repair an 8051 PCM

[space387]

So I have found the walk through on how to remove and replace the botched memory chips for the 8051 ecu but when it come to what you have to flash on them and with what device you flash seems to be clear as mud. I have a Moates Burn2, first off if I find an adapter can I use this to flash? Second I am fairly sure it was the E-side that died but would like to be able to swap them both out, can some one direct me to the bins to burn or do you send the tune like you would through EEhack?

Thanks

[NomakeWan]

If you get an adapter, the Burn2 should work just fine. I have a cheap Chinesium MiniPRO TL866 and it does the job, in case for some reason the Burn2 doesn't have a definition for the chip in question, chances are the TL866 does.

You absolutely need a BIN, because the EEPROM programmer uses this binary data to flash the chip. If the data on the chips are bad, then the PCM won't communicate over ALDL, so you can't flash with EEHack.

The BINs are not the same BINs you'd use for EEHack, because for some reason the raw chip data is "scrambled" compared to BINs flashed over ALDL. The correct BINs for burning to the chips directly are here: http://www.gearhead-efi.com/Fuel-Injection/showthread.php?1160-64K-bootable-image-for-95-LT1-PCM&p=14452&viewfull=1#post14452

Best of luck!

[steveo]

...when it come to what you have to flash on them...

yeah that's a bit unclear. there's a thread on it here but since i can never find it, i should make a page on my site about it. here ya go... http://fbodytech.com/bin-files-2/ecm-repair-bins/


...and with what device you flash...

any programmer with a socket adaptor to fit those eeproms will work fine. your burn2 is totally capable of that.

[space387]

Steveo I was searching and saw you post something about trying to flash the chip on the board. Did anything ever come of that?

Also I'm sure I'm not searching correctly but can anyone find a link to the adapter I need to flash the mem chip with a burn2?

Come to thing of it if some one can pass along the pinout cross reference of what pins match what spot on the burner I'll make my own

[steveo]

Steveo I was searching and saw you post something about trying to flash the chip on the board. Did anything ever come of that?

nah never finished that project. believe me if i did i'd be offering low cost repair services for these ecms. maybe one day when i get another test bench ecm going

[lionelhutz]

Check ebay for a PLCC32 to DIP32 adapter. Should be easily under $5. Too cheap to try and do yourself. But, if you did want to, then download the datasheets or Google Image search the 28F010 flash chip and you'll see pinouts for both the DIP and PLCC versions. The adapter would be made by matching it up pin for pin PLCC to DIP.

[NomakeWan]

That doesn't help you do it on the board itself though. I did find a potential way to do that but I haven't wired it up yet to try, not to mention I don't have a spare PCM laying around I can sacrifice since both of mine are in running cars. If you're trying to do something on the board, an adapter to try that out is gonna run you a lot more than $5. ;)

[lionelhutz]

When did this thread turn into one about programming the chips while they are still on the board? It's probably not a good idea to try that anyways. If your PCM is working then you flash it over the ODB port.

The OP asked about removing the chips and socketing the PCM and a <$5 adapter will put a PLCC chip into a DIP programmer socket.

It's not clear from the Moates description if the Burn2 would do it, but it doesn't appear it can. Possibly try the 29F040 chip selection but only do 1/4 of the size to program the same size as the 28F010?

[NomakeWan]

When did this thread turn into one about programming the chips while they are still on the board?


Steveo I was searching and saw you post something about trying to flash the chip on the board. Did anything ever come of that?
Right about here. ;)

[steveo]

wow i never realized the burn2 was such a limited programmer, that's too bad. might not work. unclear if it's a software limitation or what

[steveo]

When did this thread turn into one about programming the chips while they are still on the board? It's probably not a good idea to try that anyways. If your PCM is working then you flash it over the ODB port.

there's a header on the board with each pin leading directly to the chip, probably GM designed it for programming on-board, however it's unclear how/if it works. it's definitely not something i'd 'just go ahead and try' without someone else doing it first.

[lionelhutz]

Trying to drive the address and data lines when the onboard circuitry is also doing the same isn't generally a good idea.

[NomakeWan]

How about when they aren’t? If the PCM isn’t powered up, why would connecting directly to the flash chip be a problem?

[lionelhutz]

You can't put power on the flash chip to program it without that power also powering the rest of the circuit assuming it all shares a common power supply.

But, say the other circuitry had no power. There usually is transient voltage protection on each pin of an IC that is built to keep the pin voltage between the voltages applied between Vcc and Vss. So, with both Vcc and Vss at 0V, the pins shouldn't have any voltage applied to them at all.

You should start your own thread to discuss if it's possible to program the chip directly instead of thread jacking the OP's post asking for help to repair his PCM.

