Target AFR not following fuel map [Ecutek]

[tomm.brz]

you re doing good

start low with timing , fix afr first with a bit of speed density maybe (it helps on full throttle pulls)
it s the cam timing the sound i bet , look at cam tables from OTS stg2 tunes but prerestyling, those that have arorund 37-40 of values both in intake and exhaust at mid rpm high load
Then put down of some degrees and use 35-36 as max values for exhaust and 34-35 for intake

[SumDizzy]

Okay, the pre and post styling cam timings are quite similar, but I see the exhaust cam has way more timing in the 3600-4000 range during high load. It also sheds timing in the 2400-2800 range too. Here's the cam timing I copied off the OFT Stage2+ A00G Rom including the following changes:

• Scaled the load and rpm axis. Looks nicer to me this way + matches with the other axis changes I did from the Ecutek guide; let me know if I should not be doing this.
• Took 4 degrees of timing from 0.9 - 2.25 load across all rpm's.
• In the same section, capped the max timing for exhaust and intake: 35-36 and 34-35 respectively.
• Did not touch the low load section

I'm updating both the high and low det maps. Let me know if you see anything clearly wrong.

[tomm.brz]

that s nice start
for intake make +2 overall so it reaches 35 at 4800rpm then goes down.. blend the end and end it at 10
0.8 engine load column make max 20

exhaust: make +4 untill 2k, then all 35 (can also try 36) until 4800 then fade until 21
Talking about high load

[SumDizzy]

@tomm.brz Can you clarify how I can use a K00G rom? Ecutek says it's not compatible, and only that the 2018 (Q00G) and 2019 (S20G) EUDM roms are supported.

From sleuthing the forums I understand I can modify the calid in 3 places in the rom file to spoof the K00G rom as a different calid, but given that Ecutek saves rom in an encrypted format I would have to use RomRaider instead.

I can modify the calid of an unencrypted rom and then try opening with Ecutek, but this fails because Ecutek maps the definitions incorrectly.

Only other thing I can think of is to find a common ROM that shares both my rom and K00G as compatible roms and use it as a bridge. But it seems odd that Ecutek wouldn't make the whole set compatible.

I ask because I see other weirdness e.g. while engine braking in sixth gear, I don't get a fuel cut sometimes - so instead of afr going extremely lean, it goes very rich instead and will stay like that for as long as I'm engine braking. I'd like to use a known good stable rom instead of a maybe good rom.

[tomm.brz]

You can t do anymore, it is officially blocked by ecutek
you only can if you haven t updated ecutek since May
and NEVER calid hack an ecutek rom it doesn t work it works only for openflash and tactrix

That "bug" i do have it too with K00G ap you would nt fix it
it afflicts every restyling car
it is hard coded by subaru

You can t get fuel cut in sixth gear over 4k rpm morenor less
you can cut it manually by touching throttle very briefly during the engine braking it wil make a little jump and then enter the fuel cut
but still doesn t work at very high vehicle speed and high engine rpm
sometimes also 5th gear is affected the same

[SumDizzy]

Okay, thanks for the clarification. I couldn't find any reference to it on the ft86club or nasioc forums so I assumed it was a very unique bug.

[tomm.brz]

how is your car doing btw?

[SumDizzy]

It's in a good place where I feel comfortable driving it around and doing the odd pull while watching the parameters on my phone. I'm pretty happy with the fuel map - I couldn't find a lot of specific info for this platform, but I got a general sense of what is a safe target for a road tune and how to structure the various zones. I was going to see if I could start a separate discussion about that in the future.


Knock isn't completely resolved and I've restarted the ignition timing map multiple times. As mentioned my initial tune had way too much timing causing the audible knock. As a new base I've used the OTS base/advance map with 4 degrees removed. I've also removed about 1 degree per pound of boost. My second attempt at building an ignition map was quite aggressive in retarding knock. I almost never saw any knock in cruising conditions including 6th gear up a hill, but at WOT I had negative timing and was still seeing the odd knock in mid-high load. That didn't make sense to me and I suspected the timing was too low.


So right now I'm on my third attempt which isn't as retarded and now I see around 5-10 degrees of timing during a WOT pull - that seems to work well but I'm seeing more knock during cruise conditions I'm trying to correct. Current knock correction goals are:

• During cruise: Knock events are okay as long as they don't lower advance multiplier
• During WOT: All knock events are bad

[tomm.brz]

if you want u can send me again the tune i can give you hints

[SumDizzy]

I will share my tune and an accompanying datalog shortly!

[Compelica]

Somewhat related - does LTFT learning stop when a CEL is lit up?

[tomm.brz]

i havent understood it , STFT never stops but LTFT can delay its activation or not even start at all if some conditions aren t met
why?

[callisto]

Well, I dug into that question for a while and it is very twisty and a lot of threshold adjustments. But there's this one block that, if FFF8AAE2 is set to false, ends up seeming to maybe disable LTFT processing. I wish I could be more certain; still, it's extremely suspicious finding a block of RPM/MAF/speed/load logic with thresholds right at the top of several blocks of LTFT math. Someone could log FFF8AEE2 (on K00G) and see if LTFT only changes when it's true, perhaps.

  uVar2 = FFF8CA24_if_zero_else_four();
  fVar3 = QUERY_2dtable(engineRpm,&2dmtable_000bb0f8);
  if (((BYTE_fff8aa97 != 1) || ((1200.0 <= engineRpm && (engineLoad < fVar3)))) ||
     ((((!bVar1 || ((engineRpm < 500.0 || (900.0 <= engineRpm)))) ||
       (((BOOL_fff8aae2 = (uVar2 & 0xff) == 4, !BOOL_fff8aae2 &&
         ((engineSpeed < 0.0 || (BOOL_fff8aae2 = engineRpm < 900.0, !BOOL_fff8aae2)))) &&
        ((0.0 <= engineSpeed || (BOOL_fff8aae2 = engineRpm < 2500.0, !BOOL_fff8aae2)))))) &&
      ((((bVar1 || (engineRpm < 0.0)) || (10000.0 <= engineRpm)) ||
       ((engineMaf < 1.6 || (BOOL_fff8aae2 = engineMaf < 200.0, !BOOL_fff8aae2)))))))) {
    BOOL_fff8aae2 = false;
  }

The above is from K00G; fvar3 is 0.1 at all rpms. I checked a few different calibrations and these exact values and this exact logic seem unchanged from A01G onward.

No luck determining whether CELs disable it or not, but at least there's a possibility of something interesting.

[Compelica]

Originally Posted by tomm.brz

i havent understood it , STFT never stops but LTFT can delay its activation or not even start at all if some conditions aren t met
why?

Someone mentioned in the forums before that LTFT learning would stop upon CEL - sounds strange since that may put the car into possibly lean conditions (and I know certain owners who could care less driving around with P0420 lit up.)