfilled was that when moving the wheels must move coordinated even if they have different properties and commanded angles A bang bang controller fulfills the basic demand because it uses the largest available control signal 6 This makes the actuators move as fast as possible at all times The bang bang controller is by design very sensitive to noise in the measured signals and can easily oscillate To remedy this another controller is often used for small errors In this case a bang bang controller was used for large errors and a Pl controller which was designed using Ziegler Nichols 5 was used for small errors 1 controlled wheel controlled wheel 2 other wheel other wheel bang bang In Out Pl reg Figure 5 1 Simulink model of the wheel angle controller 45 46 Implementation The wheel angle controller can be found in Figure 5 1 Note that the PI controller is fed a zero input until it is switched in to prevent the integrator from winding up The other input to the bang bang controller is t
104. t similar to that of a conventional car The possibilites are rather limited since only brakes which only can produce friction and not motors which also can produce torque are fitted in the car If these brakes were to be replaced with motors a lot more advanced feedback could be implemented 6 3 Tasks and Messages The subsystems are further divided into distinct tasks which would be easy to implement in software Each task is then implemented as a single function when programming The tasks exchange data with each other via messages that can be global or node local The tasks and messages are scheduled in a TDMA round Tasks are run once every second TDMA round and messages have a lifespan of two TDMA rounds The TDMA round was chosen to 10 ms 1Time Division Multiple Access 1 62 Schedule Local Global read_swheel Swheel_Enc_Raw A a Swheel_ Angle Swheel_Enc_Raw check swheel tcl Angle Swheel_Angle Switches calculate_wheel_angles Steer Value_ Swheel_Angle Figure 6 1 The tasks messages and how they are related for the steering subsystem in the CL node In the schedule each task is allocated a time budget and a deadline The time budget states how much processor time the task is allowed Tasks that involve calculations are allowed 200 us and other tasks are allowed 100 us The deadline specifies when in the TDMA round the task execution must be finished The time budgets are chosen to be
105. t the application This must be transferred directly to each node via a so called BDM cable The BDM cable puts the host processor in a special debug mode which allows access to the nodes flash memory 2 All cluster modes have different schedules and thus can send another set of messages and run other tasks One use could be to set the cluster in some sort of service mode which allows for debug output and sensor calibration 94 Programming and Software Tool User guide A 2 5 Running and debugging the cluster After the schedule and application data are transferred the cluster should be ready to be started If everything works each node should blink a green led to indicate that it has contact and is synchronized with the network Debugging and monitoring can be done with a tool called TTPview which gives access to all of the global messages in the cluster These can be shown as different kinds of instruments such as LEDs and graphs Each node can additionally be debugged with a debugging software and the BDM cable mentioned earlier This approach does not however give access to the information on the network Appendix B System Power Schematic This appendix contains an overview of the power system in the car Note that only the parts involved in the X by Wire system is specified 95 96 Shunt RL steer Rear fuse box RL brake RL Amp meter Monitor RR brake RR steer RR gt Pp rr ee Se Se Dashboard 2
106. t wheel angle encoder GND black 5V red 50 pin PowerNode connector Board power supply aD Va 7 4 24V red GND black PowerNode power supply 424V red Speed sensor Parking brake lock signal red Pressure sensor GND green supply red connect this to the signal pin GND black Brake actuator set value blue set value red signal black Limit switch GND grey 5V blue pressure sensor blue 40 pin PowerNode connector Appendix E Manufacture Drawings In this Appendix all the manufacture drawings of the fabricated parts have been included In Figure E 1 to E 4 the parts of the modified steer actuator joint are shown and E 5 to E 9 displays the attachment parts of the front actuator control box The last two figures Figure E 10land E 11 shows the parts of the rear control box attachment 108 109 mm Figure E 1 Steer actuator universal joint the joint fork Manufacture Drawings a Lx 45 Chamfer mm Figure E 2 Steer actuator universal joint the joint centre cube 111 45 Figure E 3 Steer actuator universal joint the screw that joins the fork to the centre cube 112 Manufacture Drawings Figure E 4 Steer actuator universal joint the clevis pin used to attach the joints to the front wheel spindles 113
107. the adaptive electronic circuit boards all the 3Swedish bagtandskoppling 4Swedish balgkoppling 34 Modifications actuator control boxes were also fitted inside see Figure 3 4 As the control boxes generate a fair amount of heat and probably a lot of electrical noise they had to be moved Figure 3 4 The Node FR housing before the actuator control boxes where moved in front of the radiator The best solution in our opinion was to locate them as close as possible to the actuator they control i e brake and steering actuators As there are not much free space in the car where the boxes could be moved the positions below were the only practical alternatives Rear nodes A fair amount of space could be found just behind the batteries All four boxes could be mounted with a reasonable distance to the actuators see Figure 2 18 Front nodes After some investigation it appeared to be enough room just in front of the radiator fan suitable for the four boxes controlling the front wheels see Figure 2 5 5 Apparently the control boxes in the front nodes generated so much heat that the PowerNode overheated and ceased to function 3 4 Wheel Nodes Circuit Boards 35 With these locations some kind of additional housing to protect the control boxes were needed since the electrical connections otherwise would be unpro tected However most of the other X by Wire equipment is at the prese
