Fuel Pump speed control

[anonymous-user]

andygtt said:

Max_Torque said:

Yup, correct

fantastic, how much lol

With all things, how much is closely coupled to the other big question "how many".......

Obviously i am not expecting to make any money personally on it, even including 10% of the time i will have to spend on just laying out the pcb would make it unfeisible!

I am in the process of pulling some costs together, big cost is the first pcb manufacture run, which pretty much costs the same for tooling independant upon if you order 10 parts or 10,000 parts ;-( and the enclosures and connectors are V expensive when ordered in low numbers ( approx £20 per enclosure, which if you ordered 1000 would be something like £1..........)

[stevieturbo]

Although, you never know it could be a good venture.

The yanks love their boost-a-pump devices, and they also fit big pumps quite a lot and there really arent any decent controllers on the market.

So there could be potential there. As long as someone else doesnt copy the control strategies etc.

[Steve_D]

I started this so it goes without saying I'm keen to know what the cost would be.

I would think Superseal connectors would do the job. Expensive but not as expensive as military spec.
Ignore the price in that link as they can be bought for £1.42 or less with bulk (PoleVolt Ltd.).

Thank you for your efforts so far.

Steve

[stevieturbo]

Steve_D said:

I started this so it goes without saying I'm keen to know what the cost would be.

I would think Superseal connectors would do the job. Expensive but not as expensive as military spec.
Ignore the price in that link as they can be bought for £1.42 or less with bulk (PoleVolt Ltd.).

Thank you for your efforts so far.

Steve

M-Cal sell the same connectors for around £2 for single units.

But that specific style is intended for use on wires. Not sure what sort of cost a PCM mount might be.

Then there is the current considerations, at least on the main pump outputs. Despite that link claiming 14A rating....I'd be dubious about running anything at that current for any duration with those connectors.

Although for wiring looms in general, yes they are great ! Much neater and more secure than crappy old spades, and from the right sources very cheap.

Edited by stevieturbo on Friday 14th January 23:46

[Steve_D]

stevieturbo said:

Steve_D said:

I started this so it goes without saying I'm keen to know what the cost would be.

I would think Superseal connectors would do the job. Expensive but not as expensive as military spec.
Ignore the price in that link as they can be bought for £1.42 or less with bulk (PoleVolt Ltd.).

Thank you for your efforts so far.

Steve

M-Cal sell the same connectors for around £2 for single units.

But that specific style is intended for use on wires. Not sure what sort of cost a PCM mount might be.

Then there is the current considerations, at least on the main pump outputs. Despite that link claiming 14A rating....I'd be dubious about running anything at that current for any duration with those connectors.

Although for wiring looms in general, yes they are great ! Much neater and more secure than crappy old spades, and from the right sources very cheap.

Edited by stevieturbo on Friday 14th January 23:46

The PoleVolt catalogue has those listed as 8W but the same series includes a 20W version.

Had not considered PCB mount as there would then be issues sealing the box. I had thought that MTs reason for considering mil spec was to use bulkhead fittings to seal the box. I expect most installations will be in a reasonably clean/dry position so a 200mm flying cable exiting through a slot in the box joint line would suffice when suitably sealed with mastic.

Steve

[anonymous-user]

MIL spec gives you an IP67 rating (at least) so obviously makes the ecu more robust to moisture etc. They also have a much much higher current density and contact reliability (typically using gold plated contacts) and positive engagement (very vibration proof)

I think for the pump controller however, for most customers, using >£100 in just connectors would be a bit much!

The best option is i think some Tyco PE series connectors for the high current connections, using 2 pins per signal (2x 9A and improved reilability to vibration etc) and a better quality sealed connector for the low current signals (possibly JAE 12w ra connector)


http://uk.rs-online.com/web/search/searchBrowseAct...

http://uk.rs-online.com/web/search/searchBrowseAct...


With a "conformal coated pcb" for moisture resistance, and a bit of silicone sealant around the high current connector, that should give a sensible level of environmental protection once assembled (obviously for internal cabin mounting not engine bay or underfloor!) (think splash proof rather than submersible lol)

even then, by the time you have bought the matching wire end connectors to plug into the pcb connectors, you have still spent over £15 just on connectors


My current BOM, at cost, looks roughly like this:

Connector set (complete inc wire ends(note1) £15.56
Enclosure (without drilled end caps for connector mounting £18.86
proto PCB (assuming run of 10 inital batch) £23.80

electronic parts £62.84, inc:
3x 20A pwm output (synchronously rectified for max efficiency)(fwd/rev Vdrop at 20A =<0.2V)
full automotive spec "loaddump" compliance (note2) (6 to 40V input range for Vbatt)
best guess EMC compatibility (prevention/shielding)
single channel high speed (500kbs) CAN coms
3 channel current sensing for diagnostic / fault detection
ECU internal temperature sensing
Supply voltage (Vbatt) sensing and KL15 ignition line "wakeup" compatability
USB data link to PC for programming & loging
status leds on unit for rapid fault finding, + remote status lamp(10W max) driver