[steveo]

the OP is the one who asked about this in the 4th post
so i answered
others then discussed
what’s the problem here
you cant thread jack your own damn thread man

[NomakeWan]

You can't put power on the flash chip to program it without that power also powering the rest of the circuit assuming it all shares a common power supply.
Except that this isn't necessarily the case. For example, I have an ICP header on one of my prototype boards that goes to the pins on my microcontroller required to program it were it to go faulty. Those pins are also used to talk over I2C to a DAC, and for other things. But if I'm not powering the board from the input rail and am instead connecting to the ICP header, then the rest of the circuit is unaffected as it's not powered up. Sure the signals I'm sending to the microcontroller are going to also hit the DAC, but the DAC isn't receiving any power so it won't matter.

Similarly, if GM designed a header to allow in-circuit programming, why would you assume that programming the flash chip via this header would somehow power up the whole PCM? Similarly, if one were to make a jumper connector to flash the chip directly, why would the rest of what the chip's connected to matter since there's no power to the PCM?

[steveo]

my first dumbass thought was nipping the processor vss and reattaching it after. but im an ameture with this stuff and the chances of it catching fire are greater than zero.

[space387]

So this bytes. Got my parts in but found the Ebay programmer I bought requires a parallel port and the program for it wont accept a serial port. Working with Win 7 on that computer if it helps.

I looked back and the only difference I can find from the 28F512 in the 8051 ecm vs the 27SF512 used in my Corvette is pin 2. the 28F leaves it blank where the 27SF list it as A16. I am really out of my league reading these data sheets but one would think you could use the Burn 2 with an atapter and a little adjusting of some settings.

27SF512 Data sheet (http://static.moates.net/zips/27SF512.pdf)
28F512 data sheet (https://www.digchip.com/datasheets/parts/datasheet/227/AN28F512-pdf.php)

For the most part it seems like all should carry over but I'm not sure about the unused pin and how that would play into the programming.

Side note I have the sockets in and set already, that went a lot easier than I expected

[kur4o]

Don`t bother with burn2. They noted that other chips are not supported. It could be a software thing or a lack of 12v source needed for erasing and writing.

Get an old PC with parallel port and xp and use the programmer you have. There are no requirements other than parallel port, so look for the cheapest one.

[NomakeWan]

Space, when you look at the datasheets the only thing that matters is the programming/flashing section. This section explains which pins are used for this and how. I guarantee you that A16 won’t matter for programming purposes, so if the rest of the pins are the same, the reset vector is the same, and programming voltage is the same, you should be fine.

[space387]

Don`t bother with burn2. They noted that other chips are not supported. It could be a software thing or a lack of 12v source needed for erasing and writing.

Get an old PC with parallel port and xp and use the programmer you have. There are no requirements other than parallel port, so look for the cheapest one.

the 27SF512 sent by Moates denotes an 11.4-12v VppH for writing/ erasing the chip. not sure how the Burn 2 gets it. I had an older computer that I have not used in a few years but on trying to start it up for this it wont even post.

[kur4o]

They are different type of chips. The 27sf512 is eprom erasable type and the n28f512 is flash type. It is likely to use different commands set for the flashing process. So it must be a software limitation.

You can try contacting Moates with the issue. They should be able to tell if it is possible.

[space387]

Thanks Kur40, I emailed moates for the kicks and giggles but dont expect a fix.

[steveo]

amazon will have a usb to parallel or pci to parallel solution on your doorstep overnight for like 15 bucks

there is a usb to anything adapter these days

[space387]

I finally did it. I resurected my 12 year old gaming PC with a parallel port and worked my way through the files and switched but I have successfully burnt a chip, I think. About to fill the other and try to remember which one is T and E side

[steveo]

nice! the guys that originally developed this program probably had a 486. nothing wrong with old hardware.

if it doesn't boot at first, after you ensure your socket is good, remember the height of the chip in that stupid socket is critical. since you have to snip the 'floor' out of the socket, and putting it back is a crazy glue job, you might need to shim the chip until the pins make contact.

i fought with that one for a bit my first time thinking my chips were programmed incorrectly

[steveo]

blue plug tside red plug eside by the way

[space387]

nice! the guys that originally developed this program probably had a 486. nothing wrong with old hardware.

if it doesn't boot at first, after you ensure your socket is good, remember the height of the chip in that stupid socket is critical. since you have to snip the 'floor' out of the socket, and putting it back is a crazy glue job, you might need to shim the chip until the pins make contact.

i fought with that one for a bit my first time thinking my chips were programmed incorrectly

thanks, I actually just tested it in the can and it wont communicate but its late and I have to work the next 2 days. I will absolutely try shimming before going crazy though.

[space387]

IT LIVES!!! Seems I didnt have good contact on a few pins. I also noticed the data stream is more consistent.