108. the car C A model where the rear steer angle depends on the front steer angle and the yaw rate r K 6 C1 V 6 with K 1 and Ci m lr ly ITER Cy C 56 Implementation In addition to algorithm B this one takes the cars yaw rate into account to prevent the car from skidding D The rear steering depends on the product of the yaw rate and forward velocity Sig m lp ly r C2 V0 with Cp a 2 1 C Cy Like algorithm C the yaw rate is taken into account when calculating the rear steer angles This one has however a different nonlinear relationship since it depends on the product between forward velocity and yaw rate E The front and rear angles are individually and optimally controlled hypothet ical model Note that algorithm A through C give negative values for the rear steer angles when the forward speed is low and algorithm D gives positive values for all speeds If the rear steer angles are negative the center of rotation is placed inside the car This will lower the minimum turning radius thus making the car more maneuver able If the angles are positive then the center of rotation is placed outside of the car which will lower the yaw rate and therefore the risk of skidding Sridhar and Hatwal 15 come to the conclusion that model D has the best characteristics from the point of view of steering effort It is also the only of the models that produces neutral steer characterist
109. tion model works well in an application like this X by Wire Steer by Wire Brake by Wire Redundant TTP C Concept Car Abstract The purpose of this master thesis project has been to analyze and document the Sirius 2001 Concept Car In addition it has also been a goal to get the car in a usable state by implementing new software on the on board computers The car is a Tiger Cat El that is modified with four wheel steering and an advanced X by Wire system The computers in the X by Wire system consist of six TTP PowerNodes that communicate with each other over a redundant fault tolerant TTP C communications bus The computers are connected to a number of sensors and actuators to be able to control the car This project has contributed to the car in several ways A complete documen tation of the systems implemented in the car is one Another is a programmers manual which significantly lowers the threshold when working with the car Last but not least is the modifications in hardware and software which have made the car usable and show some of the possibilities with the system The results show that the Sirius 2001 Concept Car is a suitable platform for research in car dynamics and fault tolerant systems The work has also shown that the TTP C communication model works well in an application like this Sammanfattning Syftet med det har examensarbetet ar att analysera och dokumentera konceptbilen Sirius 2001 Ett annat mal har varit att
110. tment of Electrical Engineering ISY and as a consequence two versions of the report is found Still except for the title page their contents are the same Together with the report a CD ROM has been included It contains valuable information for anyone with the intention to continue to work with the system Among the included substance are hardware specification sheets and programming code The entire report has been created using the IATEX 2e package Par Degerman Applied Physics and Electrical Engineering program Niclas Wiker Mechanical Engineering program Link ping march 2003 Acknowledgment During our work we have been in contact with a lot of people helping us in many ways First of all we would like to thank our examiners prof Svante Gunnarsson ISY and prof Karl Erik Rydberg IKP and supervisors David T rnqvist ISY and Johan Andersson IKP for support during the project and useful comments on the report A special thanks goes to Christian Grante at Volvo Cars who has been struggling a lot to supply us with the tools needed for programming the network as well as valuable information on the history of the car Also worth mentioning are our opponents Jonas Elvfing and Mikael Littman who have provided us with suggestions on the content as well as the structure of the report In addition the following people have been an invaluable support throughout the whole project e Thorvald Tosse Thoor and Magnus
111. tor and the brake pedal MR brake The MR brake is however not actually plugged into the node but the wires are prepared for prospective connection Other equipment connected to the node are the clutch pedal sensor the parking brake switch the upper left switch in Figure 2 4 one brake pedal sensor and one of the steering wheel encoders Figure 2 3 The centre nodes are located on each side of the black fuel tank in the middle of the figure Node CL no 2 to the left and Node CR no 4 to the right Also seen in the figure are the carburettor choke no 1 i e the cold start enrichment device and the fuse box no 3 which contain power supply fuses for the centre nodes the front nodes as well as for the actuators connected to these 2 2 2 Node CR Node CR is located opposite to Node CL the box to the right in Figure 2 3 and controls the steering wheel MR brake the throttle servo and the four LED s in the middle of the dash board see Figure 2 4 The other brake pedal sensor the throttle pedal sensor the second steering wheel encoder and the rest of the dash board switches all except the parking brake switch are also connected to this node Please note that all switches are two way i e ON OFF except for one which is three way 5Central Left Magneto Rheological brake Central Right 8Light Emitting Diode 10 Inventory Figure 2 4 The