Thats obviously without any costs to design, code, test, or assemble the unit!



note1: Some products i could mention, either just sell the device without the matching wire end connectors (and hence look £10 or so cheaper on the ecu price but you still have to buy some overall), or even worse have short flying leads comming out of a grommet on the ecu. This is terrible for reliability, as not only do you have to solder flexible wires to a pcb (which then fatigue very quickly with any vibration) but the customer has to join to those wires, and this is usually the worst connection possible! (making proper, reliable wire to wire connections is not that easy or cheap, Skotchlocks (urgh!) or crappy soldering is usually the downfall of many home built cars!!)

If you fuel pump stops, your car stops! although with 2 pumps, i could write a "Limp home" strategy that just continues on one pump and illuminates the fault lamp to let the driver know to not use full power etc!)

note2: design does NOT include total reverse polarity protection currently (due to the high cost of RBP for 60A current), however, the electronic control parts ARE protected) I am currently investigating a suitable fuseable link that can be pcb mounted that should blow before the Mosfets....)

Edited by anonymous-user on Saturday 15th January 14:57

[anonymous-user]

I just had a thought, i wonder if the optimum control parameter (for non-pressure control version) is just to log fuel injector duty? Simply connecting to one of the fuel injector drivers, would allow the processor to simply generate a "injector on to off" ratio. Fuel flow is proportional to injector pulse width and engine rpm, which is exactly what injector duty is??

This would allow a really easily calibratable 2d table of injector duty vs pump duty ??


(and result in no requirment to input engine rpm and load (MAP etc), just one wire to an injector "switched" side ???)

[Steve_D]

Max_Torque said:

I just had a thought, i wonder if the optimum control parameter (for non-pressure control version) is just to log fuel injector duty? Simply connecting to one of the fuel injector drivers, would allow the processor to simply generate a "injector on to off" ratio. Fuel flow is proportional to injector pulse width and engine rpm, which is exactly what injector duty is??

This would allow a really easily calibratable 2d table of injector duty vs pump duty ??


(and result in no requirment to input engine rpm and load (MAP etc), just one wire to an injector "switched" side ???)

Sounds good to me.
Would it need a table or would it be a linear tracking of the duty cycle plus a margin for good measure?

Steve

[Mr2Mike]

Max_Torque said:

I just had a thought, i wonder if the optimum control parameter (for non-pressure control version) is just to log fuel injector duty?

You'd need to be carefull to build in enough extra margin under steady state conditions - injector duty cycle could change a lot faster than a fuel pump could speed up and down.

[anonymous-user]

I think a calibrateable table would be a good idea, gives you options then (i.e. you could fly a bit closer to the wind at low loads to reduce return fuel mass, and leave a bigger margin at higher loads etc)


I think i'm gonna need to build a fuel system rig to do some proper testing on! (got loads of spare pumps, rails, injectors etc to test, but it would be easier to build a nice rig layout rather than the usual bench and jamjar approach ;-)

[stevieturbo]

I think using TPS is probably the easist option.

Low pedal, low pump output. WOT, pump flat out.

That way the controller would be simple, and the pump or pumps would have plenty of time to respond, as few engines instantly make a load of power the instant you nail the throttle.

Obviously having a delay built in to keep the pumps flat out for a few seconds after WOT too....for those on/off throttle moments.

[tinker-27]

Hi guys I am building a fuel system with a areomotive a1800 pump , I've not powered it up yet but I don't want to run it flat out all The time so I would be interested in this idea .I'm running a fpr near the pump and returnless setup , out of interest it is in a rx7 which runs it pump at 7 volts until boost is present then has full 12v , it uses a resistor in the engine bay with a bypass relay .

[anonymous-user]

Mr2Mike said:

You'd need to be carefull to build in enough extra margin under steady state conditions - injector duty cycle could change a lot faster than a fuel pump could speed up and down.

I suspect the elasticity of the fuel system would actually help to damp out a lot of the fluctations, if not, a simple pressure reservoir could be used to help.

(i use a 100cc reservoir on my car with it's pressure control system to help limit the slope of the system volume vs system pressure characteristic).

on a fast stamp, the FPR will also shut rapidly (as MAP quickly rises to atmospeheric) and could actually result in a brief over pressure at the rail as the dynamic pressure stalled.

If i have the time i'll knock up a quick Matlab sim with some sensible numbers and see what happens!

[andygtt]

Whilst working off of fuel pulses will be fine, Im still keen to work off of a full table in my ecu which will take into account a number of other things... i dont want the pump slowing down just because injector pulses reduce when I change gear at 8000rpm.

