Inside scoop on CASE relearn?

[1project2many]

I'm looking for information regarding this procedure. I have plenty of service information so I know when it's supposed to be done and how to make my scantool / software do the job. But I don't know what the learned value is actually used for and there are plenty of conflicting answers on the interwebz. Anyone with GM info or firsthand knowledge out there?

Is the information learned during CASE critical to correct spark advance?
Can the pcm work out cam / crank correlation without CASE? Can it work out correct crank position without CASE being run?
Is CASE information only used for misfire diagnostics?
Any clues as to how the learn procedure determines works? Does it detemine crank position? Does it calculate other values?

Thanks.

[EagleMark]

After I did a 0411 swap that ran fine and had no issues or error codes I was wondering the same thing? Someone told me it was only needed to clear code and if code didn't set the only thing it was used for was misfire detection... it can be done without scantool...

http://www.gearhead-efi.com/Fuel-Injection/showthread.php?1375-General-motors-vehicle-computer-relearn-procedures

[ericjon262]

wow, that thread is helpful... sticky so it's more visible?

[daleulan]

I am not 100% sure whether GM uses their adaptive CKP learn to correct ignition timing but I don't use it for that in my software. The timing variations learned from what I have seen are so small that the timing isn't really affected much - less than 1/2 crank degree. I use it for engine misfire detection. Usually you can do misfire detection at heavy loads or idle and low engine speeds without the values but misfire detection through the OBD/EOBD mandated range (approximately the zero load-line up to engine maximum with a few other limits) requires such small deceleration/acceleration measurements (in the order of 10 or 15 microseconds of delta per cylinder) and machining tolerances can be much higher than that. It is a bit sobering to look at the raw tooth time measurements and see what a challenge the misfire detection algorithms have to deal with.

[1project2many]

Thanks, Dale.

This is good information. I simply haven't had the resources to connect multiple vehicle to measure then compare crank signals. It's ironic, too, because we have a fleet of similar engines here so it would be a good sample to work with. I don't envy anyone who needs to write misfire detection strategy based on crank speed. When OBDII was still new I spent a bunch of time watching crank signals during various conditions trying to get an idea what it took... big job. Then for a time it seemed like ionization detection might become popular but my understanding is that it takes quite a bit of effort to characterize those signals as well so there's little benefit from the technology in the end.

[Hog]

wow, that thread is helpful... sticky so it's more visible?
I've had that for a long time, it's been on 3 of my PC's. Here is teh link.

http://www.bsecorp.com/files/techtips/ST9303l.pdf

peace
Hog

[EagleMark]

Thanks Hog! pdf is now attached to that first post.

[Playtoy_18]

Haven't read link,but here is how I understand it.
Case relearn basically gives the pcm a chance to relearn the minute differences in the reluctor/sensor relationship.
To properly flag misfires,the computer needs to know the exact "dimensions" of the signal or a baseline.
Small variations in the reluctor/sensor tolerances mess with the pcm's ability to determine how quickly the crank speeds up/slows down between strokes. Which it uses to determine misfires.
Once this "relearn" of a basline for those variations are done the pcm can accurately guage misfires,run misfire monitor etc.
Once it is learned,it does not change unless one of those components are changed (ckp/pcm/reluctor).
It is written onto the pcm and not changed unless triggered to do so with the relearn.

I could be totally off,but this understanding has worked for me.
It is the same thing a Ford requires though they call it a KAM reset iirc (keep alive memory).
On Fords a KAM reset is needed to clear certain learned values that don't clear with a code reset.

Now,i'll read all the links and see how wrong I am :)