“Fly by Wire” Clutch actuator for Aston Martin DB5
A collaborative project by:
Wren Classics and The Fusion Workshop.
Wren Classics were approached by a DB5 owner who was struggling to operate the standard clutch due to it being quite heavy. The customer had an accident in the 1960s, when his left leg had been damaged during a motorcycle race, leaving it weakened.
A heavy clutch can be difficult to live with for anyone, particularly on a long journey. Early Astons were notorious for having relatively heavy clutches as the gearboxes were quite agricultural compared to modern cars.
The owner had tried more common solutions, but nothing could produce a clutch action light enough.
Being a true classic car, the owner placed certain restrictions on what could be done so our objectives were:
Given the above restrictions it was decided to use an electric, linear actuator operated by the clutch pedal and controlled by an electronic control system.
Phase 1:
An industry standard P.I.D. (Proportional Integral Derivative) servo control system was implemented to give precision feedback for both position and speed control.
P.I.D. control is a common method used to regulate the dynamic behaviour of a system. Examples are found in many industrial devices, where it’s employed for control of temperature, pressure, flow, speed, or position, to name a few typical applications.
Brief overview of a P.I.D. system.
Proportional
Integral
Derivative
Some key elements of the system are a Set Position input (the setpoint, or our target position for the linear actuator), a pulse-width-modulated (PWM) signal of some Duty Cycle which drives the actuator, and the Current Position of the actuator.
It’s called closed-loop control because a feedback loop relays information about the current state back into the system, allowing it to obtain the difference between this current state and the desired setpoint, which must be corrected.
In our case the Clutch sensor sends the desired “Setpoint”
The sensor mounted on the linear actuator sends the “Current position”
The micro processor calculates the error between the “Setpoint” and the “Actual position” This is known as the “Following error” The controller outputs a correction signal in the form of a PWM pulse train that determines the direction that the actuator must move and the speed it must move to reduce the error between the two signals.
The larger the “Following error” the faster the actuator moves. In the case of our project, the faster the clutch pedal is moved, the faster the actuator moves until the “Following error” is reduced to zero.
This is a continuous process with the clutch pedal being monitored every 20ms and small corrections made as required.
By constantly maintaining control, the “Following error” is actually never allowed to become large. The motor is held in position with very small voltage adjustments by switching the supply to the motor very quickly (20KHz – 20,000 times/sec) and so the current supply to the motor is very small, less than 0.5A, only increasing when a large, fast change of position is required. Immediately the position is found, the current to the motor drops to very close to zero.
In the software we have complete control of speed of response of the system and therefore the “feel” of the gear change.
A basic bench test was done to prove the functionality:
A simple motor driving a feedback potentiometer and a second control to simulate the clutch pedal. This arrangement was sufficient to prove the basic code.
The existing feedback mechanism was replaced with a new one to be compatible with the new control system.
This motor was a much higher current motor than the test rig so a dual output 10Amp “H Bridge” driver was used to drive the actuator.
A collaborative project by:
Wren Classics and The Fusion Workshop.
Wren Classics were approached by a DB5 owner who was struggling to operate the standard clutch due to it being quite heavy. The customer had an accident in the 1960s, when his left leg had been damaged during a motorcycle race, leaving it weakened.
A heavy clutch can be difficult to live with for anyone, particularly on a long journey. Early Astons were notorious for having relatively heavy clutches as the gearboxes were quite agricultural compared to modern cars.
The owner had tried more common solutions, but nothing could produce a clutch action light enough.
Being a true classic car, the owner placed certain restrictions on what could be done so our objectives were:
- To significantly lighten the clutch pedal force required to activate the clutch plate in the gearbox while maintaining the original driving experience.
- This is a classic car and every effort should be made to avoid changes to the interior or exterior of the car.
- This removes the possibility of changing the gearbox to a fully automatic or to have a paddle type gear change arrangement.
- Previous, more traditional methods have been tried to achieve the goal: changing the clutch plate mechanism to a roller bearing type and to change the Master / Slave cylinder ratio.
- Unfortunately, with the latter, the mechanics of hydraulic control would mean that to achieve less force at the pedal more pedal travel would be required. This was not possible without changing the physical dynamics of the gear change.
Given the above restrictions it was decided to use an electric, linear actuator operated by the clutch pedal and controlled by an electronic control system.
Phase 1:
- To design a servo system to control an actuator by connecting a sensor to the existing clutch pedal to read the position and to translate this position to a control signal to move the actuator to operate the original hydraulic master cylinder.
- To create a simple proof of concept model in order to code the software for the electronic system.
An industry standard P.I.D. (Proportional Integral Derivative) servo control system was implemented to give precision feedback for both position and speed control.
