I'm making some good progress, when I have free time.
The Hybrid protocol for 95-96 is ugly and unnecessary complicated and has slowed things down a little. The Hybrid protocol is wrapped up inside the OBD2 and Emissions junk. The protocol is what they call "Designed by Committee", they make things complicated so you have to buy their datasheets and manuals.
I could not control the inter-byte and inter-message timing that OBD2/Hybrid is picky about with the .NET VCP driver. So I have completely switched to the FTDI D2XX DLL. This allows a lot more things to be done and should be a little faster. This also allows more than one FTDI cable to be plugged in, with VCP having more than one cable screwed things up.
I also got some info from Innovate Motorsports about their WideBand's protocol. I will probably add support for their WideBands in the future.
Also I have found a source for some dirt cheap cables, but I want to test them out a little more.
Progress Report:
Transitioning program from Virtual Com Port to FTDI Driver.
97-99 DSM is done.
95-96 DSM Init Sequence is done, as well as some simple requests.
I have about 8 to 10 hours of work before my next release which will support 95-99 DSMs.
I would like to add support for 1Gs, if anyone has a ECU and harness plugs they would like to donate it would help out a lot.
Also IRL today, I stripped three exhaust manifold studs. If anyone in the Reading, PA area wants to help a brother out, and has a right-angle drill I could borrow for a day, that would be awesome.
what kind of info do you need for 1G?
works the ping pong way, (2G seems to be nearly the same?)
basic adresses and formulas are shown here:
The 1G stuff is very well documented. But you shouldn't trust everything you read 100%
It helps having an actual ecu, to test my own code for mistakes.
And to try to speed up the protocol by decreasing the delays between messages.
Do you think you could make that dynamic? Or maybe have the user enter in the delay as an option? I've heard that some ECUs can read faster from one car to the next. I'd like to push it as fast as it will go until I start getting I/O errors or jumbled data.
PM me when you need it tested on the '97 again. And the wideband support would be nice. I have plenty of time in the evenings to test it.
Any idea when your next update will be out?
____________________________
Wes - '97 Talon Tsi AWD
Man this is bad ass! I just bought a pocketlogger not too long ago but its kind of a pain trying to tune on a PDA, and most importantly no knock logging!
Is that blue cable on ebay thats linked the one that works? If so Ill order one asap! And just to be sure, my pocketlogger cable wont work?
You are the man! Are you taking donations? You should!
If you log less items, you will have more resolution per item.
You should get at least 15 samples a second.
Log RPM, AirFlow and KnockSum only and they will update 5 times a sec.
WideBand is logged separate from other items.
RPM, AirFlow, KnockSum and WideBand is all you should need for WOT tuning.
Your only options for a faster sample rate are a ecu swap or a complete ems.
Also if you are using EvoScan to log KnockSum on a 95-97 the RequestId should be 3E not 26.
KnockSums address is different between 95-97 and 98-99.
Address 26 will always read zero on a 95-97.
I have a 98 with a 95 ECU so what does my requestID need to be? I did a pull and it showed one count of Knock (without me changing anything).
Would it be possible to just get the logging code from a 1g on to a 95 EPROM? I love the logging on a 1g; the TMO Logger is great.
How about otherwise the logging from an Evo 8? Maybe Tephra or MrFred on EvolutionM would know how the code works well enough to be a help on porting it over to a 95.
Would it be possible to just get the logging code from a 1g on to a 95 EPROM? I love the logging on a 1g; the TMO Logger is great.
How about otherwise the logging from an Evo 8? Maybe Tephra or MrFred on EvolutionM would know how the code works well enough to be a help on porting it over to a 95.
The problem swapping the Evo code is the newer ECUs use the H8 processors, which have a totally different Assembly Language.
A 1G logger swap is possible, but thats a few hundred lines of code.
I have the 95-96 protocol done, just have to finish a few things, then I'll put out my next release. Mitsu wrapped the 95-96 Hybrid protocol inside OBD2 so it is slow like OBD2.
Jeff from DSMChips said he might do a feature thats speeds up the protocol, problem is the Emissions/OBD2 part of the code is so cryptic the top DSM code guys have trouble figureing out what Mitsu was trying to do.
If you log less items, you will have more resolution per item.
You should get at least 15 samples a second.
Log RPM, AirFlow and KnockSum only and they will update 5 times a sec.
WideBand is logged separate from other items.
RPM, AirFlow, KnockSum and WideBand is all you should need for WOT tuning.
