two smoked 04 USDM STi ecus so far :(

[autophysics]

I hope all goes well Zuczek.

Did you do a rom read before you attempted the flash? A strange thing happened to me tonight. I always do a full read of the ECU before attempting to flash it. This is to assure me that the connection is good and all is well. Well, as usual I did a preliminary download of the ECU's rom and after doing that successfully I flashed the car - it ran fine, I checked my logs and made a few changes to the maps. I started ecuflash and began the preliminary downloading of the rom from the ECU. It stopped at 30% with a read error of some sort. I tried again and it stopped at 30%. I turned everything off, tried again and it stopped at 30%, each time with the same read error. The car runs fine, no DTC codes appear to be present - and I'll try removing the negative battery cable for a while then trying again tomorrow - but has anyone ever heard of a consistent read failure well into a read?

I've read and flashed the car probably a dozen times so far with this being the first sign of trouble.

I'm extremely afraid to attempt a flash until I can resolve this read failure issue - unless someone thinks that reflashing might actually fix this.

[EDIT] Problem Resolved: After disconnecting the negative battery terminal for an hour I tried again and read successfully and flashed successfully and started ok.

I wonder what would have happened if I had simply attempted a flash without discovering the read error.

[zuczek]

I had the checkrom fix applied. How did you resolve the issue??

Same problem today on my 05 LGT. ER HC on the speedo and Get Kernel Error. Everything flashed OK and I clicked on Compare to ECU to double check. All CRC's were ok. Turned the car off. It won't start now.

What version of cable are you using? 1.2 or 1.3? I have 1.3. Flashed many times before with 1.2 and worked every time. I tried the 1.3 maybe 4 times and now this??

Is it normal for the black connector to separate into 2 pieces when removing it from the OBD plug? Does it need more glue?

I emailed Colby. I'm hope he can provide some support.

did you apply the checksum fix to the rom?
my car did the exact same thing when I flashed a non-checksum fixed rom back to my car.

since this was before the checkbox functionality, I used tinyhexer to manually fix the checksum.

Were you still able to flash/read from the ECU?? Or did you have to do a bench reflash?

[autophysics]

I was using a Tactrix 1.2 and Ecuflash 1.29a when I encountered my download rom problem but it was recoverable by disconnecting the battery negative cable and trying again. It was not corrected by rebooting the computer so I don't think that it's the fault of 1.29a then again has anyone had a problem using 1.28?

[cboles]

There seems to be a couple people having problems like this in the last few days. I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble. Mitsubishi ECUs can generally always be recovered regardless of what you flash into them, but Subaru ECUs must have functional code running to be able to reflash them again.

Colby

[zuczek]

There seems to be a couple people having problems like this in the last few days. I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble. Mitsubishi ECUs can generally always be recovered regardless of what you flash into them, but Subaru ECUs must have functional code running to be able to reflash them again.

Colby

Colby,

How compicated would it be to flash the ECU on the bench like you do it, when there is a problem? Is there a simple programmer that can be assembled/bought for those situations?

[05GarnetLGT]

zuczek:
I am actually working on getting something like that together.
the microcontroller in our ECUs has an industry-standard function called JTAG which can be used to bench flash the image.
unfortunately, I am having trouble finding a cost-effective (read: not $2000+) solution for the SuperH microarchitecture.

[zuczek]

zuczek:
I am actually working on getting something like that together.
the microcontroller in our ECUs has an industry-standard function called JTAG which can be used to bench flash the image.
unfortunately, I am having trouble finding a cost-effective (read: not $2000+) solution for the SuperH microarchitecture.

Isn't there something simple like the JTAG used in the satellite hacking community? I remember back in the days, we built a simple JTAG with some resistors and used a parallal port to flash the chip.

I wonder what Colby is using? Probably the $2000 one...

Man, I wish there was a way to disable the stupid immobilizer. That way, I can have a spare ECU and swap it in an emergency.

[05GarnetLGT]

yeah, same jtag!

btw, to the original poster: still have those smoked 04 sti ecus?
can you open em up and take pics of the pcb?

[cboles]

For the SH based Subaru ECUs I have dealt with so far, it's not so bad. There is a CPU manufacturer boot mode, much like that which is used by the Mitsubishis, where you can communicate via another serial port and bulk erase and reflash the CPU (but not read it, due to the mandatory erase). You basically have to build some custom serial interface hardware and then have some protocol specific software. The software is already in EcuFlash, in fact. I'll try to get around to publishing some schematics of the interface.