P.I.D. control is a common method used to regulate the dynamic behaviour of a system. Examples are found in many industrial devices, where it’s employed for control of temperature, pressure, flow, speed, or position, to name a few typical applications.
Brief overview of a P.I.D. system.
Proportional
Integral
Derivative
Some key elements of the system are a Set Position input (the setpoint, or our target position for the linear actuator), a pulse-width-modulated (PWM) signal of some Duty Cycle which drives the actuator, and the Current Position of the actuator.
It’s called closed-loop control because a feedback loop relays information about the current state back into the system, allowing it to obtain the difference between this current state and the desired setpoint, which must be corrected.
In our case the Clutch sensor sends the desired “Setpoint”
The sensor mounted on the linear actuator sends the “Current position”
The micro processor calculates the error between the “Setpoint” and the “Actual position” This is known as the “Following error” The controller outputs a correction signal in the form of a PWM pulse train that determines the direction that the actuator must move and the speed it must move to reduce the error between the two signals.
The larger the “Following error” the faster the actuator moves. In the case of our project, the faster the clutch pedal is moved, the faster the actuator moves until the “Following error” is reduced to zero.
This is a continuous process with the clutch pedal being monitored every 20ms and small corrections made as required.
By constantly maintaining control, the “Following error” is actually never allowed to become large. The motor is held in position with very small voltage adjustments by switching the supply to the motor very quickly (20KHz – 20,000 times/sec) and so the current supply to the motor is very small, less than 0.5A, only increasing when a large, fast change of position is required. Immediately the position is found, the current to the motor drops to very close to zero.
In the software we have complete control of speed of response of the system and therefore the “feel” of the gear change.
A basic bench test was done to prove the functionality:
A simple motor driving a feedback potentiometer and a second control to simulate the clutch pedal. This arrangement was sufficient to prove the basic code.
The existing feedback mechanism was replaced with a new one to be compatible with the new control system.
This motor was a much higher current motor than the test rig so a dual output 10Amp “H Bridge” driver was used to drive the actuator.
|
|
A board was designed to go inside the original housing, replacing the OEM pcb.
In this phase we were essentially just replacing the clutch pedal with the actuator and driving the original master cylinder. An adaptor was designed and 3D printed to connect the actuator body to the master cylinder and to house the clutch position sensor. All in a body that replicated the original master cylinder in order to blend in with the original mechanics.
In this phase we were essentially just replacing the clutch pedal with the actuator and driving the original master cylinder. An adaptor was designed and 3D printed to connect the actuator body to the master cylinder and to house the clutch position sensor. All in a body that replicated the original master cylinder in order to blend in with the original mechanics.
|
|
This new arrangement was all mounted beneath the car and connected to the gearbox slave cylinder.
After many trials we concluded that this actuator did not have the force required to reliably operate the clutch despite changing the master/slave hydraulic ratio. It was decided to abandon this phase and begin with another option.
After many trials we concluded that this actuator did not have the force required to reliably operate the clutch despite changing the master/slave hydraulic ratio. It was decided to abandon this phase and begin with another option.
Phase 2:
A new linear actuator was sourced. Many thanks to the team at Hollin Applications for their help with modifying the actuator with different feedback and gear ratios. This one had twice the calculated required force of 2200N (220Kg)with a max of 4900N (490Kg). With this much force available it was decided to bypass the hydraulic system completely and drive the gearbox clutch arm directly from the actuator.
As there was now no direct link between the driver and the clutch arm, the servo system was in sole control of operating the clutch and therefore became true “Fly by Wire” Certainly a first for a DB5 and research has not found an equivalent system working anywhere.
The software was suitably modified to control this new cylinder and an uprated motor driver installed. The new driver has a 20 Amp continuous drive capability and a Max short burst capability of 60Amps. So far, no tests have seen more than 7 Amps required.
The clutch sensor is rated for 2 Million cycles.
With this new arrangement, a separate enclosure was required for the electronics. This is now mounted beneath the car along side the actuator. Custom bracketry was designed and fabricated in house by Wren. As the actuator is connected directly to the clutch arm allowance was made for the arc travel during operation and some compliance built in to avoid potential binding. Safety features were added to ensure that the installation is safe and secure.
The enclosure is water proof to IP67
Testing proved that the system could easily operate the clutch mechanism with no sign of stalling or current draw above 2 Amps in normal use.
At this stage, external controls were added for possible future maintenance use. Two IP67 rated external button/LEDS were added and the software modified to use these. By pressing on the buttons, the actuator can be manually driven in or out and the LEDs continually give visual feedback of the motor drive direction and current draw.
The software contains over 500 lines of custom code with complete control of responsiveness, speed, bite point control and manual over ride buttons.
