Hacking/tuning the jeep ecm

[rts91tsi]

Not sure how I could find out. If your not having any luck finding out, I doubt I would have any luck. Would the numbers off my ecm help? Its the factory original ecm, and its a 93 wrangler 4.0 inline six, and a 5 speed manual trans.
Also, pardon my ignorance, but what is DSM?

The link I posted will show you where to look for the microprocessor, then get the numbers/letters off the top of the chip.http://starparts.chrysler.com/starli...n/Electric.pdf

I also found a Snap-On PDF, that has specific ranges for data items that may be useful to building an ADX. http://www1.snapon.com/display/DocMg...px?fileid=4669

Once you find those numbers, if documentation is available for the processor, it should contain a listing of the opcodes necessary for operation and execution of the eprom's bin. Then the real fun begins!
De-solder the EPROM, figure out the offset for the chip being used, read and upload the chips contents to a file on the computer. Depending on how much information you want to find out, there is a program called Binutils, that should be able to help if you want to disassemble the code. http://www.gnu.org/software/binutils/

DSM is an acronym for Diamond Star Motors, the collaboration between Chrysler and Mitsubishi, which is mostly associated with Eagle Talons, Plymouth Lasers, and Mitsubishi Eclipses. These cars were produced at a factory in Normal, IL and technically an American made car.

[EagleMark]

Then if you can figure it out? I read on TGO that a guy is commenting a disassembled code and he is up to 20,000 lines and about half done! He new were most things were. TunerPro software is 40,000 lines of code I read somewhere Mark has into it...

I have over 100 hours into making an adx for LT1 and half the code was written by TunerPro, I just entered parameter information, it's only 2000 lines... and I had the address of where things were.

Cracking GM code was done as a project at Ohio State University way back when by genius type computer geeks who knew microprocessor languages and took a team and years to get it done... luckily all the rest of GM code was similar so we have hacks for most...

I commend you for even thinking about doing this and the conversation is great! I am enjoying it... but even if you had the info you are talking about years of work... but you are getting closer and may find a way!

[hexdmy]

My specialty is calibrating the SBEC2 and JTEC engine controllers, like the one used on your Jeep. A couple of pointers ,

- Chrysler ecus use a special serial communication routine that runs at 7812.5 and 62500 baud .

- TunerPro will work for the 3D tables (yes SBEC2 ecus use 3D tables) and constants, but will not work for the functions, as Chrysler uses a 16 bit slope value between breakpoints.

- You can connect a Moates Ostrich to the ecu and emulate as long as your ecu is not using a latched chip.

[EagleMark]

So it can be done?

I just looked through TunerPro and each value in xdf (bin side) and adx (data acqusistion side) had 8, 16 and 32 bit options. Does that make it doable?

[hexdmy]

TunerPro cannot adjust the functions used in the SBEC ecu. You would need to use software like DCAL .

[1project2many]

De-solder the EPROM, figure out the offset for the chip being used, read and upload the chips contents to a file on the computer.&

Do the DSM ecu's EPROMs contain executable code? Early GM ecm's placed code on a separate PROM while the removeable EPROM contained primarily calibration data which made things much tougher in the beginning.

Cracking GM code was done as a project at Ohio State University way back when by genius type computer geeks who knew microprocessor languages and took a team and years to get it done... luckily all the rest of GM code was similar so we have hacks for most...

I uderstand the point behind this but I want to set this straight. The only connection to OSU was that the DIY_EFI mailing list was hosted there when enough like minded guys were on the list to get the project started. Most of the guys in the project were not computer geeks who knew microprocessor language. And for all intents and purposes, if the effort resembled a team it was only very loosely so. There were a couple of guys who had some really kewl skills and had done exceptional things working out their own disassemblies. There were some engineers and electronics techs and few guys who had experience with network administration. But there were also plenty of guys who were into cars and had a little advanced skill with automotive systems and believed they could get the GM code figured out. We kludged around in the beginning working out data locations and functions for a very limited amount of stuff. Every now and then someone would find a tidbit or a piece of software that gave a good clue. But the real advancements came when "Ecmguy" decided to reward the efforts being made by releasing a few of his hacs to the public. Without that, the entire effort would probably have fizzled. After that, more and more previously unknown data started coming to light, most of it from outside the group that started the project. What's important is this:

You didn't have to be a genius to make tuning a possibility, but you had to be willing to put in effort and it did require some outside help.

