[Gmecm] BLM cell confusion

[Gary Evans]

Yeah, me too. I learned a lot from this thread about the 16 cell code  
and how the BLM / INT works on those boxes. Most educational. I  
downloaded the $8d ANHT_hac and took a look. What a beautifully  
commented disassembly. Hats off to whoever put the effort into  
creating that one.

I'm not sure what you mean about the BLM never changing? The car  
rides around in cell 1 for 90% of the driving cycle. The value  
changes in response to the INT, which in turn is responding to overly  
lean/rich signals from the O2 (actually, I believe it responds to  
missing crosscounts, but the effect is the same).

So, the existing value for Cell 1 is displayed, but it is changing  
all the time. (unless the VE map is perfect, and mine certainly isn't.)

This is a *big* limitation of the $a1 and $88 code. There is only one  
global learning variable for the entire VE map above idle. If, for  
example, the VE map is lean at 1500 rpm and rich at 2500, then the  
BLM will be chasing itself all over the place trying to compensate  
for both deficiencies in the tuning.

I think the theory behind the design decision was that the VE map  
would be accurately tuned by GM Powetrain, and therefore BLM is only  
needed to handle variations in production tolerances, engine wear,  
and fuel quality that would (presumably) affect the entire fuel map  
equally.

Seems like a stupid assumption to me, but that's how they did it.

Also in a correction to my earlier post, I see that the INT must  
indeed be out by 4 before it moves the BLM, aka the "INT window":

8717        04      4 KLCITTH    4 UNITS  C/L INTEGRATOR WINDOW VALUE

FWIW, the ANHT $8d bin has a BLM update threshold of 128 +/-5 INT  
counts, so maybe the operation is really not that different between  
the two systems, simply different terminology is used.

-Gary


On Nov 14, 2005, at 3:33 PM, WopOnTour wrote:

> Thanks Gary!
> I appreciate you taking the time to clarify those points
> I will take a closer look at both items and the $88 and $A1 masks
> It's totally possible that the Tech1 is creating the cell values  
> itself in the traditional fashion, (it does a few things itself  
> that can be easily misconstrued) But that wouldn't explain why the  
> BLM stored value would ever change as according to your data, we  
> wouldn't really be switching cells, so the existing BLM value for  
> cell 1 would be that's displayed.
> Guess I can add it to the long list of "Things you thought you  
> knew" LOL
> Thanks again
> WopOnTour
>
> _______________________________________________
> Gmecm mailing list
> Gmecm at diy-efi.org
> http://lists.diy-efi.org/mailman/listinfo/gmecm