Slo_crx1

http://www.openecu.org/index.php?title=EcuFlash

After the urging of several members, I figured it wouldn't hurt to throw this up here as another alternative to newer obd2 tuning. So far EcuFlash supports most '01-'06/'07 Mitsubishi's and Subaru's (listed here at the OpenEcu main site... http://www.openecu.org/index.php?tit...VehicleSupport) and the list seems to be growing. So far there is nothing supported for Honda's K-series, but I'm sure it won't be too far off if the interest is high enough. Enough gabbing for now, on to the program!

This is the heart of the EcuFlash program. When you open a rom image there are tables listed on the left hand side that contain different vehicle options you're allowed to tweak, as long as you know what you're doing lol. Note this rom image is from my '03 Evo 8, and is an "in progress" tune. So far fueling is great and no knocking under boost with stupid amounts of advance, but copy these values at your own risk.



This particular picture shows the high octane fuel and timing maps. There are 2 sets for each, high octane and low octane. The ecu can determine through knock counts which table it should run from and during any point of knock the tables are interpolated. This can make things a hassle for initial tuning, so I usually set low and high octane maps the same as each other until the final tune. Once the final tune is reached, then I'll kick in a little spark retard under higher loads and extra fuel as well to the low octane maps, just as a "just in case" preventative. One thing you'll notice is that the fuel maps are listed in afr's instead of milliseconds that the injectors are open. Keep in mind that these aren't "actual" afr's, since the stock ecu still runs with dual narrowband o2 sensors. On my fuel map where it lists a value of "9.3 afr", my value is actually closer to around 11.3~ according to my lc-1 wideband.

Another thing to look at is load percentage at the top of the tables...these cars are MAF based and everything goes off of loads instead of manifold pressure like Honda's speed density setups. These load points are in no way correlated to boost pressure, but more or less refer to peak torque production. Peak torque is where the highest loading occurs, which in this case is the 260% column on the tables. Evo 9's have load% columns that reach into the 300's; doesn't mean they make more torque, just that the tables are tailored to the map differently.

Here's a few more tables...Maf scaling, injector size, and injector latency, and my knock cel patch...



The Maf scaling table helps to optimize the airflow readings when an after-market intake is used on the vehicle, or a larger turbo than stock is used. This table can really screw your car and your tune up if you don't understand how to change it. Datalogging is key to adjusting this table, and datalogging the flow of the stock setup (in g/s or converted to hz) compared to the flow of the aftermarket setup will usually tell you where the Maf table needs to be adjusted. As with any setup, datalogging is key to tuning these...just quite a bit more on this setup than the standard Honda speed density setup. FYI, the stock ecu outputs only 1 byte of info on maf readings (as well as load readings), and the cut off frequency is 1600hz. In order to log above that value there is a map hack called the "2 byte" hack with allocates an un-used or non-essential portion of the MUTIII table to log the higher values. I'll get into more on that later though.
The other 2 tables that are showing are Injector latency and Injector size. Stock Evo injectors are not 513cc like shown, but closer to 560cc's or a hair higher. It's easier to control the latency with the injector "downsized" instead of running it at it's real size. These features come in quite handy when upgrading to larger injector sets, and most people with good tuning knowledge can still get 1000cc injectors to idle like stock when setting these tables properly. AIT compensation is pretty much self explanatory for anyone who's ever used Crome or any other tuning software. The factory settings are usually pretty good, so most tend to leave them alone.



Here is the Maf smoothing table and the tables that allow for ecu-based boost control. You'll notice there are 4 boost tables listed in the left hand column...1,3,&4 are usually the ones the ecu turns to if it detects any type of vehicle/engine problems or for warm-up periods. Table 2 is the one that is adjusted to whatever settings you want for boost. I won't get into detail on them too much, for the beginner the fuel/timing tables are the most important. I have these tables setup on my car to run a solid 23psi all the way to redline without any of the typical Evo boost drop-off. One table that is worthwhile to cover is the boost limit table. This is the factory ecu's way of limiting over-boost, but unfortunately does it in a very violent manner via fuel cut. Anyone who's ever put a decent amount of bolt-ons on an Evo has hit this, and it feels like you've just hit a brick wall at top speed. The ecu will allow overboost for a certain period (there is a table that lists the period of time in milliseconds, normally set for 1000, or 1 second) until it cuts fuel. This table should be raised in order to prevent hitting fuel cut (all numbers shown are load%).
The boost enhancement table (or anti-lag) tells the ecu at what rpm to start the anti-lag process. This is basically where the engine will run a slight bit leaner than normal while getting into boost in order to spool the turbo faster.



