dimented24x7

iTrader: (2)

For a while Ive been looking to convert my ecm, which will be an 8746, to be romless so I can dabble with the source code. There doesnt seem to be much info on how to do this, though. I did find one site awhile ago that goes into some detail about how to get the rom to transmit the rom contents as serial data. Im sure some of you have seen it.

Link

Has anyone tried this yet? Looks like something I could do with a possible chance of success. How is the replacement eprom image used, though? Do I just open the ASCII as binary and burn it to a chip? Looks like it would be rather large to fit on a chip.

As for bypassing the rom and replacing it with an eeprom, well, there im really lost. I havnt come across anything yet. Does anyone ahve any info on how to go about going romless with a C3 ecm? Im surprised there hasnt been more talk about it on the boards given its importance.

RBob

iTrader: (1)

I thnk that reason there isn't more chatter on going ROMLess is that not too many people dabble in changing the source code.

Unless you want to change the code to add or change functionality there isn't a reason to go ROMLess. Before I continue, it may be easier and better to change to the '749 ECM with the $58 mask. You'll have a better ALDL stream, more RAM space, faster processor, everything already in EPROM (no ROM). . .

Back to the '8746, yes, the link'd method to gettig the RON dump does work (I've done it). The ROM contents are sent in an S19 format. As the data is sent over-and-over it, has 'cut here' headers and all. Just take the file trim the excess and use an s19 to bin utility to convert to binary.

To do it, burn the supplied image into an EPROM and place it into the ECM. Power it up and the data is sent.

To disable the on board ROM it is a matter of grounding a pin on the ROM chip(s). The site you linked has schematics for that ECM. Get that also.

More to come. . .

RBob.

RBob

iTrader: (1)

As posted to the diy-efi list on Dec 7 2000:

(This is from memory as I do not have
my notes here). If you can refer to the '746 schematics, it will help in following this.

The onboard ROM is disabled by tying the ROMEN line low. A wire on one of the RomI/O chips does this (U7-1 to U7-22).

Locate the 74ls138 decoder chip (U8). This is the one that has the chip select (CS) lines to the eprom & adc.

Move the ADC address from 0xC000 to 0xA000, this is a single cut and jumper (pin 11 to pin 13). This will also need to be changed in the code at three locations (no big deal, we have access to it now!).

Now the address block from 0xC000 through 0xFFFF inclusive is open. In this area a 128kb eprom can be used. The tricky part is creating the proper CS signal. I used a 74ls00 piggy-backed on top of the 74ls138 decoder already on board. Note that the eprom CS is at 0xD000 of that decoder. Wire address lines A14 & A15 to the 74ls00, use the output of the 74ls00 as the new eprom CS. All these signals are available at the above
mentioned decoder.

Basically, the eprom is selected whenever address lines a14 & a15 are both active (high).

Now expand the 24 pin eprom socket to 28 pins. I did this by putting a 24 pin socket in the board, then a 28 pin socket into that one. This allows access to the extra pins. Pin 24 of the orig eprom needs to be accessed also. This is vcc and requires being moved to pin 28 of the new eprom. (Cut pin 24 out of the 24pin socket before soldering to board).

Two additional address lines added to the new eprom and the hardware is done. These address lines are available locally as they go to the service port.

Now the area from 0xC000 to 0xCFFF is all new eprom space, with the original eprom code at 0xD000 and the original rom code occupying the eprom from 0xE000-0xFFFF.

Use RTV to glue extra chips and wires in place for reliability.

dimented24x7

iTrader: (2)

RBob, what else has to be wired to the secondary decoder chip besides address lines 14 and 15? On the schematics that I have, the ROMEN is on chip U8 as pin 22. Pin 22 on U7 is listed as 'D5'. Is this still the correct one to jumper to ground? I also heard of someone using a 24 pin 2732 chip in place of the rom. I also hear that the rom can be read 'destrictivly'. Is this basically jsut saying that the rom can be removed to be read? Im jsut curious as to what other methods have been used to bypass it.

