No sumcheck code in a '95 MEMCAL

[Roger Heflin]

On Sat, 7 Nov 1998, Ludis Langens wrote:

> Roger Heflin  <rah at horizon.hit.net> wrote:
> > There is not other rom on the board, Motorola has one in the CPU's
> > themselves, and there is the possibility that the car comes up with
> > the other rom, checks the socketed rom and resets into that rom with
> > the CPU rom disabled.  That is something I could see them doing.
> 
> The older C3 ECMs don't have ROM in the CPU.  (There is ROM in an ASIC,
> but that is in addition to the PROM, not a replacement.)  I can't (yet)
> say for certain, but the P4 generation looks to have a ROMless CPU also.
> 

I am not sure how to tell what mine is.  It has a 32kb rom.  It is
also one of those in engine computers, so it must be a bit more
hardened against heat.

> > My
> > car will run with a bad rom (read bad checksum, SES light) which would
> > imply that it may run without a rom (it runs really really bad), which
> > would imply that there is code someplace else running things.
> 
> Not code - but an entire extra analog computer "programmed" with the
> resistors in the MEMCAL.
> 
> > I will have to see if I can find that code.  I have not found any
> > loops that run through all of the rom (or a large part of it) yet.
> 
> If your ECM is a P4, look at all the places that write $FF00 to location
> $400B/$400C.
> 

I did find the checksum code (finally) it is at FC31, it is called
once in the setup code (initial setup code), this code is also called
on what appears to be a reset.  It looks like it hits the watchdog
every so often, and it does apear to go through all of memory.

> > I wonder why they check the rom on reset?  Do they think it is that
> > likely that it could have went bad? or are they just covering all of
> > the bases
> 
> It has been standard practise to checksum the ROM in computers since the
> IBM PC.  I think I have seen a ROM failure or two (in personal
> computers, not ECMs.)
> 
> EPROMs have a limited guaranteed data retention lifespan - something on
> the order of 10 years.
> 

They could have also got cute and put it in hardware (that is if
things needed to be done really fast.

> > finally decided that they had not had any of the type of
> > problems so quit doing it?
> 
> Two ideas, both having to do with switching to OBD II style error
> handling:  OBD II may require the CPU to try running even with a damaged
> PROM.  Or, while the engine is running, OBD II may require the CPU to
> recover very quickly from a hardware (i.e. watchdog) reset, leaving no
> time to sum the PROM.
> 

I don't think so far (for the f-bodys) that anyone is reprogramming
any of the ODB II computers (96+), they almost always seem to replace
the 96 compuer with a 95 computer before even trying to reprogram it.

> > I guess you might try putting a bad checksum in there and see if it
> > notices.  If it does, it would seem to be taking care of it somewhere.
> 
> Difficult at the moment.  The engine&ECM in question have no vehicle to
> call "home" at the moment.  Plus, I think the engine's owner has it torn
> down to rebuild.
> 
> The checksum was already "bad".
> 

The could also have gotten cute and used a slightly different
checksum algorithm.

> -- 
> Ludis Langens                               ludis (at) cruzers (dot) com
> Mac, Fiero, & engine controller goodies:  http://www.cruzers.com/~ludis/
> 
> 
			Roger Heflin
			93 Z28