About 2 weeks of my free time at office and home I devoted to evo ecu disassembling. The first goal was to discover load pid and writing patch for "correct" launch control.But by the way i found some interesting. Heres one of the first usefull achievements - found OBD and MUT baudrates. Heres the definition for the most popular 94170008 ecu - sometime I'll make definitions for other ecus.

How to use. If your flash is 94170008 -unzip my definition to ecuflash mitsu folder (dont forget to backup the original) , open your rom and u'll see new params under MISC. Change the BAUDRATE2_MUT and reflash the ecu.
Then start evoscan and put this value in baudrate window (15625 by default)
I tested 62500 baudrate logging - all Ok -engine runs - no errors.
Enjoy highs speed logging

copy sent to openecu.org ,aktivematrix.com

 

very nice baud rate. That is friggen fast!

 

Hopefully you can come out with a patch for 94170014

 

Catch it

 

what kind of samples per second are you guys seeing?

 

awsome thanks bez, im going to try this right now and let you know how it works

 

Flippin' sweet! Great work Bez!

 

Cool. What are the available baud rates?

 

Did I hear you say you think you can log load?

 

The idea is to find out how the ECU calculates the x axis from the maps (load) and spit that out instead of one of the predefined values.

On a seperate note, folks should remember that the higher baudrate will probably be of limited use for things like RPM which are continuous and heavily aliased when read as an 8 bit value.

d

 

1. It have to support any baudrate you enter - the baudrate u'll enter will be converted to the ECUS uart prescaler and then to baudrate again-so u'll see the nearest real supported baudrate. But remember - to set a reliable baudrate record you need to make very short wires line from ecu to obd connector.

2. I know how load is calculated - but I suppose that 2 or 3 load components are translatad trough MUT-u have multiply 2 of them and divide by third for example...
But I want to go away from MUT and try write my own universal datalogging kernel..

3.In some circumstances unpatched ecu can speak 62500 baudrate.
4. U can use higher OBD baudrate patching ECU.



It can fit the other ecus of this family - to try this with your ecu add this lines to your ecu definition after the include <evo7base> ;

<scaling name="BAUDRATE" units="units" toexpr="500000/(x+1)" frexpr="500000/x-1" format="%.0d" min="0" max="500000" inc="1" storagetype="uint8" endian="big"/>
<table name="BAUDRATE1_OBD" address="d59d" type="1D" level="1" scaling="BAUDRATE"/>

<table name="BAUDRATE2_MUT" address="d5d1" type="1D" level="1" scaling="BAUDRATE"/>

<table name="BAUDRATE3_unknown" address="d5e1" type="1D" level="1" scaling="BAUDRATE"/>

If you see after reloading your hex that baudrate3_unknown=62500 and BAUDRATE2_MUT=15625 and BAUDRATE1_OBD=10416 than it will work with your ecu.

 

96420011 ???

 

I can confirm that this works. Evom went down right when I was reading the directions and didn't go back into Ecuflash to switch the MUT numbers to 62500. If you don't do this mostly all of your readings will be off, the most noticable one is the RPM which was pegged at 8k and the rpm of the car will rise to about 2900 or so.

After I switched everything around it works like a charm, the logs become quite long in a very short amount of time. Before I was logging at 92-107 samples/sec. After is just about double that. I think my old POS laptop can't take the mass amount of datat all at once and slows down the transfer rate. Looks like its time for a new laptop.

 

Can someone try this out with a Palm? Im just curious to see if the sample rate will also increase. Since there is a table to switch the OBD baudrate also.

 

I have a feeling the baud rate bottleneck will be on the palm side, not the ecu side.