Although saying that I currently have no idea what the pulses mapping is going to look like through the range or rpm, boost and revs!

[anonymous-user]

andygtt said:

Whilst working off of fuel pulses will be fine, Im still keen to work off of a full table in my ecu which will take into account a number of other things... i dont want the pump slowing down just because injector pulses reduce when I change gear at 8000rpm.

Although saying that I currently have no idea what the pulses mapping is going to look like through the range or rpm, boost and revs!

That's not an issue, i'll just put calibratable "ramp rates" in for positive and negative pwm changes, so you can say change pwm in the positive direction (more pump speed = more fuel) as quick as you like, but on a tip out during gearshift say, have a really slow negative pwm ramp, that will effectively hold the pump speed high for some time (i would hazard a guess at say 800 to 1000ms would cover most gearshifts)

The advantage to using injector pulses as an input is that you get both injector duty (and hence system flow rate requirement) and if necessary engine rpm from the same signal (just count the number of say negative edges over a time period)


I've just started to knock together the pcb design, and although i haven't fully done the thermal loading check, it looks like i can get everything onto a 76 mm x 98mm pcb, which is reaosnably small.

The only thing i can't decide is whether to have the microcontroller on a seperate "plug in" board that piggy backs the main "power switching" board. That way, if i want to upgrade / downgrade the processor or coms etc, i can do that without redesigning a whole new pcb for the device. (it does add some complications (and a minor cost) in the interboard connector and retention etc.

[anonymous-user]

1st cut at a layout / sizing:









I need to do a proper component placement and review track widths and thermal loading, but it looks ok as a first stab. Thats with the micro on the same board as the power switchers, smoothing capacitors and filter inductors.

[Steve_D]

I don't believe size would be an issue so don't cramp your style or make life difficult for a few extra inches around the waist.

Steve

[stevieturbo]

andygtt said:

Whilst working off of fuel pulses will be fine, Im still keen to work off of a full table in my ecu which will take into account a number of other things... i dont want the pump slowing down just because injector pulses reduce when I change gear at 8000rpm.

Although saying that I currently have no idea what the pulses mapping is going to look like through the range or rpm, boost and revs!

Thats why using TPS alone with a closed throttle "hold" of a few seconds would make it incredibly simple.

[anonymous-user]

Ok, slowly pulling together a workable spec:


3 of 20Amp high side pwm outputs with closed loop current control:
These 3 outputs source current to 3 pumps. They are rated at 20Amp each continous, but looking at the thermal loading i suspect that in fact they will end with a little bit of derating if you somehow manage to use all 3 at 20A continously (can't imagine anything other than an F1 car will spend more than 60% of a lap at full throttle AND peak power rpm!!) They will derate automatically with temperature. The closed loop current control means they are protected against short circuit or pump failures etc. The Mosfets driving these actually have a peak rating of 60Amps, so are safe aginst any unexpected in-rush currents (although with the current control this should never occur)
Expected to be 1x lift pump (probably <10A) and 2 x HP pumps. Currently the unit is NOT internally fused, so i would expect a 60A inline fuse or circuit breaker to be used in the Batt+ supply cable


2 of 2Amp low side pwm outputs:
These outputs sink current, and are protected against flyback voltage so can drive solenoids / relays etc. Used to drive a fuel pressure drop valve if running returnless, and/or a external warning lamp or relay to alert driver to any system fault (or could cut the ignition if you want to be really safe!) (internally fused 3A 12v output availible to drive whatever you control with these LSD's)

2 of 12v nominal (40V overvolt protected) frequency digital inputs:
For monitoring any digital signal, like injector duty, or coil pack trigger (depending on operating mode)

3 of analogue 0-5v inputs:
able to input analogue signals for conversion to control parameters, such as TPS, MAP or Fuel Pressure (input impedance (estimate >100Kohm) and over volt protection TBC)

2 of digital inputs (12v nominal):
Used for KL15 ignition "wake up" and seperate "enable" input (immo, trigger from engine ecu, or crash sensor)

1 of 0-5V analogue output:
8bit digital to analogue convertor output (10 & 16 bit as options) to generate a 0 to 5v signal output (min load resistance is 2Kohm). Used to output an "injector differential pressure" signal to engine control ecu if required (could also be used as a fuel flow signal etc)

High speed (500kbs) CAN bus
Automotive std high speed CAN bus - Optional, would require more "advanced" settings depending on what other system you wanted to interface with etc

USB interface
Std USB connection to PC using TypeA to MiniB usb cable - allows system calibration and data logging via PC





That little lot should allow people to customise to just about any control strategy they fancy. (i may make the s/w open source if i'm feeling particularly generous ;-)

Edited by anonymous-user on Monday 24th January 19:41

[stevieturbo]

Even I am starting to get tempted lol