RalliArt ECU reflash - who wants one?

[blk-majik]

A technical note: just because the car has an ODB2 port does not mean that anyone can reprogram anything in the ECU. All that means is that you can read and reset error codes. It is a diagnostics interface. It is not ment to make modifications, and in almost all cases, doesnt even offer it as an option. Read Only. IE: you can turn off the CEL light until it gets reset by something. The ODB2 specification supplies no methods for changing any engine paramaters, period.

The flashes performed on other cars usually require the ECU to be disassbmled and the memory chip removed from its socket, placed in a flash device and updated with a new ROM. Some newer ECUs do have an interface to do this via a PC. To make the ROM, a company has to reverse engineer the data on the stock system, then figure out what new vaules will work. The risk is misinterperating something and blowing the engine because you were off by one binary digit. This has been done on the Evo because the engine is old (circa 1992, maybe earlier in the states) and well know. The ECU is pretty simple, too. The 4g69 is new. This implamentation of MIVEC is new. And its complicated. The mappings are not as linear. It wont be as easy.

I would like to see this type of solution, but dont get your hopes up. I agree with chowetime. Focus on something easier first. You wont convince a company to risk a $20k car a solid month of dedicated research for something that we dont even know will provide us with any gains.

First, ask yourself what engine paramaters you want to change. Then ask yourself how you can change them without a reflash (with a piggyback). Here is a list of some basic things you can control to gain performance:

1) Injection duration
2) Injection time
3) Ignition time

Respectivly, here is some possible variables need to be changed in the ECU to make these modifcations:
1) Target Air/Fuel ratio at given RPM, or o2 readings
2) Crank or cam position sensor at given RPM, throttle position
3) Crank or cam position sensor at given RPM, MAP reading

And again, here are some solutions that already exist that can do this for you:
1) Emanage, Split Second PSC1, SAFCII. Modifies signal to the ECU from the air/fuel sensor
2) Emanage. Delays signal sent to the injectors
3) Lots of MSD products. Reads other variable direclty from sensors to calculate new Ignition time, or delays signals from ecu output. Emanage may be able to take care of some of this, as well.

Its much easier to apply these pre-made solutions to solve your problem, than to just ask for a pretty packaged soulition. The Emanage should work just fine if you change the MAP sensor wiring. Just as Mario or Malvin, as they have both done it. Why re-invent the wheel?

[blk-majik]

I found some good resources, just to give you an idea of what is involved in created an ECU flash.

here is a page that tells how to flash the ECU of a late 80s or early 90s MOPAR: http://www.knizefamily.net/russ/elec.../moparecu.html
Since the 4g69 was developed joinly by MOPAR, mitsu and hyundi, some things may have carried over. This is a pretty basic page. However, this page assumes that the ECU is socketed and you can just take the chip from a differnet car and slap it in your car. Thats like replacing a human hear with a cow heart. Very few models will support it.

here is an advanced page about how someone used Linux to flash a 1G DSM ECU. Again, it assumes you already have the ECU ROM image.
http://esm.logic.net/public/dsm/ecu-flashing.html

Here is a page that generates a 1g/2g DSM ECU ROM image. This is very impressive! But, most cars are not this easy to create an image, but these things are pretty simple. Notice this page was made in 2005, but the cars were produced since what, 1991? It took them 14 years? But it gives you an idea of the things you can change
http://ilostmymind.com/ECU/DSM/custom.html

If any of the stuff on those pages dosnt make sence to you, that is why I think the piggyback is a more effective way to go.

[WoRkZ]

Originally Posted by blk-majik

A technical note: just because the car has an ODB2 port does not mean that anyone can reprogram anything in the ECU. All that means is that you can read and reset error codes. It is a diagnostics interface. It is not ment to make modifications, and in almost all cases, doesnt even offer it as an option. Read Only. IE: you can turn off the CEL light until it gets reset by something. The ODB2 specification supplies no methods for changing any engine paramaters, period.

