jcsbanks

96260009.

I think it may indeed be free. I have flashed an 8 image into my 9 ECU, and whilst it won't start the engine I can test the comms, DMA, RAM etc. I'm finding that the fuel map is being erased on key off, but the ignition map is not. There is a comms issue too that is inconsistent.

jcsbanks

ffff9c00-9ffff is erased on key off, this is where my fuel maps were, I think ffffa000-a7fff is OK - leaves 768 bytes for the stack to come down and gives us 2k for alt maps.

Still leaves an intermittent comms problem though, whereas the 9s I've played with have been fine.

jcsbanks

I've tested FFFFA000-A7FF and it seems safe. Comms problem seems to be resolved too by this move. I've posted an 8MR JDM version on MLR to test.

I also tested a button in my app that resets Tephra's DEAD variable, forcing a copy from ROM to RAM of my maps, should remove the need for middle ECU plug disconnects to reset, I suppose we could later add a more complete reset, but let's see how this goes first.

Appauldd

iTrader: (22)

Let me know when you are ready for a tester.....I will give it a shot.

Paul

jcsbanks

Thanks Paul, need to sort out if 96260009 is feasible or not, if it is, transfers to others should follow, although I would like to know if there is just one US Evo 8 ECU I can do to keep most people happy.

jcsbanks

96260009 is working. After the first flash from a non-Tephra ROM it copies the maps from ROM to RAM because Tephra's DEAD variable is not set. However, subsequent flashes need a middle plug removal reset because Tephra's DEAD variable is remembered between flashes as this area is not overwritten, but the area of the RAM maps is overwritten by the flashing process. I suppose I could put markers at the beginning and end of the RAM area, or just use a button in my app (tested on my IX) that just writes a non-DEAD value to Tephra's variable, forcing everything to be refreshed from ROM.

tephra

iTrader: (6)

hey John.

Since my RAM block is ok (for this rom) you could change the DEAD check to a memory location in your RAM block.

just update the pointer in my code so that it checks and resets accordingly...

cheers
d.

jcsbanks

Good idea. I suspect it is the 9s that overwrite the FFFF84xx area during a reflash, but the 8s don't.

tephra

iTrader: (6)

yeah 8 and 9's obviously use different RAM segments to store incoming ROM segments.

might be a kernel/driver thing in ecuflash...

jcsbanks

Maybe that the 8 RAM starts at FFFF8000, so your 84xx range is quite low in there, whereas the 9 RAM starting at FFFF6000 means that the 84xx range is in the middle of the stuff used by the reflash kernex, hence it gets overwitten.

tephra

iTrader: (6)

yeah I *thought* it said which RAM block it used in the ecuflash log... might be wrong thou

roger smith

iTrader: (4)

Hi John.

I think for US evo 8 the smallest number of roms that will work for the majority is 2. Someone correct me if I'm wrong.
The 94170015 and 96940011.

And just wanted to sidenote, your efforts are greatly appreciated, as are tephra and MrFred. When I come into enough money to be philanthropic I will gladly do/get something nice for you guys (ball bearing upgrade for your turbo, carbon doors, you name it). And I won't forget about this pledge as long as this community is alive.

jcsbanks

I was thinking of just going 96530006 since that seems to run all US Evo 8?

So the bare minimum set would be:
88570008 JDM Evo 9, probably runs all Evo 9, I suppose I could do 88590015 since I have already done an earlier version...
96260009 JDM Evo 8MR, also runs JDM 8, Evo 260
96530006 Probably runs all US Evo 8
90550011 JDM Evo 7 - seems to run all Evo 7, plus Evo 7 ECUs running Evo 4,5,6

Any more is really a few too many to support as my brain will melt. The patches are complex.

l2r99gst

iTrader: (2)

96530006 sounds good to me. I'm still waiting to switch to that ROM, but I need a reason, like SD or your live tuning first.

Appauldd

iTrader: (22)

Get me a copy of the 96530006. I will load it up and test its functionality against the 94170015