Phase 3:
Road tests.
Real world driving found that the “bite point” of the clutch was too low. The software was changed to give full control over the position of the clutch pedal when the actuator starts to move. We can move the “bite point”, the point where the clutch is just engaging, to suit the driver’s requirements.
A small control hidden out of site enables the driver to move the "bite point" from close to the carpet, to the top of the clutch pedal movement.
Many miles of road testing in very severe wet weather showed that the system is fully sealed against water ingress.
Gear changes can be made as quickly as required without lag in the system. A must-have as this car may compete in classic car competitions in the future.
The driver can control the car exactly as before: Hill starts, holding on the clutch, gentle pull away, or a wheel spinning jump start. Everything you would want from a classic Aston, only much lighter. The "Leg feel" is changed by simply adding or removing return springs from the clutch pedal.
The installation fits in beautifully under the car and is still protected behind the original cover panel (Not shown).
A new linear actuator was sourced. Many thanks to the team at Hollin Applications for their help with modifying the actuator with different feedback and gear ratios. This one had twice the calculated required force of 2200N (220Kg)with a max of 4900N (490Kg). With this much force available it was decided to bypass the hydraulic system completely and drive the gearbox clutch arm directly from the actuator.
As there was now no direct link between the driver and the clutch arm, the servo system was in sole control of operating the clutch and therefore became true “Fly by Wire” Certainly a first for a DB5 and research has not found an equivalent system working anywhere.
The software was suitably modified to control this new cylinder and an uprated motor driver installed. The new driver has a 20 Amp continuous drive capability and a Max short burst capability of 60Amps. So far, no tests have seen more than 7 Amps required.
The clutch sensor is rated for 2 Million cycles.
With this new arrangement, a separate enclosure was required for the electronics. This is now mounted beneath the car along side the actuator. Custom bracketry was designed and fabricated in house by Wren. As the actuator is connected directly to the clutch arm allowance was made for the arc travel during operation and some compliance built in to avoid potential binding. Safety features were added to ensure that the installation is safe and secure.
The enclosure is water proof to IP67
Testing proved that the system could easily operate the clutch mechanism with no sign of stalling or current draw above 2 Amps in normal use.
At this stage, external controls were added for possible future maintenance use. Two IP67 rated external button/LEDS were added and the software modified to use these. By pressing on the buttons, the actuator can be manually driven in or out and the LEDs continually give visual feedback of the motor drive direction and current draw.
The software contains over 500 lines of custom code with complete control of responsiveness, speed, bite point control and manual over ride buttons.
Phase 3:
Road tests.
Real world driving found that the “bite point” of the clutch was too low. The software was changed to give full control over the position of the clutch pedal when the actuator starts to move. We can move the “bite point”, the point where the clutch is just engaging, to suit the driver’s requirements.
A small control hidden out of site enables the driver to move the "bite point" from close to the carpet, to the top of the clutch pedal movement.
Many miles of road testing in very severe wet weather showed that the system is fully sealed against water ingress.
Gear changes can be made as quickly as required without lag in the system. A must-have as this car may compete in classic car competitions in the future.
The driver can control the car exactly as before: Hill starts, holding on the clutch, gentle pull away, or a wheel spinning jump start. Everything you would want from a classic Aston, only much lighter. The "Leg feel" is changed by simply adding or removing return springs from the clutch pedal.
The installation fits in beautifully under the car and is still protected behind the original cover panel (Not shown).
The Replacement Master cylinder on the left side that now holds the clutch position sensor and the control box.
|
The actuator in position.
|
The Original prototype (on the left) and the final version with custom made PCB on the right.
|
This system is completely customisable for any vehicle that requires a "Fly by Wire" upgrade, although the actuator mounting will need adapting to suit each car. Our next project is looking as though it could be a 4 1/2 litre "Blower" Bentley from the 1920s! Or it might be something like an AC Cobra, who knows..
So if you are struggling with a heavy clutch in your car, give Wren Classics a call on 01747 852899 and talk to them about a fly by wire solution for your car to bring the pleasure back into driving.
If you have a project coming up that TFW might be able to help with, please contact me through the "Contact" page and I'll be glad to try and help.
So if you are struggling with a heavy clutch in your car, give Wren Classics a call on 01747 852899 and talk to them about a fly by wire solution for your car to bring the pleasure back into driving.
If you have a project coming up that TFW might be able to help with, please contact me through the "Contact" page and I'll be glad to try and help.