...

Its much easier to apply these pre-made solutions to solve your problem, than to just ask for a pretty packaged soulition. The Emanage should work just fine if you change the MAP sensor wiring. Just as Mario or Malvin, as they have both done it. Why re-invent the wheel?

Thanks for your input (as always, it's quite valuable).

Now, as far as the wrinting capability is concerned, Mitsubishi just recalled the 2.0l Lancer to reflash the ECU... so I guess this solves it.

And yes, some external hardware options are available... but at what price?!? And do you know many car owners able to tune their car by themselves? I no I can't do that myself... and I am willing to bet that most guys on here either would never attempt it themselves and/or would not even trust the local tuning guy to do it. And anyway, since you,d have to pay a guy and a dyno for at least half a day, there's no way this is a solution for the masses.

Once again, what we need is a 200 - 300$ solution that doesn't involve splicing wires or taking appart the stock (and quite expensive) ECU. We need a solution where you take your car to the local reseller, they plug a laptop in, re-write teh mapping on the car, you pay them 250$ for it and go off on your merry way. If it can be done on other cars, I want this done to mine! Very limited risk, interesting (if somewhat limited) rewards. I'm kinda of a moderate kinda guy... and I just have to assume that just like in the real world, most guys on here are.

So once again, let's focus on 100% software solutions. My knowledge in computers tells me it can be done. The biggest unknown is: Would this be a good value? What would the gains be? If it's only 5hp, there's no way it's worth 250$.. unless maybe it's over the whole RPM range. But if it's 10hp and over, this becomes very interesting considering the way it's so easy to have done to the car and the very limited risk for the owner.

-

[jdizzy]

I agree w/ blk-majik....

[WoRkZ]

Originally Posted by blk-majik

Here is a page that generates a 1g/2g DSM ECU ROM image. This is very impressive! But, most cars are not this easy to create an image, but these things are pretty simple. Notice this page was made in 2005, but the cars were produced since what, 1991? It took them 14 years? But it gives you an idea of the things you can change
http://ilostmymind.com/ECU/DSM/custom.html


...

If any of the stuff on those pages dosnt make sence to you, that is why I think the piggyback is a more effective way to go.

Woaw! If that software really works, we must get the guys to do one that supports the RA as well! BTW: What's the size of our stock injectors?


The fact that most guys wouldn't know what to do of that specific page is exactly the reason why we need this reflash product for ou cars! Once again, no splicing and no risk to the ECU...

I won't be deterred from this so easily!

-

[WoRkZ]

-

And BTW, I'm not worried about tuning shops not having the necessary hardware to acces and flash the ROM (ECU memory). If they don't, they're not serious enough tuners to work on my car anyway. With the right budget, I can order and put together a reflash kit based on windows and USB in about a week... and I know nothing about tuning.

-

[blk-majik]

Ok, its not if we can or can not write. I could write to my ECU EPROM chip tonight if i wanted to. Being an electrical engineering hobbyist, I have the tools already. But I wouldnt do that unless I wanted my car to never run again. Its not IF we can write to it. Its ****WHAT**** we would write to it. The best auto tuner in the world doesnt know jack about PC assembly language, let alone motorola 68H series assembly language, let along propriatary mitsu/motorola cpu assembly language.

Consider this. The ECU is a computer program that takes input form the engine sensors. This program then produces an output to the Engine actuators according to the program's calculatons on these inputs. It runs on a modified Motorola processor in almost every auto ECU in the world. Mitsubishi uses a proriatary modifiecation of these Motorolla processors that we have no details about. And we will never get the details. We do know that they add what are known as OPCODES (numeric software representation of very small transistor arrays on a silicon platter the causes a software command read from the 32 data lines comming off of a register to change the physical open/clsoe state of thousands of these transistors in a specified way) that no one knows anything about. That means that the code is doing mysterious things that we can never figure out. Anyone with the ability to figure out what the other opcodes are doing is already makes enough money to buy a Porsche Carerra GT at $600,000 and still have enough money to live comforably because they work for AMD or Intel on the lead processor reverse engineering team and have access to one of the only 2 labs in the US capable of doing so. So dont expect to get any help from them =)

