NefMoto

Symptom - after getting the i-limit set up perfectly, after a while, the car starts overboosting.

Clear codes and reset adaptation, overboost goes away.

Problem? It appears as though the i-limit adaptation is a drunkards walk - i.e. it ONLY adapts upwards, since the MAP limit prevents the adaptation function from seeing an overboost large enough to cause a negative adaptation.

Opinions on best way to fix? Don't say openloop WGDC, thanks :)

I'm guessing reduce LDEIAU and TLDIAN?

Or, alternately, disable upregulation completely?  might be safest...

Fix the scaling in the ECU and run a big enough MAP sensor :P

Stop guessing, log.  ;D

I hate both of you.

gonna need a WHOLE lot more ram variables :) How much of the ldr i-limit adaptation path can you find for me ;)

Haha, was going to say the same thing, figure out how to throw a bigger MAP in there, and don't be at the limit ;)

nyet loves living on the edge!

 ;D

I have it all identified and labeled. I can give you the addresses no problem. Is there a specific sub-function you want to start with?

hmm:

ldimxr_w
ldimxrk_w

B_ldimxp
B_ldimxa
B_ldipos

B_ldimxn?
B_ldimxp?
B_stldw?
B_stldia?

(?) not sure if needed

or anything else you think i need for i-limit adaptation?

0x382744 ldimxr_w
                ldimxrk_w (written to GPR only)
0x382742 ldimxak_w
0x38273E ldimn_w
0x382740 ldimx_w

0x00FD96.9 B_ldimxp
0x00FD96.7 B_ldimxa
0x382763.1 B_ldipos

0x00FD96.8 B_ldimxn?
0x00FD96.11 B_stldw?
                    B_stldia? (doesn't exist, hex value 1-4 gets written directly to stldia)
0x380C73 stldia
0x382764 stldia_old

Thanks phila. I'll post results if i get anything that works.

Here is what I ended up doing:

First, tried shifting TLDIAPN (RPM is based on k03 spool, so it is pretty aggressive at low rpm) bad news, no matter how far I go out, any persistent lde will result in adaptation. So if you have a stock DSLOFS, and are running past the MAP limit, good luck; the ECU will keep adding I limit forever, since MAP limit is always under requested.

So I chose to nuke it entirely by setting LDEIAP to max (1270) on a locked WG RS6 turbo file (it is locked via KFLDIMX, not KFLDRL)

It is worth noting that stock LDEIAP is actually negative (-20), so even if you aren't over boosting, positive adaptation is enabled.

I'm not quite sure what my approach is going to be for my car (DSLOFS=0)... since really I just want to numb positive adaptation, not completely kill it..

Also, for debugging negative adaptation, I'd need B_ldimxn ...

I hope you're still reading phila :)

This didn't work?

No. I'm just a retard and missed it.

I'll relog when I get back on the dyno..

thanks again phila, you're a life saver.