junkcltr

iTrader: (1)

I stumbled across this site last and saw the messages about using an EEPROM in place of the factory EPROM. I noticed that the EEPROM access times for the AT29C256 part is not mentioned. I tried an AT29C256-150 (150 nanosecond) part in my 1227165 ECM. I programmed the binary image at location 4000H, and programmed FF's to the lower 16K of the memory. I looked at the image afterwards to verify it was OK. I istalled it in my truck and started the engine. It would run for a second and stall. I did this several times. The engine was being flooded with fuel (could smell it). My question is, has anyone else had problems with the AT29C256-150 parts being too slow for the 1227165 ECMs?

Grim Reaper

Try binary zeros in the lower half.

junkcltr

iTrader: (1)

Thanks for the suggestion. I am going to try inserting "00"s in the lower half of the memory as you mentioned. If that doesn't work I will try inserting the data&code in both the upper 16K and lower 16K. From the sounds of it, GM doesn't do a complete address decode of the memory space in the '165 ECM if it matters what is in the lower 16K block. I guess there is possible bus contention going on somewhere........possibly. Any recommendations on generating the 16K file filled with "00"? (MS-DOS debug, hex file editor, C program, pocket programmer???)

Thanks,
J

junkcltr

iTrader: (1)

I tried the all 00s in the lower 16k, code&data in the upper 16k. The engine started and ran very rough for a second, then stalled. Same as with FFs in the lower 16k, code&data in the upper 16k. I then tried putting code&data in the lower 16k and code&data in the upper 16k. I started the engine, it did the same thing....ran rough for a second and stalled. I can read the ALDL data with the key on and it all looks fine. When I start the engine, some data gets messed up and then it loses the ALDL link. It appears to have one of two problems, either the AT29C256-150 has to slow of an access time or the memory is appearing in another address space. I know it isn't the EEPROM not being programmed correctly because I look at the binary before I program it into the EEPROM to make sure I created the file correctly. And after programming, I do a verify and it says it is OK. Has anyone tried these AT29C256-150 parts or does everyone use AT29C256-120 parts (120 ns)??

mta227

150 might be seem a bit slow, but I doubt that it would be an issue. Consider that the processing power of the typical PCM is about equivilent to an old IBM XT...
I would check your connections between the calpak and the eeprom. For example, if you soldered in the eeprom, its pretty easy to get a bad connection on a leg not required to burn the chip. Will program fine, but totally mess up in the car.

junkcltr

iTrader: (1)

I did solder a socket in the plastic carrier. I checked all of the connections with an ohm meter (continuity tester). I also checked adjacent pins for shorts. It all seemed OK. I just double checked it and they seem fine. The strange part is that the EEPROM works when the engine is off and the key is on. I receive a perfect ALDL data stream. This leads me to think that it isn't a read access timing problem. The problem arises when I start the vehicle. Then it all goes bad....ALDL locks up and engine runs rough for a few seconds. It seems like the EEPROM is being read when other ECM I/O is trying to be read. I just ordered some AT29C256PC-90 FLASH parts and some other stuff I needed from JDR Elec.
I have read that some people are trying 00s, some trying FFs, and some making the memory a double data&code image. Does it really matter? All that should matter is that the data&code resides at locations 0xC000 - 0xFFFF. Do some ECMs ONLY work with a certain configuration?
I do agree that the Motorola microprocessor is a slow part. It is running at 8 MHz. Judging from other microprocessors I have seen, the 150 ns memory should be in the ballpark......which I had the timing diagrams.

Grim Reaper

150 ns should be more than fine.

junkcltr

iTrader: (1)

I totally agree with you an the 150 ns parts being acceptable. Just by your comment I can tell that you are into electronic type stuff........you seem like a smart guy. The thing I did not mention is that I have a few appox. 10 ns gates driving the signals. That is becuase I am not really using the Atmel 29C256-120ns parts. I had some EEPROMs lying around that were 150ns but needed logic in front of them to immitate the 27C128 parts. So my memory is really like a 180 to 190ns part. This is why I ordered the At29C256-90 parts. I won't have to use the gates with the Atmel parts. If there is a problem with using Atmel parts then I know it is a manufacturing error (my soldering). Again, thanks for the help.
J