Peak And Hold Drivers
So, people feel there are issues with running PWM on Low z injectors, and i am aware of this. From what I have researched is that the OEM use Peak and Hold, so i thought, how can i do this? Low and behold, i found something on Megsquirt forum by another Mini enthusiast, infact its the same chap i mentioned in one of my other topics on dual table firmware and siamese port injection, Jean Belanger. He has designed and built an add on board for the Megasquirt to convert its lowley two injector outputs to P&H.
http://www.jbperf.com/p&h_board/index.html
And before anyone says just use a resistor in series with the injector, I am, I want to bin it, as I know for a fact that the injector characteristics have drasticaly changed!! so much so that the injector thats installed was the one when used on the OEM ECu, flat lined and went lean at full throttle. Ramp up times are huge with maybe slower ramp closed, even with transient diode, dont forget High z ingectors resistance is the injector coil where as Low Z and resistor, more than half the resistance is the resistor, it must play some part in the equasion of things, and i think it is this thats 'increasing' the physical flow rate of the injector. I recon i could get away with using the original injector that was fitted to the car which would originaly max out at around 80bhp. Dont get me wrong, the resistor works a treat, its just not what i want. I now have the engine running well enough to start experimenting, i just dont want to smoke the ECU.
This could be an interesting project
sprox
Please publish table (and config) dumps before/after the PH driver install. Most interesting values to watch:
- VE at low load loadsites
- injector opening parameters (if you care to tune them: not absolutely necessary as they can be compensated with VE table)
I need to tune a few things, then I will post the config and tables, I just got it running, but I now have an issue with a wheel bearing so the car is off the road again,lol.
To the practicality of it. Would simply removing the driver chip, soldering a wire onto the appropriate pads of Pre_Inj for that chip, taking these wires to the 'inputs' on the P&H board, maybe then returning another wire from the P&H driver circuit output to the INJG pads of the removed chip, and solder the P&H FETs inplace of the VEMS Injector FETs. This just leaves the reistors R18, R21,R39 and so on, what to do with these? this would keep the FET heat sinks all in the same place along with the original flyback circuit, with only the add on board needing mounting. Seems daft taking the switched current off the board and then bringing it back again, when you could achieve the same on the 5v signal side of things.
One thing I am concerned with is there is no protection on the processor ? Boom and bust
What about increasing resistor value of R18, R21 and so on, to pull the voltage down, and leave the driver chip well alone, but giving the processor some protection, the driver chip being a rather expensive fuse??
Sprox
You can't use the FETs on the VEMS board because the current needs to go through the sense resistor on the p&h board. If you wanted to do that anyways, you'd have to disconnect the FET's pin that goes to ground, connect it to the corresponding pad for the Darlington transistor on the p&h board, and then connect the injector ground on the p&h board to the VEMS ground pad. That could probably be done but that would make for a pretty screwy installation. Another point is that I'm not sure if the VEMS FET would behave correctly when being driven by the LM1949 on the p&h board. FETs like to be either on or off but they'd have to be partially on during the hold phase and they'd probably heat up quite a bit. That may not be a problem but I'd have to check the specs and/or do tests to have an idea.
As for using higher values for R18,..., in itself, that wouldn't be a good solution. What's needed is a voltage divider which can be done by 2 resistors. Fortunately, the p&h board has a pull down resistor (R10-13) which could be used for the voltage divider together with R18,... If you were to use 1K Ohm for R18... and 430Ohm for R10-13 you'd have a signal that would be within the spec of the LM1949 with a battery voltage from less than 8V to more than 16.5V which should cover pretty much all cases from cold start with weak battery to a poorly functioning voltage regulator.
So the ideal setup would be to connect the p&h board in place of the FETs while replacing R18,R21,..., with 1kOhm resistors and replacing R10-13 on the p&h board with 430Ohm resistors (Digikey part# 430EBK-ND) and using the VEMS flyback with the p&h board Darlington transistors. With that, you'd need to find a way to heatsink the Darlingtons. If you don't mind drilling holes in your case, using the case as heatsink is a good solution if the case is thick aluminium. You'll need to use mica insulators to keep the transistors from grounding to the case.
Jean
Jean thanks for that, i see a solution aproaching.
I had intended to mount the P&H board inside the Alubos case above the Genboard and sinking the Darlington transistors ( I had always considered using these for the reason you lay out) in a similar manner to how the Genboard FETs are sinked, I have plenty of insulator sheet, this would not have been left out.
Ok then, this sounds like a plan.
I will try this with the altered onboard resistors, fairly easy to do, hard wire from the FET pads to the P&H, swap the resistors on the P&H board and give it a go. That way i still have my expensive fuse Grin protecting the processor some what.
Need to see if I can get these parts from Maplins or RS components before I go away next week, i want to do this mod, when im in the hotel, it'll give me something to do.
This the pinacle of the R&D
Boom and bust, maybe with a bit of smoke
Sprox
We have progress
I have been busy swapping the Genboard into the intended Alubus 'Vems' case. While at it I soldered new EGT2, VR2 and P259 chip on board, so far everything checks out, although i have no way of checking the VR chip.
Anyway. I source some 0.6w Metal film resistors and soldered everything onto the P&H board excluding the flyback diodes and transister as discussed with Jean. One thing i did do was solder on the resistor and zener diode which is also part of this circuit, This wont be a problem will it Jean?
I also changed the resistors on the Genboard which is after the FET driver chip and before the FETs, changed them to 102s (1000 ohm). see picture, resistors circled in yellow
I removed the four FETs from the middle row and moved them to the ignition chanels to keep the output capebility. At the same time I installed two new IGTBs for possible future COP/ sequential ignition.