Just figured I'd throw this one in there for you math heads. The scaling manager will allow you to add math options to the tables, such as changing the Maf flow from g/s to hz. This is also where to add definitions for map hacks such as 2 byte maf flow and 2 byte load.
If you notice on the left hand side, I also have an add in hack that I didn't mention either...the knock cel add-on. Basically it's a bit of coding added to the base xml tables that allows me to set my cel light to flash when the ecu records over 3 knock counts (or whatever you set it at). Pretty handy little feature, and could save your *** down the road.

That's enough for me for now, I'll continue adding on to this thread as I get time, for a good write-up on an intro to tuning via EcuFlash, check out this link... http://forums.evolutionm.net/showthread.php?t=302895

Chris Harris

Nice work!

Slo_crx1

Thanks! I'll continue to add to this thread and try to keep it as updated as possible.

Just a quick note to add, subaru maps look slightly different in setup compared to the Mitsu/Evo maps. In the subaru maps you can physically disable cel coding, and there are a lot more options available. When I get a chance to pull my WRX rom off my laptop I'll post of some more dealing with them.

Guy-Fast

I know nothing about these cars never even rid in one so pardon the tard questions. Do you guys disable sensors like the Honda crazed owner or have the options in your guys software?

Slo_crx1

Mitsu's MUTIII coding doesn't allow for any sensor disables, but...there are ways to "manipulate" the hex coding to use them for different sensors (i.e. more usable ones) or to replace the sensor's "output" (in reference to the MUT table, usually two seperate hex values that the sensors physically input to. They can be changed to be used for more useful tasks such as datalogging outputs for 2 byte load, 2 byte airflow, 2 byte rpm, etc.). For Subaru it's a little different, as they use just a standard 16bit obd2 coding, so for Subie's you can physically remove a sensor and then disable that entire portion of the coding that might control it or the cel light.

One good example for Mitsu's MUTIII code being manipulated is the latest patch/add-on coding that replaces the specific coding for the Maf sensor. It uses a JDM Evo8/9 3 bar map sensor that replaces the stock 1 bar usdm piece, and manipulates the coding for the Maf to convert it to a speed density system. It really hasn't caught on too much yet, as there is a decent amount of work involved to re-write the MUT table. Just to give you a quick preview of what the MUTIII table actually looks like, here is a very small section of the coding, specifically the area and 2 bytes that I change in my rom ID to allow for 2 byte load logging (for instance in this case, 2 byte load positions would be MUT0X on the first grid, and 0 and 1 on the top grid...so the resulting hex coding would be 0x888F and 0x888E)...


The set of "FFFF"s before the actual hex code is the "reply" areas. Think of it in terms of a modem connecting...a piece of code is transmitted and looks for a return of the same code transmitted and a response. The "FFFF"s are basically the holding areas for the info that is returned and thus ecu responses based off of.

A lot of this coding is still semi-unknown, and it takes a lot of work sitting down with a hex disassembler to figure out the patterns as to what is what.
Hope that helps a little bit Chris. As far as tuning specifically is involved, it's rather straightforward and easy, especially since now you're dealing with afr numbers instead of injector pulse width in m/s. Use a wideband's narrowband output directly to the ecu and those afr numbers become very close to what the real numbers should be.

Slo_crx1

Well it's about time to continue with a little section on Subaru. This particular rom I pulled from my '04 WRX m/t. This is the stock rom before I started tuning it, so keep it in mind if some of the fuel and timing values look a little strange.
The WRX rom you will notice has a lot more tables than the previous Evo 8 rom I showed everyone. Basically there is a table for almost everything imaginable on here, and even some tables that you probably never heard of. There are many tables that correspond to boost control, and with these tables there should never be a need to use any type of external MBC or EBC again.


These tables show boost target in bar absolute, so keep that in mind. Boost limit is also shown that shows even in stock ecu form there is a decent amount of room for increasing boost without hitting a limit or fuel cut. These values of course can be raised if needed, but you'd better have one serious machine to need to do so.



Wastegate duty cycles and compensation by IAT's, these can be adjusted as needed but I'd suggest if you're just starting out to leave these as they are, and honestly I haven't needed to touch them yet as my WRX is still sitting around the 15psi mark for daily driven status.




On to the really important ones...fuel table(s) and injector scaling. Also sort of shown off in the back is the table to adjust front o2 voltage to match a certain AFR range. You'll notice the fuel table shown is the same setup as the Evo 8 table in that it lists AFR's instead of injector m/s. The only main difference between these maps and the Evo 8 is the way they list load...instead of load % Subaru uses airflow in g/rev. Fortunately these numbers can be logged so it's not a guessing game to try and convert values with different formulas. The same rule applies to these maps as with the Evo's...raise the number in the AFR box and the mixture goes leaner. Lower it and it enrichens the mix.