RBob

iTrader: (1)

What else has to be wired to the secondary decoder chip besides address lines 14 and 15?
Pin 14 goes to VCC (+5V) and pin 7 goes to ground. I usually glue the added '00 on top of the current '138 decoder chip.

On the schematics that I have, the ROMEN is on chip U8 as pin 22. Pin 22 on U7 is listed as 'D5'.
Chips U6 & U7 contain the ROM. Pin 22 on both is ROMEN, grounding either chips pin 22 is all that is required.

I also heard of someone using a 24 pin 2732 chip in place of the rom.
This will work. Each of U6 & U7 contain 4kB of space, same as a 2732 EPROM. In total 3 2732's would be required, one for each ROM and then the main one.

I also hear that the rom can be read 'destructively'. Is this basically just saying that the rom can be removed to be read?
Chips U6 & U7 can be removed and read, not as easy as the ROM dump via IAC port method. Plus it will most likely render the ECM dead as it will be difficult to re-install them after being removed. Forty pin DIP chips along with the conformal coating, blah. These chips are required for other ECM functionality, so they can't just be removed and left out.

RBob.

dimented24x7

iTrader: (2)

Ok... I was jsut curious as to how the other methods worked. I hooked the ecm up to an old atx power supply and downloaded teh contents of the roms. Now I jsut have to get another '8746 that actually works.

dimented24x7

iTrader: (2)

Ok, last question... Im not an expert in electronics by any means so bear with me if you could. As for wiring the whole thing up, I ground pin 22 on U7, cut pin 13 on U8 and wire pin 11 of U8 to pin 13's location on the board. On the eeprom, address pins A0-A11 are connected as they would be for the 24 pin prom. Pin A12 on the eeprom is wired to pin 13 on U8. Then pins 14 and 15 of U8 are connected to the inputs of the 74LS00 and the output of the gate in use is hooked up to pin A13 on the eeprom to give access to the last half of the chip. Is this the correct way to hook it all up? Just want to make sure since Im not too terribly knowledgable when it comes to all of this.

dimented24x7

iTrader: (2)

ok... Now I know why I havnt been following. There are errors in the schematics set for the 8746/7747. Here are the schematics.

dimented24x7

iTrader: (2)

Here are the pinouts for the 74LS138.

dimented24x7

iTrader: (2)

From the above schematic, I take it the chip marked U12 is really U8. Is the chip marked U8 really U7 instead? That would make more sense since, from the schematics, 'U8' is the one that contains the ROMEN on pin 22. From this, I take it that the procedure for bypassing the rom would be the following:

Ground pin 22 on U7

Cut pin 13 on U8 and wire pin 11 to 13's location on the board

Wire pins 3 & 6 (a14 & a15) of U8 up to the 74ls00 and wire pin 1 of U8 up to A12 on the eeprom to gain access to the 32 k of space for C000-CFFF

Wire the output of the gate to a13 on the eeprom to get it to select the last half of the 128 k of memory.

Finally, wire up the eeprom pins A0-A11 as they would be for the standard 2732 eprom.

Is all this correct?

RBob

iTrader: (1)

dimented24x7, the one you posted in the incorrect schematic. Looks like Ludis has some mis-linked info. To get the correct one, once you are at the schematic page (where they list the links to the pics themselves) scroll down until the header has 1228746, then grab those pics.

This would explain why nothing made sense. The group will look something like this:
RBob.

dimented24x7

iTrader: (2)

Yes, thank you. The previous ones didnt make any sense at all. As soon as I get another 8746 that actually works Ill give it a try and see what happens.

RBob

iTrader: (1)

If you are going to go ROMLess then any of the '7747, '8746 or '8063 ECMs will work. They are all the exact same hardware, just have different ROM images in them. Once the ROM is disabled, they are all the same.

The output of the 'LS00 goes to the ~CS input of the new EPROM (forget which pin). Then the address lines A12 & A13 of the ECM get wired to A12 & A13 of the new EPROM.

I think the rest you posted is correct.