If you could learn how an engine worked chances were you could learn enough to work with code.

You had to pay attention to detail and you had to be veeery patient. Many hours can be burned up with very little apparent progress.

There are some quality hacs around from unexpected sources:

An Avionics tech for a DOD facility. This guy told me he made it a point to spend 20 - 30 hrs per week working on disassembly. He had great access to Motorola documentation, much of which described example circuits and code which exactly matched GM circuits and code.

A Surgeon who taught himself Motorola machine opcodes and used previously existing hacs to do a few uncommon masks such as a turbocharged lotus and a ZR1. He found it relaxing.

A dentist and a network admin who spent a bunch of time working out code for the 7749 / $58. I believe they were the first to release a 3 bar calibration for the SyTy world.

An Apple computer guru. He liked Fieros and got curious abuut what was in his car's ecm.

An engineering student who wrote some interesting papers because he was curious about how the ECM code actually worked.

It would be unfair to believe you couldn't do this because it takes some god-like ability or talent. But it does take time, dedication, help, and desire.

/soapbox

[rts91tsi]

Do the DSM ecu's EPROMs contain executable code? Early GM ecm's placed code on a separate PROM while the removeable EPROM contained primarily calibration data which made things much tougher in the beginning.

Yeah, the DSM's EPROM contains executable code.

[JeepsAndGuns]

TunerPro cannot adjust the functions used in the SBEC ecu. You would need to use software like DCAL .

Still learning. What is ment by functions?

[1project2many]

Thats why I am willing to pay someone to do it.

You probably do not make enough to pay someone to do this work. It takes a lot of time.

Still learning. What is ment by functions?

Math. Mopar systems used math equations instead of lookup tables in some areas. You would need to edit the equation instead of tables full of data.

[JeepsAndGuns]

My specialty is calibrating the SBEC2 and JTEC engine controllers, like the one used on your Jeep. A couple of pointers ,

- Chrysler ecus use a special serial communication routine that runs at 7812.5 and 62500 baud .

- TunerPro will work for the 3D tables (yes SBEC2 ecus use 3D tables) and constants, but will not work for the functions, as Chrysler uses a 16 bit slope value between breakpoints.

- You can connect a Moates Ostrich to the ecu and emulate as long as your ecu is not using a latched chip.

So can I use the moates autoprom to read the EPROM? And for that much, if you can use the ostrich to emulate, I'm guessing I could also use the autoprom?
If so, do you know what offsets I need to use to read the chip. I can get the chip out, and probably read it, but as far as actually cracking the code and writing definations is over my head. Thats why I am willing to pay someone to do it.

I am currently visiting family for the holidays, so once I get back I will get the numbers/letters off the microprocessor.

[JeepsAndGuns]

The link I posted will show you where to look for the microprocessor, then get the numbers/letters off the top of the chip.http://starparts.chrysler.com/starli...n/Electric.pdf

Ok, back home. I pulled the top off the spare ecm I have, and it looks a little diffrent on the inside than the one in that link, but I think I got the correct numbers.
The numbers goes as follows:

SC80571VFN
672T02
C96N
ZQESI9312

Also the eprom is a 27C256-20

[rts91tsi]

Ok, back home. I pulled the top off the spare ecm I have, and it looks a little diffrent on the inside than the one in that link, but I think I got the correct numbers.
The numbers goes as follows:

SC80571VFN
672T02
C96N
ZQESI9312

Also the eprom is a 27C256-20

The SC80571VFN is a Motorola processor. I'm not familiar with it, but there is info out there about them.
http://www.seekchip.com/icstock-S/SC80571VFN.html

The C96N is also a Motorola Processor which also has documentation.
http://www.alldatasheet.com/view.jsp...68HC11A0P-C96N

There are also a few tools out there if you are serious about making your own hack. This IDE should work for the C96N, and has colored syntax for opcodes.
http://autos.groups.yahoo.com/group/...u/message/8159

[shelgame]

I do a lot of stuff with the turbo Mopar ECU's. All of the SBEC's and SBECII's use a 68HC11A series processor.

I have many of the Jeep bins already and have kind of started dis-assembling the code. But, I need to get back into it. I had a '95 GC last year that I had intended to do a cal for, but just didn't get back to the code. Too busy with other projects. Anyway, the GC was too rusty for me to use as a project (I know, big shock, a rusty Jeep in Michigan). But, I'm still interested in hacking the code. I'd be happy to help as I find stuff...