Thought I would share one of those tuning "mystery" stories. Sometimes you've got to play detective. I have been busy tuning my LT-5 for the last 2 months. Over the winter I added a set of ported cylinder heads (AFR or Trick Flow don't make DOHC 32V high flow heads) and decided to take advanyage of the additional airflow by installing larger single pattern intake cams. This is all on a stock block 5.7L LT-5 motor w 1 7/8" Watson headers and 3" Magnaflow exhaust. These babies LOVE to breathe. Very much a top end motor. Not much tops an LT-5 at full song around 7000rpm.
Anyway, have been playing with the rune for a couple of months when I noticed the BLMs are not updating. The LEARN flag is OFF although motor was going into C/L. I thought possibly there was a problem with my new copy of TPRT 5.0. I also "noticed" that the bin where all this started had a Checksum that reset, IOW it went beyond the 65xxx. I do multiple changes to a bin but none were of the kind that should affect LEARN control. Or so I thought. It took me a couple of days of eliminating the parameters that had changes between the bin that DID work from the bin that DID not. It turned out to be coincidence that the Checksum reset at the same bin change but had nothing to do with the LEARN staying off. Using Craig's Ostrich 2.0 and Mark's TPRT 5.0 really made the sleuthing a much more effcient process. Still took me most of the morning today BUT I finally got it. With the larger intake cams, you invariably get the rich smell at idle and I had lowered the O/L change v CTS tableto take some fuel out. Apparently the code doesn't like too much fuel pulled out because it appears to prevent LEARN from turning on. Essentially you are running O/L although the ALDL says you are in C/L.
Made the change to the bin, download and bingo LEARN ON. I re-input the stock values but for grins I would like to finnd out how much I can decrease fuel before r-creating this issue again. Maybe one of the hack gurus can tell us what the logic is behind this outcome. After going thru this, I thought I would have LOVED to have an ECM bench.

 

Without knowing squat about your code, I'll bet there is a "CL o2 volt" parameter that is being violated (unsatisfied) enough to interrupt learn.

 

With some parts that are unique to running a motor with 16 injectors and a secondary throttle mechanism, the def file is pretty recognizable by someone familiar with an SD system. The LT-5 code is actually a derivative of the L98 code or $8D that existed. Too expensive to start from scratch and not enough time for a run of 6900+ motors. So they took what they had and modified it to fit with some compromises. I thought about the O2 volts since I did lower them at idle due to the increased overlap mimicking a lean condition.
But that was only at idle. The Learn flag was off regardless of airflow/rpm etc. Cruise voltages were left as stock. I had lowered the Prop Gain v O2 error and Prop Gain v airflow to help with the surging usually encountered w larger cams. Perhaps this affected the O2 swing or transitions. However, the datalog shows the C/L flag coming ON. Hmm?

 

I've been reviewing the def file for my 92 ZR ($D0A) and I noticed there is a constant for O/L Change Bias v CTS of -13.67%. Coincidently?, the values in the O/L Change Bias v CTS do not go below -13.67% beginning at 80C on up.
I had set the O/L Change to -11.67% from 80C on up in an attempt to lean out O/L idle. I have a suspicion that anything less or should I say leaner than the Change Bias number prevents the LEARN flag to be flipped ON. At 80C and above with a Change Bias of -13.67, the ECM won't modify the AFR produced by using the appropriate VE table value (RPM x MAP). Makes me wonder if there's a way to run O/L idle. Perhaps setting the O2 idle voltages to the point where the O2 sensor doesn't swing thru the O2 window at idle.
Hmmmm?