Fuel Pump speed control
Discussion
Max_Torque said:
A basic single bridge PWM controller for a brushed motor (like a conventional fuel pump) can be adapted to read a PWM input easily enough. They usually read in a voltage that is supplied by a potential divider. They have 3 wires normal for the speed signal:
1) 5v
2) signal
3) 0v
So a crude input for such a controller could be your TPS signal. I would probably want full power to the fuel pump at anything over 50% throttle. But at least at idle and around town the pump is running slower.1) 5v
2) signal
3) 0v
eliot said:
So a crude input for such a controller could be your TPS signal. I would probably want full power to the fuel pump at anything over 50% throttle. But at least at idle and around town the pump is running slower.
Exactly one of the options I suggested at the start for a simple controller.The only addition would be a full power delay. Just so that when going up through the gears etc, the pump isnt cycling hi/low every shift. Maybe a delay of just a second or two before it reverts back to a lower setting.
The problem with any Dumb controller is that that fuel flow and pressure are NOT linearly proportional to pump voltage! There is a dead band at the lower and upper end (due to mech friction etc) and the pumps electrical current (which is equivalent to output pressure) depends on the difference between the pumps forward and reverse voltages, which are in turn dependant upon things like speed, temperature, supply voltage etc.
So you could find that sometimes the pump doesn't supply enough fuel when you need it, unless you leave a healthly margin between pump output and engine demand. Added to which the non linearity of most mechanical fuel pressure regulators also will change the actual fuel injector deltaP.
Fuel consumption by an IC engine is proportional to LOAD x Speed, so using TPS in itself would not be enough, unless you can set the %PWM output of your controller to effectively go max at say 25% throttle rather than 100% throttle etc.
The more "complicated" control solution (closed loop control) only takes 1 more sensor (a fuel rail pressure sensor) and completely removes any significant change in fuel injector deltaP (for example, on a turbo engine if you boost pressure control is a bit flaky, during any overboost, fuel injector DeltaP is maintained at your target level automatically.
This is why the OEM's use it, you install it, cal it, then it just works!.
So you could find that sometimes the pump doesn't supply enough fuel when you need it, unless you leave a healthly margin between pump output and engine demand. Added to which the non linearity of most mechanical fuel pressure regulators also will change the actual fuel injector deltaP.
Fuel consumption by an IC engine is proportional to LOAD x Speed, so using TPS in itself would not be enough, unless you can set the %PWM output of your controller to effectively go max at say 25% throttle rather than 100% throttle etc.
The more "complicated" control solution (closed loop control) only takes 1 more sensor (a fuel rail pressure sensor) and completely removes any significant change in fuel injector deltaP (for example, on a turbo engine if you boost pressure control is a bit flaky, during any overboost, fuel injector DeltaP is maintained at your target level automatically.
This is why the OEM's use it, you install it, cal it, then it just works!.
http://fuelab.com/fuel-pressure-regulators/529-ser...
Here it is. Just what we are looking for. Due to be released any day.
Here it is. Just what we are looking for. Due to be released any day.
MarkWebb said:
http://fuelab.com/fuel-pressure-regulators/529-ser...
Here it is. Just what we are looking for. Due to be released any day.
Interesting. Pity no details whatsoever about it ? Clearly it is only 1 part of a system that is required.Here it is. Just what we are looking for. Due to be released any day.
Check US patent 7810470 for all the bloody (did I use that word correctly? I'm a Yank) details on the FueLab regulator. It is intended to be used with the FueLab pump, which incorporates the ECM. But I bet one could do their own ECM to drive other makes of pumps.
I'm working on a control for my Aeromotive pump on a street rod. Now it runs at full speed, drawing 10 amps, and most of the fuel just goes back to the tank. I like to work in analog and am going to use MAP as the primary input, plus maybe TPS if needed.
I'm working on a control for my Aeromotive pump on a street rod. Now it runs at full speed, drawing 10 amps, and most of the fuel just goes back to the tank. I like to work in analog and am going to use MAP as the primary input, plus maybe TPS if needed.
Bit of a topic "re-awaken"!
I've been looking at this sort of thing again, and after checking out the availible aftermarket devices (of all kinds) i'm not sure there are actually any "Good" ones to buy?