Your only options for a faster sample rate are a ecu swap or a complete ems.
Also if you are using EvoScan to log KnockSum on a 95-97 the RequestId should be 3E not 26.
KnockSums address is different between 95-97 and 98-99.
Address 26 will always read zero on a 95-97.
I'm going to have to try this out. I haven't changed anything in Evoscan (both v2.5 and v2.6 beta 7) and I know I don't get a continuous reading on 0 on my 97. It fluctuates betwee 0 and 1 under load. And judging by my timing curve, I'm not getting any timing retard due to knock.
____________________________
Wes - '97 Talon Tsi AWD
I'm going to have to try this out. I haven't changed anything in Evoscan (both v2.5 and v2.6 beta 7) and I know I don't get a continuous reading on 0 on my 97. It fluctuates betwee 0 and 1 under load. And judging by my timing curve, I'm not getting any timing retard due to knock.
I would try logging both 3E and 26. I'm not really sure about the 97s they are kinda in between the two.
Be safe, I don't want to mislead anyone. None of this info is documented by Mitsubishi.
All I know for sure is that 3E is KnockSum on the 95s, because I have the code disassembly for the eprom 95s.
subed, dang fine work here Ceddy, and from PA to boot. Might save me till i get the funding to grab DSMlink V3. I'll wait till the 95's are done, keep up the good work.
____________________________
-Chuck
~2gb'ed Talon Tsi Awd A/T
Have you checked out what that Brazilian BIN was like that had no OBDII? It must have some type of communications. I tried the Brazilian BIN, and my car seemed super rich with the same maps as my normal EB23C/E BIN tunes, but since I couldn't log OBDII I couldn't even see what the trims or anything were like.
Have you checked out what that Brazilian BIN was like that had no OBDII? It must have some type of communications. I tried the Brazilian BIN, and my car seemed super rich with the same maps as my normal EB23C/E BIN tunes, but since I couldn't log OBDII I couldn't even see what the trims or anything were like.
No, I haven't played with that yet. I only have one eprom ecu and its in my car.
I suspect it uses MUT like the 97-99s. Thats what the euro EVO 4 - 7 use.
Your probably running rich because the Brazilians put a lot of ethanol in there gas back then. 25% I think. Not sure if Mitsu compensated for that or not.
No, I haven't played with that yet. I only have one eprom ecu and its in my car.
I suspect it uses MUT like the 97-99s. Thats what the euro EVO 4 - 7 use.
Your probably running rich because the Brazilians put a lot of ethanol in there gas back then. 25% I think. Not sure if Mitsu compensated for that or not.
I'll probably give evoscan a try with the Brazilian code then and I'll post how/if it works.
I got my $13.88 cable today (from China on ebay)! I haven't tried it on my Eclipse yet, but I did try it on my 00 Galant 2.4 and it worked great with Evoscan -- lots of data and it was running around 100 samples per second. I haven't opened the cable to see how it looks inside, but on the outside it looks a heck of a lot nicer than the Evoscan (openport 1.3u) cable (hopefully looks nicer on the inside too).
Do you use a palm for this software while in the car or its laptop only?
Laptop. I highly doubt he'll work on a PDA version, especially palm. The only reason I ever got a Palm was because at the time laptops were still so expensive, but now decent laptop is 500 bucks, or you can get a nice little netbook for 300. I hope to never use my Palm again.
I got my $13.88 cable today (from China on ebay)! I haven't tried it on my Eclipse yet, but I did try it on my 00 Galant 2.4 and it worked great with Evoscan -- lots of data and it was running around 100 samples per second. I haven't opened the cable to see how it looks inside, but on the outside it looks a heck of a lot nicer than the Evoscan (openport 1.3u) cable (hopefully looks nicer on the inside too).
I tried the cable on my Eclipse today. It was running around 10 samples per second in Evoscan on Hybrid mode. I tried the EC95 Brazillian code -- I couldn't get a connection with Mitsu or Hybrid in Evoscan. I also unsuccessfully tried Mitsulogger, MHI Scan, and MUT Monitor.
I tried the cable on my Eclipse today. It was running around 10 samples per second in Evoscan on Hybrid mode. I tried the EC95 Brazillian code -- I couldn't get a connection with Mitsu or Hybrid in Evoscan. I also unsuccessfully tried Mitsulogger, MHI Scan, and MUT Monitor.
Thats disappointing the Brazilian code didn't work.
I would really like to get my hands on a dealership MUT2 or MUT3 scantool and see how it communicates.
