RomRaider

After some additional reading, the reason you may not want to do that may have to do with pulling loads of manifold vacuum on deceleration, which may be able to pull oil through the valve guides, if bad enough.

In the thread I'm referencing, people were messing around with the IACV on cable operated WRXs, forcing it to close completely during decel and wound up having oil seepage through valve guides due to the higher vacuum. Could have just been bad valve guides though I suppose!

Since our throttle plate also doubles as the IACV, I assume forcing it to close completely during deceleration would have similar results.

Finally had a chance to load the ROM up in IDA and take a look. My own search, however, shows only 5 addresses with the right string (i.e. 3F 00 00 00 BF 00 00 00).

Those addresses are:
2EB8C
2F874
30424
30C58
322CC

Of those addresses, I believe the two correct ones are 30424 (AF3 Correction Limits) AND 2F874 (AF3 Learning Limits). This is based on a comparison of this ROM to other ROMs and similarities in nearby data, but I could certainly be wrong.

I have not yet flashed this map for testing because it turns out that my vehicle has an undefined version of the ROM installed (EZ1GB10I), and I am not 100% positive that two are compatible (I have a separate thread here asking that question, i.e. whether or not I can flash EZ1GB10K over EZ1GB10I and, less importantly, what the differences between these ROMs are). I will confirm these table addresses ASAP though.

So, I've managed to COMPLETELY get rid of rev hang on my 2010 Forester 2.5i, although vacuum under decel seems a bit excessive (i.e. peaks at about -27inHg under hard decel, but no puffs of smoke when getting back on the gas, so valve guides seem to be able to handle it!). The revs actually decrease almost too quickly, so I may work on these values further, rather than completely zero'ing them.

The table that had the biggest impact was "idle airflow min target decel adder." This table appears to add airflow, in g/s, to the idle airflow calculation, resulting in higher throttle openings during the transition to idle for a duration that appears to be controlled by "Idle airflow min target decel initial idle activation max mode counter." I believe ramping is controls by the ramping adder increasing/decreasing tables (also in alpha idle control in ECU Flash). The throttle opening table ("Target throttle angle idle airflow target base minimum), which determines the throttle opening angle based on measured airflow + the adder values, also had an impact, but mostly at higher RPMs/Loads (see 100's in attached table).

"Overrun initial injector enrichment (pulsewidth)" also had an effect, but softened the transition to off-throttle a bit, rather than delaying it. I will likely increase this value some from my tuned table, as the transition to off throttle in first is very abrupt as-is.

The below-copied table shows my primary changes, with the top four tables being my tuned tables and the bottom four being the stock tables.


I have also attached my current EZ1GB10K tune. Keep in mind that this is using stock timing for now, but, other than rev hang changes, modifies fueling in CL and OL; disables CL/OL delay; and changes throttle mapping to the EU STi (2010?) throttle mapping, scaled to not exceed the EZ1GB10K max requested torque, since other tables utilize those values (not all defined). I could certainly be forgetting some changes too.

If you do decide to make these changes to your own ROM, please do keep an eye on engine vacuum changes and make sure you aren't pulling oil through your valve guides!

Quick update - I cleaned my throttle body and PCV valve today and saw decel vacuum decrease from ~27 inHg to -25inHg; I believe that my PCV valve was clogged and causing the high vacuum reading, NOT the changes to my tune.

Update to the update - I wound up at about half of the stock "idle airflow min target decel adder" values to get the car to drop RPMs how I prefer, although I flattened out the adder values after about 4k RPM to encourage faster drops at high RPM. This has left me with a peak decel vacuum in the high -23 inHg range that drops to ~-22 inHg under constant decel, which I believe is quite good/acceptable.

Also, if someone else finds this tune/method of reducing rev hang helpful or finds more optimal settings, I'd love to hear about it!

Hey all,
Picking up on this thread, I am trying to resolve the rev hang in my 09 NA forester. I've grabbed the 32BITBASE and EZ1G107M files, but when I open my ROM, I don't get half the tables accessible that everyone is talking about. Can anyone help out?

Also wondering where the base CL fueling info is drawn from. I don't seem to have a VE table in my current definition.

Cheers

I'm playing with EZ1G107M and this is what I found.

Idle Airflow Min Target Decel tables values have biggest impact to rev hold upon shifting. But also sometimes the car revs up when in-gear at low revs. It can rise the revs up to 2000 rpm, then it switches off injectors, revs come down to 1000 rpm, it switches on injectors and it's repeating. This is the most annoying thing it does. Sometimes everything returns to normal after about half an hour of driving in traffic jams, sometimes not.

I'm pretty sure it's because ECU learns fast, unlearns slowly.

I found that that airflow adder is calculated using the following formula:



Airflow_Adder_1, 2 and 3 are calculated somewhere in the deep of ROM - I could not track them God damn indirect adressing! Short tests show that zeroing Smooth Factor #1 has biggest impact. Zeroing Smooth Factor #2 and #3 together with #1 sometimes causes RPM drops to 400 - 500 RPM when lift-and-coast uphill. I'm going to play with #2 and #3 set to somewhere between 0.1 and 0.05. It needs time - sometimes the issue comes after couple thousands of kms.

I also attach EcuFlash defs with these defined in case someone wants to try too.