After moving the above FETs, i soldered on wires from the P&H board to the FET pads, checked everything was good with no shorts. The P&H board is mounted, inverted, on the top part of the Alubos case, with one end of the board in the second pcb chanel, with the other end fitted with two stand offs. The FETs are clamped in a similar fashion to the Genboard with insulator sheet between the case and the FETs, FETs checked for any shorts to the case or each other. All good. I think its quite neatly packaged to be honest. Maybe worth considering a longer board to use the pcb slots and just clamp the FETs, obviously VEMS specific then, but how many people would use these to justify the need?
It may be possible to get two of these boards in the case back to back, as Jead had designed for Megasquirt, but it will be damd tight with the wiring, but not impossible.
I now need to rewire the Econoseal connectors and reconfigure the P259 outputs back to my original configuration, rewire the injector to the correct chanel, remove the resistor, and start again with the calibration.
Oh one other thing, I removed the Transient surpression diode from the flyback wire inside the case so I can install it in the loom instead, that way I can keep an eye on its temperature and remove it if required.
Just a note on the config of the injector, currently the single injector is configured in the H[0] tabele as 0-1 with the divider as 1. The question i am asking, is if I set the divider to two, will this reduce the injector firing by half? IE instead of firing every 360 degrees per cycle, it fires once every 720 degrees per cycle??
Reason I ask this is, as i have mentioned previously, the injector Im using is providing far more fuel that it did on the OEM system and wodering if the divider is a factor in this, not knowing how many times the OEM system fired the injector per cycle.
Onwards and upwards, I aint smoked it yet
sprox
The resistor and zener diode won't be a problem because without the flyback diodes and TIP42, they're not connected to anything but ground. As for the transient suppression diode, you need it if you're not using the p&h board's flyback. I think that Rob was saying to remove it and use the p&h board's flyback components which is another way to do it.
As for board size, if I get enough demand for it I could do it. Do you have exact dimensions for it or a link to the case dimensions? I could also adapt the board for a similar installation by removing the flyback components and adding the resistor for the voltage divider.
I look forward to hear how things go from here.
Jean
Ecu is back in the car and running. Before I swap over to the peak and hold drivers, i have changed the config of the injector. The required fuel value has been doubled and the 'divider' in injector settings has been set to 2. At idle i havent noteced any major difference, so its a matter of testing that and letting the VE learn settle the VE map acordingly. I'll have a play with the acceleration enrichments to see if this change improves that situation, which is the main reason for the change really. I'll do that tomorrow evening.
Once thats done, i will rewire the injector and set up for the peak and hold driver. Hopefully the VE map will be close enough allowing EGO correction to do its thing again using VE learn to settle the VE map. Hopefully have this done by the end of the week
Sprox
Been out with the new injector config as above, and everything is good, lol it was fun to get out in the Mini and remember how well it pulls, lol
injector PW
- at idle is now about 2.4ms : not very low at all, should not be very hard to get good quality idle
- and max 11.5ms at 52.5% duty
- The injector's potential is now starting to be used
- RPM=5478, but appr. what MAP, what power?..... 98 to 100kpa and 100bhp MAX power at last dyno print out
Its amasing how many ways there are to achieve exactly the same fueling
I have also tuned the idle Lambda to 1.04 and fast idle to 1.02, you know why Wink The idle seems much smoother now, though its difficult to tell as i've not had the engine running for a while.
Never got round to looking at the acceleration enrichments, but its about the same as before, still some work required there.
Coolant temps have settled out and no longer fluctuate stupid amounts, so i think thats fixed to a degree, still need to get the INC files for these sensors compiled into HEX format.
Next stage now is the swap over to the peak and hold driver. I'll rewire the injector tomorrow evening, its only a matter of running a wire from the Injector output to the injector, I'll leave the other wire with the series resistor where it is and just disconnect it at the injector for now. Testing will start on Wednesday.
Looking at the injector ramp up times, i was thinking of starting with 800 and starting from there ?? remembering this is a low ohm injector with the peak and hold driver, the injector will open fast. Im currently using a ramp up time of 1296
Sprox
It works!
Injector ramp up is set at 552 at the moment and idle is good. Warm up enrichments and idle air need tuning fully again. VE map isnt far off, and the agressive VE learn i have set, is pulling it back fairly quickly. Its just down to tuning now.
Id like to thank those for their help, especialy Jean for taking time out to get involved And also Rob of http://vems.co.uk
This table is the same table found here GenBoard/Manual/DigitalOut/Table but it has been altered to represent the new Peak and Hold Drivers, its more for my reference as the injector configurations use so many different numbers, its hard to remember, lol.
| Family | Number | h[0] mask | Pin | Name | Max Load | Comments |
| INJFET | 0 | 01/1 | EC36 pin 7 | Inj 1 | 10 A | Injector driver |
| INJFET | 1 | 02/2 | EC36 pin 19 | Inj 2 | 5 A | Peak & Hold driver |
| INJFET | 2 | 04/4 | EC36 pin 8 | Inj 3 | 5 A | Peak & Hold driver |
| INJFET | 3 | 08/8 | EC36 pin 20 | Inj 4 | 5 A | Peak & Hold driver |
| INJFET | 4 | 10/16 | EC36 pin 9 | Inj 5 | 5 A | Peak & Hold driver |
| INJFET | 5 | 20/32 | EC36 pin 18 | Inj 6 | 10 A | Injector driver |
| INJFET | 6 | 40/64 | EC36 pin 6 | Inj 7 | 10 A | Injector driver |
| INJFET | 7 | 80/128 | EC36 pin 17 | Inj 8 | 10 A | Injector driver |
Back to MembersPage/Sprocket