late 95+ ECU performance (on turbo Talon & Eclipse)

[Scott Ritter]

Hi. I've just joined the list and am currently trying to debug a performance 
problem that likely resides in the ECU's of late 95+ turbo Talons & Eclipses. 
Apparently, early 95's, which sport a visibly different ECU board (=> 
different ECU code?), don't have the problem. The problem is that when the 
boost pressure is raised by as little as 1psi above factory nominal, the 
engine suffers from an abrupt power loss due to major timing retard. The 
problem is not consistent at any specific boost level, though; and the timing 
re-advances about a second after the throttle (boost) is reduced.

I've been hypothesizing what might be the (changed) ECU logic behind this 
symptom & some potential hardware work-arounds, and have come up with the 
following list. Any feedback or additional input would be greatly appreciated! 
(I've sent a copy to the Talon Digest already.)

a) MAS air flow/RPM, (computed boost).
Perhaps the new logic includes a test for boost above 12-ish PSI & retards 
timing as the response (prior to the full thrash of fuel cut, at a still 
threshold). If this is the case, then an FCD (commercially available unit to 
limit the airflow sensed) might help some, but the lean fuel condition would 
probably make it too dangerous a fix. This situation is the worst case, since 
the only real "cure" is to change the actual ECU code.

b) Rich exhaust from the post-cat O2 sensor.
I'm going out on a limb here, please excuse the ignorance-factor... Assuming 
that the 2nd O2 sensor is in place to let the ECU know that the cat's doing 
its job, and that going WOT removes careful stoich control on the mixture, 
then
the engine's gonna tend rich (like in the 1st gen's) to handle the big air 
coming in. The high volume of rich exhaust will cause an apparent degradation 
in cat performance, which the ECU will detect. *If* the ECU reacts to reduce 
emissions at this point by reducing timing(?), then there's the problem! If 
this is the case, then putting an ECU-appeasing voltage gain & level-shift 
circuit between the pre-cat O2 sensor output and where the post-cat O2 sensor 
used to be ;-) should eliminate the effect. A software fix would still be the 
cleaner solution, but until that's available, this might be a viable & 
affordable solution.

Question: Would simply pulling the post-cat O2 sensor from the exhaust stream 
(but leaving it attached to the ECU) be likely to confirm/disprove this theory?

c) Pinging
If the ECU's ping-level tolerance has decreased, then we might simulate 
reduced pinging. We could insert a divide-by circuit between the knock sensor 
& the ECU to reduce the number of pings sensed by a jumper-programmable 
factor. Again, there's nothing like a nice software fix, but a divide-by 
circuit should solve this one.

Thanks,
/Scott Ritter
--------
"In the beginning was the thing. Then one thing led to another." T.R.