112. turation Integrator Figure 5 2 A somewhat simplified model of a wheel angle system A Simulink model of this can be seen in Figure 5 2 where the block called variable saturation is a block that restricts how fast the wheel can move to model friction and other disturbances A more complete view of two wheels can be seen in Figure 5 5 where the delays of the network messages can be seen 5 1 2 Wheel Angle Predictor The simple model that was defined in the last section is used to verify the sensor readings If this model is fed the same signals as the actual wheel it would generate a good estimate of the wheel angle This predicted value is used to verify that the values read from the two wheel angle sensors are correct If the sensors deviate too much they are invalidated 5 1 3 The Bang Bang Controller The bang bang controller type was chosen because it makes as much use of the hardware as possible by always using the largest possible control signal To make the wheels move synchronized the control signal that the bang bang controller would generate was adjusted so that the wheels would reach their com manded angle at the same time This was implemented by taking the wheel with 48 Implementation the greatest difference between commanded and actual angle and calculating how much time it would take for this difference to vanish if the wheel traveled with its maximum s
113. uipment differs between the PowerNodes three different types of circuit boards are found in the car one type for the wheel nodes and one type each for Node CL and CR Although there are some differences between the four wheel nodes Node FL and FR lack speed sensors and the parking brake lock is only implemented in Node RL and RR they still have the same type of circuit board All circuit boards are supplied by the 24 V system and the PowerNodes are in turn supplied by the cards However in the two centre nodes Node CL and CR a 12 V power cord is added to supply the MR brake controller cards the steering wheel encoders and the carburettor choke servo Another common factor between the different boards is the analog filters used They all are first order low pass filters3 with a cut off frequency at 16 Hz Two different types of operational amplifiers are also used on the boards Al though they differ in the allowed temperature range and surrounding components their task with one exception see below is the same to amplify a signal by a factor of about 2 To supply power to these amplifiers all boards have a 12 V voltage regulator and a DC DC converter fitted The circuit diagrams of the installed board types are found in Appendix C and in Appendix D a detailed description on how to connect the different wires to the circuit board is found 36Pulse Width Modulate A square
114. ures 6 5 and 6 6 Supervising is done by checking the status of all messages that are produced by the node The resulting message is calculated as the sum of the messages from CL and CR The supervising subsystem also reads out the dashboard switches and controls the indicators to inform and warn the driver 6 3 Tasks and Messages 67 Local Global read_switches Switches di A Switches Calibration Mode a actuate_indicators Node Status_ Swheel_Angle Swheel_Angle Pedal Brake Value supervise cnode Pedal_Throttle_Value Pedal_Brake_Value Node Status C Swheel_Enc_Raw Pedal Throttle Raw Pedal_Brake_Raw Switches calibrate_cnode ent Calibration Mode fa FF Swheel Angle Pedal Brake Value calculate feedback Swheel Feedback Brake Feedback actuate feedback Swheel_Feedback Brake Feedback Figure 6 6 The task messages and how they are related for the supervising calibrating and feedback subsystems in the CR node The tasks and messages in the supervising subsystem on the center nodes are Tasks read_switches Creates a message containing bits for all switches on the dash board actuate_indicators Controls the various indicators on the dash board supervise_cnode Checks the total status of the node by analyzing the messages that the other tasks on this node produces Messages Switches State of the dashboard switches Each bit in the message represents one switch if the switch is
115. useable and functional concept prototype is obtained To achieve this the following goals have been set e The different parts included in the control system should be identified and well documented e The implemented controllers and algorithms should be modified so the car behaves in a consistent manner when driven 1 3 Report Structure 3 e The system should be able to handle redundant components in order to detect faults These goals should be implemented in hardware and software in such a way that the car can be used as a platform for laboratory or research projects in the future 1 2 2 Limitations As the objective definition is fairly open and the available time limited a few limitations on the scope of the project are applied First no evaluation of the installed network or the protocol is performed See 14 33 where the TTP C3 and the CAN4 protocols are compared and 22 for a brief introduction to alternatives to TTP C TTCAN FlexRay etc As the first of the above goals states only the parts which build up the X by Wire system are covered in detail All other parts like the engine power train ignition system etc are only mentioned briefly The implementation of fault tolerance is also restricted to manage only fault detection not handle any failure modes This means that the system should be able to detect if for example a sensor is malfunctioning but not react in any special way if or when that happen