RBob.

dimented24x7

iTrader: (2)

Really? So basically, say, if I want to use my '8063 I can just use the entire source code for the 8746 and itll work just like an 8746? Huh...

I have a partially dead 8746 that someone sold me and I cant get my money back for it so ill experiment with that.

dimented24x7

iTrader: (2)

I wired up the 8746 I had and it worked as advertised. Thanks for the info. RBob, I sent you a PM. Id definatly be interested.

Heres what the end result looks like, The adapter plugs into the 'RN' socket and uses a 28C256 so bin switching is a possibility. Looks kinda ratty but since ill never use this ecm in the car it doesnt really matter.

RBob

iTrader: (1)

Well I'll be. . . You are one of very few that have done this. Heh-heh, have any idea of where in the code you will start first? Seriously consider building a test bench next. Congrat's.

RBob.

dimented24x7

iTrader: (2)

Hmmm... One of the things Ive always wanted to do was control a 4 barrel tbi with an ecm and have it run as a 2 barrel at low loads and have all 4 barrels (and injectors) operating at high loads so I can have the best of both worlds. Granted the 4 barrel units are mad expensive, and I could have the same benifits with an aftermarket ecu, but I still want to try it anyway. Im not real knowledgable with how the ecm's source code operates as a whole yet so I wont be doing much in the enar future but hopefully ill get the hang of how it all fits together so I can make modifications as I please. And yes, I do plan on a test bench. Wouldnt want to try anything crazy in a full sized car when im not entirly sure of what effects it may have. Anyway, thanks for the help

RBob

iTrader: (1)

ttt

RBob.

Grumpy

Run them all full time.
With the right code you can run huge injectors and mantain proper manners. I have 6, 60 PPH injectors on my lil v6.
Any engine that really's going to use a 4 brl TBI to it's full potential isn't going to be an econo car.

Ken73

I've been looking for a thread like this.

RBob, could you not setup address decoding so you wouldn't even have to move the ADC around?

What other ECM's besides the '8063, '7747, and '8746 share the same hardware..? Seems like there might be a few more..?

RBob

iTrader: (1)

It just works out cleaner by moving the ADC address. If it dosen't get moved then the address decode for the EPROM gets more involved. But the kicker for me was when I looked at how the conversion of the S19 file to a binary file got complicated. The first 4KB block would be from $A000 through $AFFF, then the 2nd-4th 4KB blocks would be from $D000 though $FFFF.

Now the S19 conversion utility needs to slice & dice to get the BIN parts in the correct locations. Besides, all I do is to swap out the little board and do the one cut & jumper (a short jumper at that), and the ECM is converted.

RBob.

JPrevost

The 8747 is the same c3 TBI ecm.

Teeleton

Looking at the photo of the ECM posted, and the guts of my 8062 ECM, it looks like the same board layout. Can anyone verify if it's the same? The PCB has the #36E858 molded into it. Can anyone confirm if this is the same board? I'd love to convert this over to a 7747, without having to go source one.

I've managed to figure out a lot of the 8062 ECU tables, but there are still a lot of constants that I can't locate or don't properly understand.

If I could swap over to the 7747 code, or better yet, the Ultimate TBI code, I think my task would be quite a bit easier.

As a reference, I currently have a 4.3L V6 running 10:1, ported 96+ heads, 222/230@.050 cam and a holley TBI with 85pph injectors. I've gotten it to idle properly, but many of the tables in the 7747 code that I would like to use to further tune it, I haven't be able to locate in the 8062.

Eventually, I plan on swapping to Ken's speedreader setup once it's ready, but since I've finished my lastest build on my engine, I'm itching to get it on the road, and the 8062 cal seems to be turing into a handicap.

Teeleton

RBob

iTrader: (1)

Good possibility it is the same ECM (except on board ROM). One '747 I have here has the code 36B858 on the PCB. The B/E difference could be anything from a date code to a rev level.

RBob.

