I was bored Saturday and in desperation to solve my spread out lean ltft lo and rich ltft mid I started looking at the USDM(513) and JDM(542) MAF scaling and smoothing tables as a possible solution.
For practical purposes I will refer to the USDM tables as 513 (inj scaling) and 542 for the JDM . Both the 513 and 542 have the same scaling but the smoothing is different. If you add the scaling + smoothing you get an interesting result. See graph.




This morning I loaded up the 542 smoothing table to my rom and my cold start problems went completely away. Ltft lo went from +1 to -9.5 (p0172)and mid stayed the same at -12.4. At least I got ltft lo going in the right direction. An interesting thing is that AFR richened across the RPM ,IDCs went all the way up to 90%

Below: 513 vs 542 Maf smoothing latencies and inj scaling the same, car is pig rich.



Below: I then changed scaling from 622 to 650, afr leaned some but ltft mid stayed the same -12.4 and lftf lo changed to -8. (back to latencies again)



In the MAF scaling thread it shows how to scale the maf through the scaling table. The smoothing table gives another adjustment option.
to be continued...

 

on a MAF note,

I recently swapped a Evo 9 MAF onto my 05 evo 8. The car went about .8 leaner overall. My Evo 8 MAF on the Evo 9 caused the 9 to go richer.

Up until then, I thought all the Evo MAF's where the same.

Also, if you compare MAF scaling and Smoothing tables between a Evo 9 and Evo 8 they are different, with the Evo 8 Scaling and Smoothing with higher numbers through the whole tables. MAF size tables are the same.

Just FYI.

 

Be sure that when you are trying to see that the LTFT mid is changing that you are at a steady cruise for a good amount of time, like 10 - 15 minutes or so. I forgot the exact requirements for the LTFT Mid to change, but I know you have to be above a certain RPM (2500 maybe...rings a bell) and hold a steady RPM for a certain amount of time.

An easier way to see what the LTFTs will change to is to simply log the STFT. The STFT will be pushed into the LTFT when the criteria is met for the LTFTs to change, if that makes sense. During a cruise, the ECU will switch what LTFT it is using from LTFT Lo to LTFT mid, so simply by logging the STFT during a steady cruise will show you what your LTFT mid will eventually be.

So, a quick example: if your LTFT mid is -12.4 like you say and you go for a steady cruise and your STFT at that cruise is +10, then your LTFT mid really is -2.4.


Eric


Edit: Also, if your whole goal is to just change your idle LTFT (LTFT Lo) and you don't want to deal with setting up the latencies, etc, then just alter the first value in your MAF scaling table.

I think Bez posted information on what the smoothing was used for, but I never really tested the smoothing settings. I'm interested to see in what else you find.

 

That's an interesting fact.

Wonder why the difference? If they could just change the scaling/smoothing why change the sensor? Seems like a cost-ineffective way to do it. ?

I haven't messed with the airflow tables that much so far. I haven't had to, but, I haven't had time to either.

I'm still learning some of this stuff too.

 

I wonder, with the MAF's being confirmed different now if it is in any way advantageous for and of the two scenarios:

Evo 8 MAF in a 9, or a 9 in an 8.

Can you see any physical differences in the size of the MAF? I would assume no.

 

Interesting results this morning, I lowered latencies 6% and was able to richen idle by adjusting uint8 values of the 25 (115to 123)and 50(123 to 127) cells (airflowinhz) of the Maf smoothing table.

Before I did the change stft were in the +10 after around +/-3.

Edit 1 uint= aprox 2%

 

Interesting. I know that before when I used maf scaling to tweak my afr, i would hit different load scaled dependant off changes. I wonder if the maf smoothing allows to tweak without affecting load under partial and wot.

 

Yep, it does not affect load just AFR. See graphs on post#1

 

So load is dependant on scaled hz and possibly baro too(?)
and afr (ipw) is dependant on AFRMAP, baro, temp, lean spool/2D antilag table (while activated), injector tables and smoothing. Need someone from Mitsu to tell us the formula

 

I am not sure they'll be able to tell it .
It seem there few groups written the code.
Firs the done some basic code. Then next group added some trims (to compensate somthing-let say air temperaure) then next group added some other compensations. Then someone added compensations to compensate compensations...Thats crazy...
Some compensations are time-dependent and not easy to write as exact math formula.
There are about 40 countdown trims used in the proccess.(>300 !!! total )
They use different 'loads' for AFR and ignition tables.

Now Im trying to sort this crap out or at least understand how they process MAS into load and FuelPULSE, find a 'real' accel enrich tables-i need it to finish my total conversion to speed density. They ,for example have tables ,that limits MAS signal momentary increase (they chop it if its raising faster than they think-and these filters are coolant, dependent and depend on how hard you press the pedal)...

 

One group had too much Sake at one point