PWMing

[Sprocket]

After having good results with running a low ohm injector with the inline resistor, i am now looking at Pulse Width Modulation.

One reason behind this is that the VE table is very flat and the values are low, infact i have changed out the injector for a smaller one where it was too small on the OEM  ECU, not making that much difference, i doubled the VE entries and halved the Required fuel value to gain some resolution. Ramp up time has had to be increased to gain idle stability, obviously the injector now reacting more like a high ohm injector. The flow rate of the injector now unknown as the characteristics are drasicaly different.

Any way, i am looking at this as more of a development sort of thing in the theory of putting together a plug and play conversion loom for the standard wiring harness, therefore doing away with the resistor for simplicity. Ive had a few people asking me now about VEMS for the Mini and a plug and play convertion loom making it easier for them to get the benifits of VEMS without the complications, easier to fit, more chance of Rob getting some sales .

Anyway with that out the way, i need to find the injector characteristics for PWM. I have been reading up on this for where to start and the best basic (we always like things baisic ) information i could find was in the Megasquirt MegaManual  

I also found this relating to VEMS genboard V3 which could be important http://www.vems.hu/wiki/index.php?page=GenBoard%2FVerThree%2FLowZInjectors more importantly this part -

good to know for v3.0 v3.1 v3.2

    * v3.0 : PWMing kit had to be ordered separately. FETdriver supply must be 12V_F, not 5V. Use D6 for 12V_F to MC34151D again: this is very important: MC34151 does not work from 5V, you'll need to select the 12V_F option on v3.0. Connect 1 and 2 at SJ1. To be sure, measure supply voltage of the FETdriver.
    * v3.1 : everything needed for PWMing was in the rescue1 kit. 2 MC34151D FETdrivers were onboard, 2 in the rescue kit.
    * v3.2 : everything required for PWMing is onboard, assembled in the picknplace run with 4xMC33151D FETdrivers. SIL5 R-network NOT needed, as 8 individual resistors are used instead (picknplaced). Therefore the first 3 requirements in the next section are satisfied, just verify them (shouldn't be any problem). Note that R154 10kOhm (marked 103) was accidentally left out from the picknplace, so some board were shipped without it before 2004.Oct.30. It's hard to notice, because its effect is only for a few millisec after the processor boots (before it switches the OC0 pin to output). Still recommended to solder it (at next maintenance, no need to rush to the car).

PWM-ing requires

    * That FETdrivers are installed (see notes above)
    * The R network (under the onboard MAP sensor) is populated (optionally 4k7..10k resistors between OC0 and all of the used injector control pins)
    * It is recommended to pull the OC0 signal high with a 10k pullup resistor, R154(under the AVR, on the bottom side). This ensures that injectors are turned off when the AVR boots (pins are input).
    * MY_CONF += -D FETDRIVER_INVERTING is defined in my_make (since the MC34151D FETdrivers are inverting)
    * PWM parameters are configured (InitialConfig?) according to the injector. PWM-duty 30% at 14V (60% at 6V) and 1024 usec peak-time can be good starting values for many setups (but ask if unsure)
    * low-voltage FlyBack is used (DR, R=0) because FlyBack will activate each time the injector voltage is switched off during the PWM-ing phase (eg. 100 times for a 6.4ms injector pulsewidth!!!). With the high-voltage flyback this will be much heat in the box, not good.




In the above, the red area is reasonably understood, which says that the area in blue should already be ready for PWMing on V3.2 boards.

THe bit im not sure about is the bit in green, what flyback is installed on the v3.2 board as standard, do i need a additional flyback board, should i be concerned about this when using the transient surpression diode, do i need to do anything with the hardware i have before changing to PWming? Just for clarification, i dont want to smoke the Unit, not so much bothered about the injector, i have a couple of those knocking about. Is this a low voltage fly back??

What is the inverting thing all about?

There are so many bits of information about flyback and difficult to see what is current up to date what to use on what version boards, when to use which one, is not at all clear unless you know off hand which works where.

simple answer would be nice such as "Standard Genboard V3.2 with transient surpression diode and low ohm injectors requires no further modifications to run PWM"

Would the above statement be true



The settings.

In the MegaManual it says to initialy set DC% to 30 for low voltage flyback, VEMS Wiki also says 30% (low voltage flyback on V3.2 genboard?) so i would say that this would be an ideal place to start.

Aagain MegaManual states a time threshold of 1.0ms and VEMS wiki says 1024us so agian a good place to start

Tuning the settings

http://www.megamanual.com/v22manual/mtune.htm#pwm seems to cover it pretty well.




Can some one confirm the Genboard V3.2 hardware compatability for PWMing please and or any likely issues I may encounter

Thanks

Sprox

[rob@vems.co.uk]

PWMing a low-z injector is an absolute pain you get all manner of ill effects (noise, rising ground voltages), please do not do it, its kept their for compatability reasons for the people who've managed to get their injectors to work correctly with it.
The 50W resistor (Resistor packs are common on many OEM installations) and Transient Voltage suppression diode will give better injection characteristics than you will ever hope to get with the PWMing.

If your injection map is flat, thats a good sign that the injector(s) are modelled correctly and that the speed density algorithm is working correctly. (Antonch was impressed with the flatness of his table: http://www.vems.co.uk/forum/index.php?topic=221.msg1943#msg1943  )

The way of getting better resolution is by doing the recommended method is to divide req_fuel by 2, and double your fuel figures, which you have done, but thats not a hard and fast rule, you can adjust the figures as much or as little as works for your setup.
I have found that you cannot trust Lambda on idle if its running erratically as I've seen cases where lambda is spot on, but unburnt fuel was spitting out the exhaust (hydrocarbons were in the thousands).  I found that the injector ramp-up time became hugely significant, and that increasing it 100 at a time identified a range where there was a sweet spot for idle, then returning to the lowest number of the sweet spot and incrementing by 20 you'd suddenly hit upon a value where the idle would become perfect.  Initially for this I was relying on Gary@APT's Sun Exhaust gas analyser, but as your ear tunes in you finish up listening to the engine first and confirming with the analyser.

The flow rate of the injector will still be the same as those figures give the throughput of fuel @4bar at 100duty.

[PeepPaadam]

I've used number of Low-z injectors and even though they often require different parameters (PWM DC % is the most varying one) I've got all of them work very smoothly.  For example 1000cc injectors on 1.8liter car. Using PWM sometimes interferes with EGT analog input resulting it showing higher values than normal but not always.
Hopefully new board design resolves such issues.

[rob@vems.co.uk]

If you are going to try PWM, then you need to remove the transient diode from the flyback and use a good wire for flyback.

[Sprocket]

Thanks

Now that i am aware that there may be issues, i can be on the look out.

Remember that this is a development thing for me more than anything, i like to see what happens and why.

Watch this space