Rogue_Ant

No problem

And just to make sure no one else has a similar issue, I've updated the package. So should be no more issues.


-Rogue

Rogue_Ant

Ok, attached is an Excel worksheet that allows you to play with the three AFM Transfer Function Tables. This worksheet will output a graph of the transfer function, and will hopefully allow people to get a better understanding of the 944/951 workings by manipulating the tables.

Transfer Function Worksheet

Furthermore, those with aftermarket MAFs, can try to get the transfer function to match their aftermarket unit (and see why the 944/951 stock code is limiting).


-Rogue

V2Rocket

i wonder if i could mess with the timing values to advance it alot, run it on 100 octane and see what it can put down. or would that be too dangerous? lol

Rogue_Ant

Haha - it is your engine Spencer. Use it however you want - but I recommend moderation...

phil823

knock.....Knock......KNOCK........BOOM!

just wait for rouges N/a Knock Sensor and mess with the timing and try to maximize the output. or higher a deff midget with a stethoscope to sit on the engine while you tune..... =0.........

(hopefully that doesnt affend anyone....)

anyways the real good of all these DIY tuning it as mentioned on the 951 post is port/polish,big throttle body, and Forced induction on N/A, stock....i really dont think that they can do much more with whats given....mabey im wrong......

itll only eat so much fuel, suck so much air, and combust so much.

thats where more air, denser air, cooler air, more fuel (to an extent), more displacement,quicker and stronger valve opening and closeing....

the tuning is almost endless with DIY tuning for custom power per addon

barks944

It is quite a coarse method, are you struggling to match the curve of the MAF sensor? I thought the DME would have just used a mapping to relate the voltage of the sensor to the air flow.

Rogue_Ant

This is the DME's mapping to relate voltage to airflow...
This worksheet describes visually the problems with the internal transfer function calculation, and why the actual code needs to be changed in order to properly match up to a MAF.


-Rogue

barks944

I would call it a function rather than a map. No matter though...

All I will say at this point is it is hard enough to work out how the values in the table relate to the graph, nevermind mapping it to a new MAF sensor!

Looks like it divides the curve into 8 segments which allows it to multuply by more or less to give the different gradients.

barks944

Seems like theres a limit on how steep the curve can be? You end up with big steps in between the 8 segments if the gradient is too high.

Rogue_Ant

Indeed, it is a transfer function

Correct, 8 segments. Each is defined by vertically by T1 & T2. T2 being the coarse adjustment, and T1 being the fine. T3 defines the actual curve, which gets applied the same to each of the 8 segments.

Rogue_Ant

Exactly the issue. When trying to match an aftermarket MAF, no matter how you play with the stock tables, you will have to make compromises, else get large steps in between segments.

barks944

Could we say plot 8 points on the curve and then interpolate between the points?

Rogue_Ant

That will require actual code changes. And that is what I've been working on, but with a lot more then 8 points

barks944

Good man, how much spare mem is on the chip? Its only got 4k right?

What software are you using to get the assembly from the binary? Had a quick go the other day but failed .

Rogue_Ant

Yes, 4K on the early, 8K for the late. Though really much of that is taken up by the necessary (and unnecessary) routines. I've made room to play with by deleting redundant maps, so not a big deal.
I used IDA Pro, mostly, to get the assembly file.