Seijuro

What does MAF Scaling and MAF compensation really do? and MAF size for that matter...

I say this because I thought MAF Scaling was used to define the air volume through the MAF at a given Hz. I also thought compensation was then minor refinements to the Scaling.

Well a little while ago I moved the MAF Scaling tables and MAF Compensation tables to RAM so that I could make live changes using livemaps. (With much thanks to l2r99gst )

I did this because my idle LTFT was at -12.5 and while idling my STFT would hover around -8. So the ECU is pulling a considerable amount of fuel at idle to keep the mixture stoich.

I've also found that trying to get this right while in closed loop is a nightmare. I've switched to open loop, and I'm adjusting values to get my recorded AFR to match my AFRMAP.

Here are the values I'm starting with. In open loop Idle is at 12.5 AFR, cruise is between 9.8 and 10.3.




Using Evoscan I see that at idle my Hz are between 35 and 40.

So, I modified my MAF Scaling in both the 25 and 50hz fields, from ~230 down to 175.5 which is the lower limit. When I do this, my STFT would go from -8 up to about +12... but only temporarily.

After about 3 minutes of idling, that +12 STFT would trickle back down to -8. When I read from RAM, my settings of 175.5 are still there.

Needless to say this behavior the last thing I expected or wanted to see. A static change without a static result is a real b****.

So far I've found that only changes to the MAF Compensation table actually have lasting effect.

This is what I've got so far, it's still on the rich side. I'm not even messing with scaling any longer because I have no idea what it's doing, it's stock scaling that comes with the ROM. Idle is stoich, cruise is still rich at upper 12's to mid 13's. Better than before anyway.



And then as a related side question what is MAF size doing? Does it serve as a limit, or is it some sort of global modifier much like injector size is?

iTune

iTrader: (3)

Are you still using stock injectors? What intake? You should put a list of your mods in your sig so we can see it.

Ceddy

(MAF_Scaling + MAF_Adder) x MAF_Comp / 128 = Injector_MAF_Comp


I think easiest way is to set MAF_Comp scaling to Percent128, and leave MAF_Scaling as stock values.

Then just edit the MAF_Comp table, viewing it as a percent of fuel added or subtracted.

oldevodude

iTrader: (3)

What about basic injector scaling? If injectors are stock then maybe you would need to do all this but I'm at a loss without knowing all the mods etc.... I have never touched mine and my trims plus/minus 2.5 . this is with 1200cc injectors and injen intake

Fast_Freddie

iTrader: (50)

I would recommend not changing the Maf tables period to dial in injectors... that is what your scaling and latency adjustments are for... Considering you are at 35-40hz at idle you are where you need to be there...

Seijuro

Good point, I forget my vehicle is very different than what most here are accustomed to.

This is a 1991 Eagle Talon. Block is bored 40 over, 8.3:1 CR. Stock Evo3 turbo which is a little bigger than the factory DSM turbo.

Everything else is stock 1991 DSM "1g" stuff. Meaning 450cc injectors, rubber intake on the feed, hot, and cold sides. Small side mount intercooler. Stock 1g MAF.

The 1g DSM MAF flows about 30% less than the 2g DSM, which itself flows considerably less than the Evo 8 MAF. I don't think the characteristics of the 1g MAF are at all like the Evo 8.

Seijuro

Ah I see, so MAF_Comp is a multiplier, that would explain why it has the greater impact that I'm seeing.

I think the problem I had before was due to my scaling already being near the bottom of the spectrum. I was already pulling the numbers down a full 130 points and more to get it in line with some data I have about the 1g DSM MAF. In some places, that only had me 50 more points away from the bottom. Of course, each point you go down doesn't just reduce the calculated total, it also reduces the final effect of the multiplier.

So in reality, the MAF_Scaling and MAF_Compensation work the opposite of what I've read. Changes to MAF_Compensation produce large results relative to the value in the Scale. Changes to MAF_Scaling will have a smaller effect on the calculated result.

That also corroborates with what I've done now that has been so much more effective. Now I'm running factory Evo 8 values in MAF_scaling which are much higher values than I ran before. Because these numbers are higher, my changes in MAF_Compensation are more exaggerated.

Once I've hit the point where for a given Hz, one point down in MAF_Compensation is too lean, and one point up is too rich, I can use MAF_Scaling to bring it in line.

The only question I have left, is what is MAF Size doing?

Ceddy

MAFSize is used to calculate Load.

Load = AirCnt x MAFSize / 65536
(AirCnt = Number of MAF ticks per Rev)

Seijuro

Oh I see, but I'm not sure what a MAF tick is?

That's interesting, so changing MAFSize would change the rate with which you transition horizontally across load cells.

Ceddy

MAF Ticks is the signal from the MAF.


MAF Hz = MAF Ticks Per Second

AirCnt = MAF Ticks Per Rev


MAF Hz = AirCnt / Time Per Rev
(My math might be off on this one, but you get the idea.)


And a lot of other things as well, very important that MAFSize is equal to the MAF you are using.

Seijuro

Got ya. Thanks!

Is there a place where I can see all the calculations these variables are used in?

Seijuro

Well knowledge really is power. I've got my idle LTFT low to +0.2%, and with about 200 miles of open loop logging today my cruise LTFT mid is within %5. It's close enough now that I can return to closed loop and tweak the map based on STFT feedback.

My logged AFR is also in line with my fuel maps now too so I can finally start the business of optimizing power!

...which I'm sure will conclusively show that I need more injector, and round and round around we go.