going lean

[bermo]

So I thought I found my issue.

 

Poking around hoping something obvious hits me in the face, I notice fresh dirt around the throttle body the looks like a leak. Pulled the throttle body out (bolts were tight) and measured the o-ring height at 6mm. Got the new o-ring and measured it at 7mm. Hmm.

 

Bolt her back together and notice more o-ring crush tightening the bolts. Good sign.

 

Reset ECU with the Cobb and take her for a spin. Went to my usually trouble spot where I can get the stutter and voila! No stutter. Sweet.

 

All AF learning tables were low if not zero. Then, either the radiator fans kick on or the A/C compressor kicks on and the AF numbers start to climb. Hit +10 on AF Learning C all the while the AFR on the wideband is nice and steady. I'm not making this up, honest.

 

Taking it easy on the car but driving it around I noticed AF Learing A showed -5ish. So one table is pulling fuel while another is adding fuel?

 

Numbers are still moving a bit, maybe another ECU reset?

 

 

Ugh.

 

 

New 7mm on the left, old 6mm on the right.

[solidxsnake]

If AF learning/correction is doing it's job, your wideband AFR should be steady. After an ECU reset, all of your learning tables (fueling and knock related) will be cleared to zero. This could also explain why that stumble is gone, if the stumble was in fact caused by learned knock correction.

 

 

Assuming everything else is consistent, after a reset your numbers should settle pretty much where they are now. Still looks like a leak to me, since I assume Dave's taken a look at your logs and is confident the sensor calibrations are all good.

[covertrussian]

That's why I don't trust smoke tests, if you can do a smoke test + pressure then it would be awesome. Most boost leaks don't show up until your pipes are pressurized.