There seems to be a couple people having problems like this in the last few days. I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble. Mitsubishi ECUs can generally always be recovered regardless of what you flash into them, but Subaru ECUs must have functional code running to be able to reflash them again.

Colby

Colby,

How compicated would it be to flash the ECU on the bench like you do it, when there is a problem? Is there a simple programmer that can be assembled/bought for those situations?

[zuczek]

Thanks Colby,

It would be very helpful if you could do a small writeup with the schematics. It would probably help alot of people out there that got into the same situation as I did.

BTW thanks for looking at my ECU. Should get there today.

For the SH based Subaru ECUs I have dealt with so far, it's not so bad. There is a CPU manufacturer boot mode, much like that which is used by the Mitsubishis, where you can communicate via another serial port and bulk erase and reflash the CPU (but not read it, due to the mandatory erase). You basically have to build some custom serial interface hardware and then have some protocol specific software. The software is already in EcuFlash, in fact. I'll try to get around to publishing some schematics of the interface.

There seems to be a couple people having problems like this in the last few days. I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble. Mitsubishi ECUs can generally always be recovered regardless of what you flash into them, but Subaru ECUs must have functional code running to be able to reflash them again.

Colby

Colby,

How compicated would it be to flash the ECU on the bench like you do it, when there is a problem? Is there a simple programmer that can be assembled/bought for those situations?

[05GarnetLGT]

For the SH based Subaru ECUs I have dealt with so far, it's not so bad. There is a CPU manufacturer boot mode, much like that which is used by the Mitsubishis, where you can communicate via another serial port and bulk erase and reflash the CPU (but not read it, due to the mandatory erase). You basically have to build some custom serial interface hardware and then have some protocol specific software. The software is already in EcuFlash, in fact. I'll try to get around to publishing some schematics of the interface.

There seems to be a couple people having problems like this in the last few days. I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble. Mitsubishi ECUs can generally always be recovered regardless of what you flash into them, but Subaru ECUs must have functional code running to be able to reflash them again.

Colby

Colby,

How compicated would it be to flash the ECU on the bench like you do it, when there is a problem? Is there a simple programmer that can be assembled/bought for those situations?

ok, theres half of the battle.
but my question if you do that: does the immobilizer interfere?
right now you cannot "clone" an 05+ ecu, or even fully back one up.
I would really like to do this so I have a spare incase something happens, and I am sure others would be interested in this too.

[zuczek]

Colby said that the Immobilizer isn't part of the ROM so once flashed with the alternate method. the immobilizer will still be programmed to the car.

For the SH based Subaru ECUs I have dealt with so far, it's not so bad. There is a CPU manufacturer boot mode, much like that which is used by the Mitsubishis, where you can communicate via another serial port and bulk erase and reflash the CPU (but not read it, due to the mandatory erase). You basically have to build some custom serial interface hardware and then have some protocol specific software. The software is already in EcuFlash, in fact. I'll try to get around to publishing some schematics of the interface.

There seems to be a couple people having problems like this in the last few days. I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble. Mitsubishi ECUs can generally always be recovered regardless of what you flash into them, but Subaru ECUs must have functional code running to be able to reflash them again.

Colby

Colby,

How compicated would it be to flash the ECU on the bench like you do it, when there is a problem? Is there a simple programmer that can be assembled/bought for those situations?

ok, theres half of the battle.
but my question if you do that: does the immobilizer interfere?
right now you cannot "clone" an 05+ ecu, or even fully back one up.
I would really like to do this so I have a spare incase something happens, and I am sure others would be interested in this too.

[bradweingartner]

I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble.
Colby

This is KEY and needs to be highlighted accross the boards. Many people are headed for trouble by downloading random ROMs and burning them. I personally would rather analyze the downloaded roms and apply those parameters to my original rom. That way I'm 100% sure that I'm not flashing an incompatible rom.

[bmorrisj]

so colby... whats the word???

[chuckdez]

I want to emphasize that the only way to be safe when reflashing is to read, modify, and write your own ROM image from / to your ECU. If you are downloading ROMs from this and other sites that may be different than you original ROM, you may be headed for trouble.
Colby

This is KEY and needs to be highlighted accross the boards. Many people are headed for trouble by downloading random ROMs and burning them. I personally would rather analyze the downloaded roms and apply those parameters to my original rom. That way I'm 100% sure that I'm not flashing an incompatible rom.

The open community should agree on a standard difference file. This would allow a tuner to export only the changes from his rom tables. Then the user can import the difference file and apply changes only to his tables within his rom. Not only would this eleminate all of the large rom dumps on the boards, it would prevent alot of these problems.