JPrevost

Teeleton, the code is likely different. Code being what's located in the 8k of rom. The 4k being the eprom with very little code and mostly lookups.
I'd call up the local j-yard for a 7747. When the speed reader comes out you can use it on your current ecm to run 7747 code. This way you'll have 1 ecm that's an spiffy and the 7747 you pick-up can be the cheap backup (reason for keeping GM electronics).
Going source code is the only way to run any code you want. When I say "any code" I mean ANY. Just like the 7730 and 7749 being about to run completely different code, all code is on the eprom with those ecms.
The catch 22 is that there are only a few ecms that have had the code downloaded from the rom. The 7747 and the 8746 being the only 2 I can remember ever seeing.

Teeleton

That was the goal, to run romless. Seems to me I could just take my 8062, make the necessary changes to run romless, and then fire up the 7747 code & cal in a larger eprom. I found what I did in the 8062 by simply starting with a 7747 ecu & bin, and then taking an image of my chip, and hunting for similar tables in the hex. Later, I found a commented disassembly of the 8062, and verified things I had already found plus found a bunch of new stuff. The thing is that I still have a lot of the tables from the original 7747 ecu in the 8062 ecu marked with question marks, so as I'm tuning I'm constantly remind of tables I could be tweaking, that I can't and it kinda grates on me.

The ALDL stream for the 8062 is exacltly the same as the 7747 according to WinALDL (they both use the same settings & tables).

Eventually, I'd like to get one of the test benches on a board setups I saw elsewhere in the forum, and possibly even a NB-o2 meter also in the forum. I've thought about swapping to a 7747, but I don't really have any convenient bone yards where I am. It's about an hour and half round trip to the nearest yards.

Teeleton

Teeleton

Well after thinking about it for another couple minutes, I decided to dash out to the bone yard to see if I could find any 7747s. I got there about 15 min. before they closed, and it's a good thing, because that exactly how long it took to locate and liberate two of them.

So I guess I'll just transfer some maps to one of them to get the blazer going again, and then start work converting the other run romless and putting a test bench together so I can start playing. I haven't pulled the ECMs apart yet, but you can bet I will and attempt to answer my own question.

Thanks guys,
Teeleton

xplane

anybody have these kind of schematics for the ecm from a 85 2.8L v6?

JPrevost

Sorry I don't, look on Ludis's website for similar ecms.

xplane

Um what website is that?

JPrevost

http://www.cruzers.com/~ludis/

xplane

ok thanks so its the fan chip that keeps exploding i wonder why

Teeleton

1228062 is close but no cigar. The main & daughterboards have the same revision #'s molded in, but U7, U8, and U15 are different part #'s. I tried putting the 1227747 mempak & prom in the 1228062 anyway, but it wouldn't give me any data on the ALDL. Oh well. The 1227747 plugged right into the 1228062 harness with no modification and gave me good data on the ALDL when I keyed it up, so everything looks good at this point. U15 is burnt in one of the two ECMs I picked up. Guess it's toast. Gonna have to pick up another one before I can start toying with a romless one.

Teeleton

dimented24x7

iTrader: (2)

Been a while since I did it but any C3 ecm can be downloaded with ludis' program. Basically what you do is snare teh computer with BRA commands when it goes to run the coded portion of the prom. The interrupts are disabled and the computer is then directed toward the program that transmits the code as .S19. Its pretty small and fits in the main SA table. This is how I got teh stuff off my 8063. I disassembled the rom xmit and planted it inside my prom. Hook up the IAC output to the serial port and it started transmitting the code.

Another bonus to having an oddball ecm is that its a great opertunity to learn assmembly. I learned assmembly by commenting my 8063 code. It was different enough to prod me to think a bit and actually get some insight into how it all worked.

dimented24x7

iTrader: (2)

Its probably teh same hardware. Theres certainly one way to find out . Although I probably could have turned some profit by selling my 8063 and using something more common rather then converting it over since they didnt make that many of them.

RBob

iTrader: (1)

Check the fan relay, the coil is probably shorted. Questions/issues such as this should probably have been in a new thread.

RBob.

xplane

oh i did start a new one. i just put that in for no reason after i found the schematics