Basically, it seems, when it comes to aftermarket vehicle/motorsport electronics, there are two options;
1)Cheap, but crap
or
2) F-ing expensive, but good
Things like CraigDavid 'lecy water pumps fall into the first cateogory and things Pectel / Motec Ecus etc fall into the 2nd. Between those, there seems to be nothing at all!
Looking at the pump controllers, all of the ones i have looked at fall down on the basics (not waterproof, poor & underrated connectors (if any), no EMC/noise control, No loaddump/overvoltage protection etc, No obvious thermal control (potted into cheapo plastic boxes without heatsinks, and almost certainly full of non automotive rated components! if you sit down and put together a Bill Of Materials for a device that actually stands some chance of working well, and continuing to work for more than about 5mins, you pretty much immediately get to the sort of cost as the current asking Price for those devices. Hmmmm. It makes me wonder just how much test, development and work is being put into these existing devices etc??
The "decent" stuff, but the stuff you need to remortgage your house for does look to be reliable and properly spec'd, but the end result of that seems to be price tags starting in the thousands rather than the hundreds......
Just wondering if anyone has "Horror" stories for fitting the cheap stuff, or any "success" stories too for that matter?
I've been looking at this sort of thing again, and after checking out the availible aftermarket devices (of all kinds) i'm not sure there are actually any "Good" ones to buy?
Basically, it seems, when it comes to aftermarket vehicle/motorsport electronics, there are two options;
1)Cheap, but crap
or
2) F-ing expensive, but good
Things like CraigDavid 'lecy water pumps fall into the first cateogory and things Pectel / Motec Ecus etc fall into the 2nd. Between those, there seems to be nothing at all!
Looking at the pump controllers, all of the ones i have looked at fall down on the basics (not waterproof, poor & underrated connectors (if any), no EMC/noise control, No loaddump/overvoltage protection etc, No obvious thermal control (potted into cheapo plastic boxes without heatsinks, and almost certainly full of non automotive rated components! if you sit down and put together a Bill Of Materials for a device that actually stands some chance of working well, and continuing to work for more than about 5mins, you pretty much immediately get to the sort of cost as the current asking Price for those devices. Hmmmm. It makes me wonder just how much test, development and work is being put into these existing devices etc??
The "decent" stuff, but the stuff you need to remortgage your house for does look to be reliable and properly spec'd, but the end result of that seems to be price tags starting in the thousands rather than the hundreds......
Just wondering if anyone has "Horror" stories for fitting the cheap stuff, or any "success" stories too for that matter?
Wow..cant believe that thread was 2 years ago !
Some ecu's are now starting to offer PWM control of the fuel pumps, either open or closed loop. I know Syvecs now offer this on some models.
Aeromotive has a simple generic PWM controller based on a pulsed input, usually rpm. Not overly cheap or sophisticated, but it would do the job, and allows you to do away with relays etc.
The only closed loop, dead end setup Ive seen from the aftermarket, and it's still targeted at a specific vehicle is from Vaporworx
http://vaporworx.com/
In the US another doing something similar I think is Squash, never looked into them much though.
http://www.squashperformance.com/
Some ecu's are now starting to offer PWM control of the fuel pumps, either open or closed loop. I know Syvecs now offer this on some models.
Aeromotive has a simple generic PWM controller based on a pulsed input, usually rpm. Not overly cheap or sophisticated, but it would do the job, and allows you to do away with relays etc.
The only closed loop, dead end setup Ive seen from the aftermarket, and it's still targeted at a specific vehicle is from Vaporworx
http://vaporworx.com/
In the US another doing something similar I think is Squash, never looked into them much though.
http://www.squashperformance.com/
Yeah, i looked at both those systems you mention, neither looks, er, how do i politely put this, very good? (things like flying leads coming out of modules, non waterproof terminals, lack of EMI controls etc)
It seems difficult to find a price for them too, although around $200 looks to be fairly typical, which i guess might cost you the same in £ once landed in this country?
It seems difficult to find a price for them too, although around $200 looks to be fairly typical, which i guess might cost you the same in £ once landed in this country?
Richyvrlimited said:
Max_Torque said:
But what does it drive the pump with? The clever bit, is in the driver, rather than the logic tbh
It'll just be using one of it's generic PWM drivers on the board. Max_Torque said:
Sounds like a "Noise" nightmare to me! (20Amp pwm inside your engine ecu, and having to run long (chances are, your engine ecu is at a different end of the car to the fuel pump!) cables that are switching at maybe 15kHz............