Final version with laser engraved front facia
Waiting for facia
Initial prototype without case
|
wiring almost complete
Prototype with 104 pin
loom attached
Display panel wiring
Wiring complete with 104 pin connector hooked up
|
Zytek DB7 ECU tester
My regular client - Wren Classics - www.wrenclassics.com is one of the pre eminent Aston Martin restorers in the country and we have worked together on some very interesting projects over the last two years. This latest one pushed the boundary's that's for sure! The Aston DB7 has an ECU based around a Ford unit from the 90's. Ahead of its time with the level of engine management coupled with a separate EDIS unit to manage the ignition advance. A car not starting can be many things, but the ECU and the alarm are often the ones most difficult to diagnose without actually substituting the unit. This particular ECU has been obsolete and virtually unobtainable for many years. With less than 200 in circulation finding one to try is very difficult indeed. Wren receives ECUs from around the world to prove that they are defiantly the root cause of a particular problem, but up until now they have needed a DB7 in the workshop to test them. Something not always possible. Andrew, the chief engineer at Wren asked me to design and build a one off custom tester for this particular ECU. It needed to be able to mimic the sensor signals sent from the car engine, the EDIS module and the alarm system. The tester needed to simulate different engine RPM under idle and full load conditions. Air and coolant temperatures, exhaust lambda sensors etc and display the resulting outputs from the ECU under test. All in a simple to use, bench top tester that would be robust enough for a workshop environment. Many hours were spent collecting data from a running DB7 car. 50 data points were measured at different RPM and load and from these the final inputs and outputs were decided. Crank and Cam angle sensors, Tacho and TPS (throttle position sensor) all needed to be electronically generated as complex pulse trains. Sensors inputs like air, coolant, exhaust Lambda, MAP (manifold air pressure) all needed to be adjustable to trigger the correct outputs. The outputs included the fuel injectors, two fuel pumps, Lambda heaters and cooling fans. The ECU needs to have the correct alarm pulse train coming in at exactly the right time to even start the cranking process. The test box has an internal alarm simulator and I designed an external simulator module to be used when the ECU is fitted in the car to ensure that all is working perfectly. I chose a Raspberry PI Pico micro processor to be the heart of the system and wrote the code in MicroPython. I needed all the pulse trains to be synchronised as they would be in a real car and this took some clever programming to generate the five concurrent pulse trains while keeping microsecond accuracy. As this is a one off at this point in time so I didn't design a PCB keeping it on prototype board until Andrew needs another one! Wiring up the 104 pin connector was a nightmare in such a confined space, but in the end we got there. To make the final product look the part I laser etched a white on black facia with Wren's logo and all inputs and outputs identified. All in all a very interesting project that has already proved it's usefulness to Wren and customers around the world. |
"Minion"
A single 0.5 W spherical led
Good ergonomics
|
"Dougal"
Multiple design iterations
0.85mm slot and 1.4mm wire holes.
|
Two 0.5 W Spherical leds
Room on the back for a name or a logo
|
Beautiful and functional product prototypes aren’t just for big-budget design firms anymore.
3D printing technology puts great-looking prototypes and quick and cheap design iterations, within reach for practically any design sector. Be that a hobbyist or a professional. If you’re an inventor, or just have a great idea for a product, your prototypes should look and feel like the final product. This little torch was formed from an idea for a small bedside table light for those nocturnal wanderings.... The concept was that it should be as small as possible built around a PP3 battery, but not so small as to be unergonomic for those who may have difficulty gripping or operating a tiny switch. It needed to stand on its own, could lie down if required and without rolling off the table. I came up with a couple of designs - The white one we nicknamed "Minion" and the other "Dougal"(for those that remember the magic roundabout..) It sort of looked like him when lying down. While the CAD model looked great and was really close it wasn't until the first print came off and it could be handled that some minor issues came to light. Those were corrected until it "felt right" I was also able to try several different front pieces, slide switches and colours. Printing new pieces was less than 30 mins and soon the final two designs came to life. Dimensions can be very tightly controlled. The slot for the slide switch was adjusted by 0.05mm for a perfect sliding fit. The two holes showing are 1.4mm dia and conceal the wires. |
|
|
Complex enclosure and mount for an oil level sensor
This is a Fusion Workshop project that combines a small IOT electronic module with a complex enclose and mount to go inside and on top of a large heating oil tank. The mount design was complicated by the fact that it had to go through the skin of the tank which is curved. There is a "tunnel" piece that goes through the skin at an angle that puts the sensor enclosure parallel to the surface of the oil inside the tank. The top circular part of the enclosure you can see in the picture contains the main control PCB and the radio to connect to a base station. The design allows the top section of the enclosure to be removed leaving the sensor module inside the tank. There a large 33mm thread created in the angled tunnel piece to attach the main enclosure. In the main enclosure there is a completely printed battery holder and retaining clips for the PCB. There would have been no way to create such a complex structure without 3D printing. Certainly there would be no "off the shelf" solution. |