116. wave signal where the average voltage is controlled by changing the width of the pulse 37 The filter is composed of one 2 2 uF capacitor and a 4 7 kQ resistor 38The voltage regulator stabilizes an input between 15 to 35 V into a constant output of 12 V 39 One input signal of 12 V is converted into two output signals one 12 V and one 12 V 26 Inventory The circuit board for Node CL is fitted inside the node housing and is seen in Figure 2 201 The connected digital steering wheel encoder has an output signal voltage of 10 V which has to be reduced to 5 V before going into the PowerNode s I O pins This is done by letting the 10 bit signal pass through a resistor bridge and an electronic protection circuit Figure 2 20 The circuit board with adaptive electronics for Node CL Starting from the upper left corner the following components are pointed out operational amplifiers no 1 relays no 2 12 V voltage regulator no 3 resistor bridge no 4 and protection circuit no 5 for the encoder signal DC DC converter no 6 To reduce high frequency ripple from the analog brake pedal sensor the signal passes through a low pass filter the same is true for the clutch pedal sensor These two sensor signals are connected to the analog ports on the PowerNode The only actuators actually connected to this node is the clutch actuator via its own control box and the carburettor choke The required input to the box is
117. with negative values rep resenting that the wheel should be steered to the left and positive values for right Wheel_Actuator_Speed_ The value that should be sent to the motor control box A negative value mean that the wheel should be turned left and a positive that it should be turned right Calibration Mode This is used to prevent the actuator from moving when the system is in calibration mode 72 Schedule Local Global read_velocity Wheel Velocity Raw_FL read_brake_pressure Brake_Pressure_Raw_FL read_brake_limit_switch Brake Limit Switch FL Wheel_Velocity_FL A Wheel Velocity Raw FL Wheel_Velocity_FL E I Wheel Velocity FL Brake Pressure FL Brake Pressure Raw FL Brake Limit Switch FL Brake Pressure FL check brake Oo _ Prake Pressure FL Brake_Value_FL Brake Value FL Park Brake FL Park Brake FL calculate brake force bBrake Force FL Calibration Mode k Brake_Force_FL Park Brake FL actuate_brake Brake_Limit_Switch_FL Calibration Mode Figure 6 11 The tasks messages and how they are related for the speed controlling subsystem in the front left wheel node In figures 6 11 6 12 6 13 and 6 14 the speed controlling subsystem is specified It begins by reading the speed sensor and the brake sensors After that those values are checked and the validated messages are broadcasted The second part controls the brake motor by first calculating a control value an
118. wn slot in the TDMA round For this to work there must be some kind of scheme which all the nodes can access which defines the TDMA round and the messages sent This scheme is called the MEssage Descriptor List MEDL The MEDL is stored within each node and must be consistent on all nodes in the cluster To create the MEDL TTTech has supplied a tool called TTPplan This tool aids in the creation and definition of a cluster The user defines the nodes in the cluster and which subsystems they should run In addition messages that subsystems produce are defined in the temporal that is when and the value that means type and size Using this information TTPplan can produce a communications scheme which in turn is the basis for the MEDL TTPplan also includes utilities to view and modify the schedule graphically which significantly simplifies the task A 1 2 Host Subsystem Each node in the cluster runs TT Tech s own proprietary operating system TTPos The OS hides some of the complexity involved in communicating with other nodes and handles the scheduling for tasks that run on the host processor The schedule for a node consists of rules when a task is run and information about how much CPU time it is allowed This schedule is made with another tool from TTTech namely TTPbuild TTPbuild takes a cluster database created by TTPplan as input and lets the user define individual tasks for the subsystems on a node A 2 Using the TTPtools Th
119. ystem before the axle is locked but that is handled by the control system The motor power cables are in turn connected to a motor control box specified later in this chapter similar to the ones used for the wheel actuator motors Figure 2 12 The different parts of the rear left brake actuator pump Starting from the left the protective housing no 3 for the DC motor and the gear box is seen followed by the parking brake lock mechanism no 1 Thereafter comes the DC motor no 2 and gear box no 4 with the gear wheel no 5 just below Then there is the cylinder block no 10 with the limit switch no 7 in the top end and the pressure sensor no 6 together with a nipple no 12 for connecting the hose in the other The last three parts are the piston no 8 the hydraulic tank no 9 and the non return valve no 11 In Table 2 2 specifications regarding DC motors and used gearboxes when appropriate are found 19Ostergrens Elmotor AB FSBO003 32 2 3 Mechanical Components 17 Table 2 2 Specifications regarding the DC motors and gear boxes Property maxon RE 40 maxon RE 35 SKF D24C Fcon Steering Braking system Cuich Powerrating w OW na Nominal voltage 24 VDO 30 VDO avoe Gear box ratio n a J maxon GP 324 231 n a Noloadspeed T580rpm oo na _ Max continuous Motor control maxon ADS maxon ADS SKF CAED box

Analysis and Design of a Redundant X-by

Contents

Download Pdf Manuals

Related Search

Related Contents