I've been working on a tune for a ME7.5 1.8T 2003 VW GTI.  I am currently getting hung up on an issue where the ECU often goes into limp mode during part-throttle.  I am having no issues with WOT.  It seems as if when I boost into the 20's (PSI) and then ease off the throttle, there is something that the ECU is not happy with, and I can't put my finger on it.  It happens rather sporadically, and often enough to annoy me.  Normal operation is restored when turning off ignition and back on again. (Soft limp).

I usually find one or both of the following DTCs after the problem occurs:

The problem occurs regardless of a cold start or warm engine.  Diverter valve operation appears to be normal.  I swapped out the N75 with another with no change.

I am attaching some log excerpts I grabbed with ME7Logger.  If you take a look at Column L, (N75 duty), you'll see that the magic "limp" number is "5.07813".  Apparently 5% duty cycle is all you get in limp mode.  You'll also notice at the same time, my pedal request is sometimes over 80% or near 100%, which obviously indicates the ECU has intervened with component protection.  But looking at the Boost values, Req vs. Actual, I really don't see what's wrong. 

Maybe the boost isn't falling fast enough when I ease off the throttle?  Could an incorrect KFMIOP table cause this?  Also, my specific ECU does not contain a KFDLULS, so I'm not sure how to modify the table for overboost protection.  I'd rather address what the ECU thinks the problem is.

My WinOLS project and logs (xls) are attached below.  Check out the charts also in the xls.

I'm almost certain I've read of this issue on here before with the 1.8T cars... I have no idea what it was, but search and you will find your answer.  It's in here somewhere.

Good luck

Thanks very much.  Zeroing out FVPDKLDUS and testing this evening.  Also, since I do not have a KFDLULS, I raised the value in DLUL which will hopefully prevent the overboost code.

Here was the thread btw:

http://nefariousmotorsports.com/forum/index.php/topic,318.0.html

I'll follow up with my results.

Yeah, I've had it scaled up between about 30 and 50% while the problem has been occurring.

This is most likely due to either failing DV, vacuum leak which results in DVs operating "slow" (no vacuum assisted actuation of DV) or you deleted your N249 valve (like me).

Fix the vacuum leak/replace DVs or modify maps if you deleted your N249. For N249 deletion, you need to neuter some Boost LDOB maps to not react on overboost condition anymore....

In properly operating boost system, DVs open immediately after the driver stops requesting boost by N249 switching on and connecting DV vacuum hose with vacuum reservoir. This forces vacuum into DV immediately and it purges charge. Vacuum reservoir is located behind check valve, this allows it to stay under vacuum even when manifold sees positive pressure.

If you delete N249, you lose the ability to immediately purge DVs via N249 action and must wait until the manifold goes into vacuum all by itself which could take appreciable amount of time... or never if you just slightly let the gas pedal go. In such a case the desired boost might sit at 1000mbar while you're still kicking tires with 20 psi for a second or two before everything calms down. This in turn might cause the overboost condition if the vacuum doesn't kick in quick enough.

KFDLULS defines maximum delta which is easy to exceed without properly operating N249 (or boost leak at DV). Playing with time tables for LDOB might make more sense. One table in particular is very popular with tuners for disabling overboost detection.... TLDOBAN. You zero that one out and it never goes into effect.... Overboost protection with anything larger than k04 is pointless anyway since you will be operating above MAP limit all the time hence the system doesn't work anymore... only gets in the way.

YW.

Last thing to find out is to figure out what causes timing oscillations. It is definitely related to my tune exceeding/howering around max torque and is directly related to my insistence on having working PID... I didn't have this issue when I run underscaled on MAF and only reared its ugly head after I scaled the maf properly...

Well yeah. Load, LoadDesired and CorrectedLoadDesired. That's the TQ as engine sees it or wants it.