* it lived. After one startup it wont communicate again. maybe the chips moved,a project for another time

[space387]

So steveo I am now being met with an issue sending a tune. In EE hack I open up the program and connect just fine. I see data like Map fluctuating and TPS responds when I push the pedal. When I go to flash and hit stability test it says it can not get the access key to release( I think) and says to cycle the key but nothing changes it. Any Ideas? I am using the starter bins from your website.

[steveo]

your aldl comms might be sketchy, but is it throwing any DTC errors?

[steveo]

oh also make sure the ecm has been booted up for a little bit (30 secs i think?). before flashing

[space387]

your aldl comms might be sketchy, but is it throwing any DTC errors?
no codes that I can see.


oh also make sure the ecm has been booted up for a little bit (30 secs i think?). before flashing

I will try to let it sit for a little bit next time.

[space387]

Letting the pcm "boot up" made all the difference. One thing I am noticing, one of my chips keeps hanging at the same address. Is this likely a failing chip? I am also seeing data breaks while data logging, this proceeded the first crash.

[space387]

so its crashing at adress 2D80 on 2 different chips. I am going to try another bin but is there a way to burn the bin directly to a chip or is the bin we use to revive a chip just pulled from a living chip?

[steveo]

yeah it’s pulled from a living chip
ill build a converter soon
i have a new 8051 ecm coming thanks to a really awesome eehack user
ill backtrace those stupid address lines and eehack will be able to spit out a bin in the either format in like 0.5 ms

[space387]

That would make this soo much easier. I had this stupid thing crash 5x today trying to update a spark table.

so cant get it to take a tune now. Looks like I'll be driving to corvette for a while

[kur4o]

That would make this soo much easier. I had this stupid thing crash 5x today trying to update a spark table.

so cant get it to take a tune now. Looks like I'll be driving to corvette for a while

I have seen similar behaviour with bad cable. The flash goes smoothly till some point, when the cables freaks out. Also a PC or laptop crap power supply can totally destroy usb communication. And the last reason is not enough current or low voltage supply. Are you trying to flash on a bench with some shoddy power supply?

[space387]

I have seen similar behaviour with bad cable. The flash goes smoothly till some point, when the cables freaks out. Also a PC or laptop crap power supply can totally destroy usb communication. And the last reason is not enough current or low voltage supply. Are you trying to flash on a bench with some shoddy power supply?

The cable I am using is also used on my 1990 corvette with a clean signal. I am putting the pcm back into the car to burn as it's the only way I have to connect and power it. I have an extension cable to power the laptop to make sure it doesn't run out of battery.

[steveo]

it really sounds like the chip to socket to board connection might be imperfect
are both eehack and tunercat/winflash producing the same error

[space387]

haven't tried tunercat yet but I will absolutely check tomorrow to be sure I have good contacts within the socket.

[space387]

Any luck with the bin splitter steveo?

[steveo]

nah haven't worked on it yet sorry

i'll probably be more motivated when my test bench is running again

[space387]

nah haven't worked on it yet sorry

i'll probably be more motivated when my test bench is running again

ok, I was finally able to get my pcm working again. Not sure if positional or the chip but the last iteration flashed both sides without a single error so I'm going to leave it alone as long as I can.

[space387]

Steveo I was watching a video on youtube about building a 16bit ecu and saw them using an arduino to record the address and data sent to in and out of the ecu. Because we have access to the memory chip is this the info you need for the bin splitter?

The arduino I have only has enough ports to give you address only but knowing the order they write should allow us to match to the locations used. I'll see what I can do this weekend to replicate what I saw.

[steveo]

ok, I was finally able to get my pcm working again. Not sure if positional or the chip but the last iteration flashed both sides without a single error so I'm going to leave it alone as long as I can.

i feel your pain. i think my socketed ECM behaved best with two folded business card chunks under the chip. it's some real voodoo getting that thing right.

honestly i think someone just needs to bust out an ohmmeter and trace the address lines back to the ECM's processor. we have data sheets for the processor pins and eeprom laying around. descrambling it iteratively proved really annoying but i could try that way again too...

[space387]

i feel your pain. i think my socketed ECM behaved best with two folded business card chunks under the chip. it's some real voodoo getting that thing right.

honestly i think someone just needs to bust out an ohmmeter and trace the address lines back to the ECM's processor. we have data sheets for the processor pins and eeprom laying around. descrambling it iteratively proved really annoying but i could try that way again too...

I'm off this weekend, I'll see if I can get a record of the order each address is accessed during a write and we can match it to the order transmitted to the ecm.

[steveo]

that’d be cool. i have a new ecm in the mail so if that fails i’ll do a hard trace of the address lines to the ecm and see whats going on there. the possible permutations for arranging 16 address lines is totally insane. it’s basically encrypted.