Timing. Timing is the devil on a Subaru. The values shown are degrees of advance BTDC, but these values can be tricky. Subaru has a way of "safeguarding" their setups, sort of like the Evo's "low octane" maps in theory, but a much worse design lol. Subaru's add timing on top of the primary ignition table with a dynamic advance table. IAM (ignition advance multiplier) is another table that determines how much of values called for in the dynamic advance map are actually used. On any kind of a well tuned car which runs knock free, the IAM should be at its' max value and consequently the ecu will run the full dynamic advance called for in the map UNLESS it encounters knock events that will lead to dynamic advance being pulled at specific load/rpm points. You can zero out the dynamic advance table, but to do so will remove the first option mechanism that the ecu has to retard timing in the event of knock. Not a good idea. If there is significant and persistent knock the IAM will get knocked down and less than the full mapped values for dynamic advance will be run across the map. There is also a learned knock correction table which will pull out from the total timing run in specific areas in which the ecu has detected persistent knock. The are also fine learned knock correction tables which do a similar task. All in all there is a heck of a lot more in tuning a Subaru and coping with it's timing strategies and learned corrections than what is currently understood about the Evo ecu. A well tuned Subaru ecu should have timing values in the primary ignition map that are low enough that the full dynamic advance is run, so that the sum of the two are what the tuner wants the car to actually run. The trick is setting them up in a manner that the ecu wants to run that full timing value. Sometimes that means taking away from the primary ignition map and adding to the dynamic advance map. Other times it is the other way around. Either way you go it's a very steep learning curve, and honestly trial and error seems to be the only way to find out what works for your individual setup. That's why it is imperative to log knock counts when datalogging, and if knock persists it's a good idea to stay off the throttle until the issue is resolved. Also shown are tables for individual cylinder timing compensation. Play with them at your own risk.



A few of the fine error correction tables I made note of above.



MAF table in g/s, can be rescaled (when properly datalogged) in the same manner as the Evo allowing for adjustments such as intakes or aftermarket turbos which would affect airflow readings. Also adjustments for MAF Intake Temps.



Like scaling? Different sensor setups? You'll be right at home with these tables then...Coolant Temp sensor scaling, Intake Temp Sensor scaling, EGT sensor scaling, and rev limit. Quite a few more listed in that group that aren't pictured.



Hate emissions? Hate dealing with emissions components and want them gone? Well, here's a few tables that ought to catch your eye. And of course, it wouldn't be complete without...



Disabling sensors! Perhaps you got a nice test pipe and you want to get rid of those dumb o2 sensor codes...easy as pie, just click on the appropriate code number and disable them...easy as that! There's a listing for almost every major obd2 cel code that might plague your WRX.

WTF

so is this ---- free ? wish someone would make it work with a obd2 bmw computer :1

Slo_crx1

The program is free, the only thing you have to buy is the Tactrix cable itself. BMW could possibly be in the works, so far this program supports Mini Cooper, and for Mitsu it supports the Airtrek, Evo 5-9, Lancer, Eclipse, and Galant. On the Subaru end it covers Baja Turbo, Forester GT, Forester turbo, Forester Sti, Forester XT, Impreza Wrx, Impreza Sti, Legacy Gt, Liberty GT, and Outback XT.

If you know how to pull the rom map off your vehicle and feel like dealing with a hex editor for hours on end or some other type of disassembler, then there's no reason why you couldn't write new XML tables to work with this program. The hardest part is syncing EcuFlash with "vehicle X" to read/test write/compare/flash the roms. So far there is no support for obd2 "can-bus" protocol ecu's, but it is in the works for the new Evo X.

Adam Hopkins

You have done an EXCELLENT job with this write up and thought I would just say thanks and good work. I am going to probably mess with my friend evo now

Slo_crx1

Anything to help out dude. As it sits I'm pretty much the only active Evo owner on here, but I could see that changing in the semi-near future. If you have any questions or issues feel free to contact me and I'll help out any way I can. The biggest secret to tuning these cars is datalog, datalog, and more datalogging. Evo 9's also have mivec maps for the adjustable timing on the intake maps, so they're a little more complicated and tend to have much less timing advance than the 8's do. Also there are different rom ID's for different year Evo's...for example mine is a 2003 and came originally with an ID of 94170008. This number code comes up when you first download the rom from the ecu. Since the 0008 rom had issues with a certain p0300 code always popping up (random misfire code), Mitsubishi updated it's roms to a coding numbered 94170015 that updated both the '03 and '04 models ('04's rom ID was 94170014) to get rid of the p0300 issue. Most newer roms have other features not found in the originals, like lean spool enhancements and another timing element that the EcuFlash community is currently working on dealing with a type of timing advancement similar to the Subaru's setup, basically where it ramps timing up a set number of degrees at a time until it encounters knock and then drops back to the original base timing and completes the cycle over again. Over time these values write themselves to the rom so the ecu doesn't have to change timing as much and as drastically as the first few times.