Re: Auto learn code-Where?

["Jeff M" <tystorm@email.msn.com>]

I think Scot's idea would work better/faster for him if he were able
change/increase the update rate of his OEM chip and/or in the added learn
code, if that can be found as well, might need more than a disassembler to
find those values from running the code.  I like your idea, I have upgraded
my heads and cam quite a bit so I know my VE tables will need some work, and
I anticipate running my DiacomP to watch the BLM and INT (and other values)
so I can make the necessary changes to the chip while I drive via the
emulator, provided I take a few other fueling calcs temporarily off line to
not have them skewing the readings.  I know my local tuning buddies are
itching to be the driver while I tune with 2 notebooks in my lap ;-), they
also feel they owe me so much from all the fixes and chip tuning I have done
for them and their TGPs, but I more so enjoy the shared passion for turbo
cars and the excitement when the race stories come in, "Jeff, I beat a
(stock) WS6" or their favorite, Type R Integra and NOS Hondas ;-).  Jeff M.

> You could also (without changing the code) use a scantool such as
> diacom and record really really long drives, and write a fairly simple
> program to collect the same data out of the data stream.  You would
> probably need quite a few hours of driving but that would give you a
> really really good ideal of where the voleff's need to be for each
> combination.    The one advantage of doing it this way is you can
> change the way you analyze the data (adjust the program).  I have a
> basic (qbasic really) program that shows all of the normal 16 cells
> (4x4) and averages the BLM's/Ints and show the range and stddev.
> Just a few changes and you would be able to get the above info out of
> that data stream without changing the assembly in the actual car
> computer, it with only a little modification for each different
> diacom/scantool data stream (code to disect the parameters) you could
> apply it to a large varity of GM type ecms.
>
> Roger

> >
> > Thomas Matthews wrote:
> >
> > >> Scott, would love to hear details about the code changes you made
> >
> > >I found the idea in some GM code I was looking at.  Took the idea and
> > >made it work in my ECM.  What it does is break the engine operating
> > >range into 256 cells (16x16 matrix).  The RPM range on one axis split
> > >into 0 to F, and MAP on another axis, also split into 0 to F.
> >
> > >It checks to see if it is operating in the same cell for one second.
> > >If it is, it looks at the integrator, and checks for low error.
> > >If low error, the present VolEff number is corrected by the BLM
> > >correction value and stored in the matrix at the appropriate
> > >place.  The air temp is also recorded in a separate 16x16 table as
> > >a check on the air density table.
> >
> > >Quite easy, really.
> >
> > >Scot Sealander




----------------------------------------------------------------------------
To unsubscribe from gmecm, send "unsubscribe gmecm" (without the quotes)
in the body of a message (not the subject) to majordomo@lists.diy-efi.org