Fantastic thread. I am going to be attempting this. I have a Pectel SQ6 ECU which can PWM control upto 6 seperate channels, upto 10A peak each. The ECU can be directly next to the fuel pump.
My engine is normally aspirated V8, about 400bhp. I have calclated that extremely worst case AFR of 10:1 @ 9000rpm it would potentially need 172 L/per hour of fuel.
I plan to fit a 200 L/hour pump with no return line, and electronically control the pressure in the line to around 3.6bar/50psi.
A walbro 255 L/hour is between 7-8amps peak current at between 12-14V at 3.0 bar.
My questions are related to the strategy for control. Discussed earlier in this thread are two options, one to use a pressure transducer in the fuel rail, post filter, pre-injectors, and use this as a feedback loop to the ECU to control a PWM input table.
The second, to run a more 'dumb' system where above 2000rpm the ECU just uses a PWM table to target a constant current (say 8amps) as it appears that the relationship between current and pressure is fairly linear.
what are your thoughts Max Torque?
cheers,
Patrick
Pat_T said:
Fantastic thread.
I am going to be attempting this. I have a Pectel SQ6 ECU which can PWM control upto 6 seperate channels, upto 10A peak each. The ECU can be directly next to the fuel pump.
My engine is normally aspirated V8, about 400bhp. I have calclated that extremely worst case AFR of 10:1 @ 9000rpm it would potentially need 172 L/per hour of fuel.
I plan to fit a 200 L/hour pump with no return line, and electronically control the pressure in the line to around 3.6bar/50psi.
A walbro 255 L/hour is between 7-8amps peak current at between 12-14V at 3.0 bar.
My questions are related to the strategy for control. Discussed earlier in this thread are two options, one to use a pressure transducer in the fuel rail, post filter, pre-injectors, and use this as a feedback loop to the ECU to control a PWM input table.
The second, to run a more 'dumb' system where above 2000rpm the ECU just uses a PWM table to target a constant current (say 8amps) as it appears that the relationship between current and pressure is fairly linear.
what are your thoughts Max Torque?
cheers,
Patrick
If you do not have the ability to run with proper (ie, high band width) closed loop pressure control, then you will want to retain a basic "overpressure" regulator and local return, probably mounted at the pump end of the car. This is because your open loop settings will not be able to be close enough to actual fuel delivery quantity under all circumstances, and so pressure excurions will be regular and of a large magnitude. In any case, you will want to measure injector delta pressure and apply a fuel mass correction to suit, to maintain accurate fuelling.I am going to be attempting this. I have a Pectel SQ6 ECU which can PWM control upto 6 seperate channels, upto 10A peak each. The ECU can be directly next to the fuel pump.
My engine is normally aspirated V8, about 400bhp. I have calclated that extremely worst case AFR of 10:1 @ 9000rpm it would potentially need 172 L/per hour of fuel.
I plan to fit a 200 L/hour pump with no return line, and electronically control the pressure in the line to around 3.6bar/50psi.
A walbro 255 L/hour is between 7-8amps peak current at between 12-14V at 3.0 bar.
My questions are related to the strategy for control. Discussed earlier in this thread are two options, one to use a pressure transducer in the fuel rail, post filter, pre-injectors, and use this as a feedback loop to the ECU to control a PWM input table.
The second, to run a more 'dumb' system where above 2000rpm the ECU just uses a PWM table to target a constant current (say 8amps) as it appears that the relationship between current and pressure is fairly linear.
what are your thoughts Max Torque?
cheers,
Patrick
Having the engine ecu, generally mounted at the other end of the car from the fuel tank, doing the high current circuit control never works well, unless you can manage to engineer a super low inductance and well shielded cable run to the pump. I would highly recommend placing the PWM driver as close as possible to the pump to minimise the RF loop area etc.
If your ecu can do output current control, then this tallies well with absolute fuel pressure, usually with a slight offset dependent on fuel temperature (fuel kinematic viscosity & pump element clearance effects)
Max_Torque said:
If you do not have the ability to run with proper (ie, high band width) closed loop pressure control, then you will want to retain a basic "overpressure" regulator and local return, probably mounted at the pump end of the car.
