Ive reached teh final stage of my MAF project. The intake ducting is built, the code is pretty much complete, and I just need to finalize how to get the MAF signal in.

Ideally what Id like to do is have a pulse accumulator that I could just sample 80 times a second to determine the flow rate. This will give the best resolution (within .5% of the actual flow). Basically, I want one pulse per cycle and to have the pulse accumulator just keep counting. 1000 Hz = 1000 additional pulses added every second. I could then just subtact the current number of pulses from the previous number of pulses to get the flowrate.

Is there any way to do this with a C3 ecm? Id like to basically have it work like the knock counter, but Im not quite sure how to do it, or if its even possible. Heres the schematic of the VATS/MAF input + knock counter input. I could just plug the signal wire right into the ECM and it would sort of work, but be useless due to crappy flow resolution. I sort of get the feeling that theres no way to only make it just count once for each cycle.

The fallback is a freq->voltage converter, but the resolution isnt nearly as good with only an 8 bit A/D converter. The best way would be pulse accumulation.

 

I think the output may already be a series of short pulses produced by a transister switching to ground but its anyones guess as to teh exact nature of the output. I dont have a scope...

 

I don't think you can do it in that manner. The VATS/MAF counter is setup to report the number of 64KHz tics between input pulses. This is basically the same way the DRP counter operates. May be worth a look at the BUA code ($32) to see how GM handles a frequency MAF.

RBob.

 

~1 gm/sec resolution isn't good enough?.

2917, to an unused TPS/MAP, would seen fine, IMO. The MAP systems only slice things 255 times.

You can design software, that has much greater resolution then an injector can react too.

FWIW, I couldn't find any difference in drivibility from going to a 2 to 3 bar MAP, which reduced the resolution signifigantly. Thou, I have enough table size to get things pretty close, in any event.

IMO, as long as you don't peg the MAF, and have a reasonable DC for the injectors, the rest is just overkill. Engines just aren't as exacty as people think they are, again, IMO. Not to mention that with a HEI the best ignition rate you'll get is 94-97% according to some folks. With CNP, you can ratchet that up anoter couple points, but everyone's still fighting to get the perfect ignition system.

 

So the knock counter is teh only one that can function like that?

 

One thing about the MAP is that the tables are also based on rpm as well, so the tables are sliced and diced by both the MAP and rpm. With the MAF, im only going to have airflow to base the fueling off of. If thats off, the error can carry over in magnitudes to the fueling due to the fact that the sensor is non-linear. Right now, the smallest division I have for airflow is 1/16 of a gram. The factory thought it was necessary to have the division be 1/256th of a gram, with scalars to squeeze the max resolution out of the tables. With only one gram as the min resolution, Im afraid that the engine would have a rollercoaster idle and buck with the TCC locked up.

Another reason I wanted this to work is that Id need less external hardware, if any at all. Maybe just a few passive components in line with the signal wire rather then the whole converter bit, or a pricey MAF translator.

 

I looked at the VATS input. Looks pretty useless...

On teh face of it, when the input is triggered it looks like basically a snapshot is taken of a free running counter. I dont really know how to make use of something like that. Right now the pulses from the MAF are so short they dont even trigger it with the RC filter in the way.

 

Yeah, look at a carburetor. Sure, it's resolution is high to an almost infintecimal degree, but, people "slap one on" all the time, and can live happily with it. And (with efi) during normal closed loop driving, the integrator is hopping around happily without the driver noticing. But, when you come to the limits of the hardware, (idle PWs a little close to the border, for example) LOOK OUT, you'll be in for a tuning headache (malignant cancer) unless you have work arounds.

On a PCB hardware note, you could change the resistors and caps to whatever you like, it's not rocket surgery.

The code can be changed, I assume, and some tricks are out there to keep track of looping counts (assuming the loop doesn't loop too many times between sampling).

 

Ive already done my fair share of butchering to both the ecm and the code so Im no stranger to that.

Back to the counter, what confuses me is that each one seems to operate differently, like teh knock counter accumulates knock, while the DRP gives the reference period, the VATS does that thing above, and the VSS does *** knows what, yet theyre all grouped together on teh same chip on successive pins. Is each really unique and different or are they the same thing with the way the input is set up defining how it behaves? Regardless, I would like to know what they actually do. Im an ME so all those chips are a bunch of little black boxes to me. (This is on a C3 tbi ecm, BTW)

 

Generally what happens, is there is a control register that determines how the pin behaves... but since this is a custom built chip for GM, back in the day when asics were really expensive, I could see them making each pin do something specific.

 

May be able to use the cumulative counter at $BC08. I don't think the pin is tied to the harness connector, but that would only be a wire away. Also, I am not sure of the clock rate for that counter, it may be 16KHz as is the ESC cumulative counter.

I haven't looked at the schematics but I'll bet that pin is at least labeled or shown.

RBob.

 

LOL, ya, that's why I just ecm cases as project boxes. At one time I had a real collection of cases, from failed attempts at simple surgeries. What's easy to one guy can be a real challenge, or impossible for another.