Is it possible that an intermittent code 36 that triggers under the very specific conditions of shutting the car down before it has reached operating temp, but long enough that the the burnoff cycle triggers could just be unfixable with modern parts? My MAF is one of the later Bosch units with the probe sticking into the open space. (Not a platinum wire strung around the thing.) I've read that some aftermarket MAFs with modern style sensors do not need the burnoff function and instead send a BS voltage reading during the burnoff cycle that the ECM is expecting to see. My code 36 clears itself upon the burnoff cycle successfully completing once the car is shut down with the temp gauge reading at least 200. I just wanted to put this out there before people tear their hair out chasing down a problem that cannot be fixed without finding old stock parts. I have also verified activity within range on the pins for the MAF and at the ECM as well as using an ALDL cable connected to the computer to check everything out. As long as I remember to let the car warm all the way up before shutting off, it doesn't give me a lick of problems.

 

I believe the burnoff cycle does not initiate every time the engine is stopped. The run timer and other inputs are involved. It's been so long since I reviewed the binary that I am not completely clear on the sequence, but those are points I do remember.

 

My solution was to find an MAF sensor from a wrecked Corvette. Original 1989 Bosch MAF. Paid $120, but no more CEL. It was worth it.