Ok so this suggests the closed-loop option with the pressure transducer is preferable, as I don't want a mechanical regulator or return lines if I can help it.Max_Torque said:
Having the engine ecu, generally mounted at the other end of the car from the fuel tank, doing the high current circuit control never works well, unless you can manage to engineer a super low inductance and well shielded cable run to the pump. I would highly recommend placing the PWM driver as close as possible to the pump to minimise the RF loop area etc.
Max_Torque said:
If your ecu can do output current control, then this tallies well with absolute fuel pressure, usually with a slight offset dependent on fuel temperature (fuel kinematic viscosity & pump element clearance effects)
I thought so. I'm assuming the offset is negative with temperature (i.e fuel pressure drops away as temp increases due to viscosity decrease and pump rotor clearances increase)?I need to confirm the ECU can do output curent control. If it can't are there any other options to easily drive the pump at a constant current?
Thanks for your help, wait 'till I start asking you about PWM controlling the Electric waterpump I have stolen from a BMW 335i too...
Understand what Inductance is, and how the "loop area" of the wires to your pump affect it! Basically, you are going to pulse the battery voltage on and off to the pump at high frequency (probably around 10KHz, although you can go a lot slower because you don't care too much about torque ripple in the pump motor, so down to around only 1Khz maybe). The wires that go to the pump become, in effect antennas. It's critical that the pump has a power feed and a ground wire that are as closely coupled as possible, and hopefully your ecu also has a dedicated "ground return" input for the PWM output. This means you can run a twisted pair cable run to the pump, meaning the outgoing and returning wires are extremely well coupled. Also, you will want to shield both those wires with a wire mesh RF shield that is grounded (at one end only) to chassis earth.
if you are luckly, and you can keep the wires shortish, you may get away without any snubbing or common mode filtering on the pump output or the ecu input.
In my experience, you need a control bandwidth of around 500Hz minimum to accurately control fuel rail pressure in a "deadheaded" port injected system. Lower than that, the electrical and mechanical constants of the system start to dominate and make control difficult.
RE: BMW water pump. If you have the "early" all metal body type pump, the Pierburg CWA200, then controlling it's speed is easy, because all it needs is a +12v, >45 to ~100Hz PWM signal (see link) on the speed control pin. The later pumps (with the plastic housings) use CAN to control their speed and so take more effort to use (The later pumps are more powerful, more efficient and so far, more reliable however)
CWA200_datasheet
if you are luckly, and you can keep the wires shortish, you may get away without any snubbing or common mode filtering on the pump output or the ecu input.
In my experience, you need a control bandwidth of around 500Hz minimum to accurately control fuel rail pressure in a "deadheaded" port injected system. Lower than that, the electrical and mechanical constants of the system start to dominate and make control difficult.
RE: BMW water pump. If you have the "early" all metal body type pump, the Pierburg CWA200, then controlling it's speed is easy, because all it needs is a +12v, >45 to ~100Hz PWM signal (see link) on the speed control pin. The later pumps (with the plastic housings) use CAN to control their speed and so take more effort to use (The later pumps are more powerful, more efficient and so far, more reliable however)
CWA200_datasheet
Edited by anonymous-user on Monday 24th November 20:29
Thanks again M_T. I'm learning a lot. I normally design camshafts and cylinder heads, specialise in engine simulation and dabble in ignition mapping and emissions development from time to time. It's fair to say full on electronics is definitely not yet my forté though!
The ECU spec is here http://www.cosworth.com/media/199295/29E-071516.pd...
If you get a chance to have a brief look over, let me know if there is anything that will be a show stopper in terms of capability?
Yes I have the Pierburg CWA200, that's good to know, I had already read it was fairly easy to control. I've already installed it in the car. I may just run it at constant 12v when I first get the engine running and for mapping, then develop more optimum cooling strategies over time. That one is going to be some distance from the ECU however, but I am right in thinking I don't need to pulse the EWP at anything like the frequencies we're talking about for fuel pressure control, so the inductance in those wires shouldn't be an issue?
I'll probably end up having a full motorsport spec wiring loom professionally made to suit my set-up but I want to know exactly what I'll need to ask for. I know I'm not necessarily doing things the simplest way, but hopefully the lightest and close to most efficient way
(aiming for 400bhp/750kg)
http://s1elige.blogspot.co.uk/
The ECU spec is here http://www.cosworth.com/media/199295/29E-071516.pd...
If you get a chance to have a brief look over, let me know if there is anything that will be a show stopper in terms of capability?
Yes I have the Pierburg CWA200, that's good to know, I had already read it was fairly easy to control. I've already installed it in the car. I may just run it at constant 12v when I first get the engine running and for mapping, then develop more optimum cooling strategies over time. That one is going to be some distance from the ECU however, but I am right in thinking I don't need to pulse the EWP at anything like the frequencies we're talking about for fuel pressure control, so the inductance in those wires shouldn't be an issue?
I'll probably end up having a full motorsport spec wiring loom professionally made to suit my set-up but I want to know exactly what I'll need to ask for. I know I'm not necessarily doing things the simplest way, but hopefully the lightest and close to most efficient way
(aiming for 400bhp/750kg)
http://s1elige.blogspot.co.uk/
The Ecu spec for the SQ6 is fairly open ended, and contains significant potential issues.
For example, the PWM outputs are quoted as "10A peak" and no continuous rating is given. However, the power supply pins to the ECU are only rated at a maximum total of 10A (Vbatt) & 20A (gnd) (most ecu actuators, eg injectors, coils etc, are Low Side switched to ground, and do not source current from the ecu (the ecu sinks power to ground to turn them on)
Also, only the "DC motor driver" output has a specific return pin (HB3B) but as you do not ever need to control the pump motor direction, just it's speed (never need to drive pump backwards!) you will be using half the bridge as the normal ground path.
A genuine 400bhp fuel pump is going to be pulling 10A on it's own, and will be taking between 3A and 5A all the time.
Knowing the typical PWM Full and Half bridge drivers used in these devices, i would not want to source more than 5A continuous from the ecu itself.
Driving a fuel pump really requires a dedicated external amplifier, using either a specifically designed driver IC (the Infineon BTN7960 for example) or a descrete mosfet half bridge.
I suspect that if you talk to Pectel they will suggest you use an external driver box for the pump........
The Pierburg CWA200 only uses the PWM signal input to set it's target speed. It has it's own internal power drivers, so the PWM signal pin will only sink a tiny amount of current (<10mA) This means you can drive it with just about any PWM output. The only complication is that the input does NOT accept TTL logic levels, and requires >8Vdc to set a "high" state. This would require a simple voltage level conversion circuit (1 transistor, 2 resistors). However, if you drive it from the "PWM" outputs on the SQ6, these should put out Vbatt as their "High" logic level voltage. Because of the low currents and slow frequency (~100Hz means you can have nice slow rising/falling edges without loosing pwm resolution), you can run this signal line down a tiny "twisted" pair shielded cable and have no issues with noise over long cable runs.
For example, the PWM outputs are quoted as "10A peak" and no continuous rating is given. However, the power supply pins to the ECU are only rated at a maximum total of 10A (Vbatt) & 20A (gnd) (most ecu actuators, eg injectors, coils etc, are Low Side switched to ground, and do not source current from the ecu (the ecu sinks power to ground to turn them on)
Also, only the "DC motor driver" output has a specific return pin (HB3B) but as you do not ever need to control the pump motor direction, just it's speed (never need to drive pump backwards!) you will be using half the bridge as the normal ground path.
A genuine 400bhp fuel pump is going to be pulling 10A on it's own, and will be taking between 3A and 5A all the time.
Knowing the typical PWM Full and Half bridge drivers used in these devices, i would not want to source more than 5A continuous from the ecu itself.
Driving a fuel pump really requires a dedicated external amplifier, using either a specifically designed driver IC (the Infineon BTN7960 for example) or a descrete mosfet half bridge.
I suspect that if you talk to Pectel they will suggest you use an external driver box for the pump........
The Pierburg CWA200 only uses the PWM signal input to set it's target speed. It has it's own internal power drivers, so the PWM signal pin will only sink a tiny amount of current (<10mA) This means you can drive it with just about any PWM output. The only complication is that the input does NOT accept TTL logic levels, and requires >8Vdc to set a "high" state. This would require a simple voltage level conversion circuit (1 transistor, 2 resistors). However, if you drive it from the "PWM" outputs on the SQ6, these should put out Vbatt as their "High" logic level voltage. Because of the low currents and slow frequency (~100Hz means you can have nice slow rising/falling edges without loosing pwm resolution), you can run this signal line down a tiny "twisted" pair shielded cable and have no issues with noise over long cable runs.
Edited by anonymous-user on Tuesday 25th November 13:18
Gassing Station | Engines & Drivetrain | Top of Page | What's New | My Stuff