And even if we did have their help, we could then only READ the program in binary form. To help understand what means, do the following. Here is an experament you can do right now. Hit your start menu and click run. Enter 'cmd' in the box (or command, if you are using windows98 or 95) and hit Ok. Now, at the prompt that pops up on the black screen, type 'edit c:\windows\winhelp.exe'.

Does any of that crap make sense to you? probably not. its binary code that only the computer understands. Now if i gave you a file with 32,768 characters of that random mess of theseb(each character might be telling the CPU to do a handfull of things), would you be able to tell me which part of which single character is representing the idle RPM stored in the program?

THAT is what you are asking a vendor to do. I am usually not one to discourage asking vendors to do something new for us, but this is something that I worked very hard to understand. I'm a computer programmer by trade. I work with x86 Linux kernel drivers, old VMS servers filesystems and Solaris servers all day. I know assembly inside and with the x86, MIPS and older Motorola processors. And I cant do this. Dont expect anyone working at a tuning shop to be able to do it either. It takes more knowledge of motorola assembly programming than auto engineering. The 4g63 ecu in the old eclipse was done by a group of people online like myself who invested years of their spare time and alot of luck into figuring out a very little bit about the ecu. this carried over to the Evo, since it is the same engine. All the flash companys are standing on the shoulders of such hackers.

If you really want to have your mind boggled, here is a walkthough of what the guys managed to disassemble in the DSM ECU. It just tells what things are, and not even how they figured it out. So, this is just the easy part. Enjoy: http://users.wpi.edu/~ktarry/dsmtech/dsmromedit.html

Here is an excerpt for those too lazy to click but interested:

MAF Compensation
Address: 7B6E - 7B82

The MAF compensation map is what the ECU refers to in order to correct the airflow signal coming in from the mass airflow meter. This code also allows you to install a 2g MAF and have the ECU compensate for it properly, or to compensate for hacked MAFs, or aftermarket MAFs that do not have quite the same airflow curve as the stocker.

The stock MAS compensation code is:
5B 5B 59 59 60 65 6C 6E 6E 6F 73 76 7A 81 82 84 87 85 7F 7A 7F

This will vary if the car is a GVR-4, for example, because they have a slightly different MAF.

The MAF compensation code is essential a correction map, based on airflow. The airflow (Hz) points for each of the values are as follows:
0, 25, 50, 75, 100, 125, 150, 175, 200, 225, 250, 275, 300, 400, 500, 600, 800, 1000, 1200, 1400, 1600.

So, for example, if you wanted to increase the airflow signal from zero to 75Hz input by 20%, you would increase the first 4 values in the map by 20%.

Since one of the most popular changes is from a 1g MAF to a 2g MAF, this is how you want to change the code:
First, the map needs to be changed to:
85 85 A8 B6 BE C3 C8 CC D0 D4 D7 DA DC E3 E6 E8 EB EA EA E8 E7

NEXT, you need to change another byte the serves as a multiplier. This byte is located at the address 5669, and the stock value at this point is "64" hex. For a 2g MAS, change that to "40" hex.

In addition, in order to make sure the timing and fuel maps still line up, you need to change the value at 549A from "5E86" (1g) or "62B9" (GVR-4), to the 2g value which is "7A03."

[WoRkZ]

-

Okay, so what you are telling me is that none of the Linux and Windows based tuning software already on the market can read the ECU code for our particular engine? That's a bummer...

This makes the project much more difficult because this is needed in order to be able to read and subsequenltly write to the ECU's ROM.

Isn't there a diagnostic (read only) program out there that already reads ECU mapping for our engine though? I thought I saw one doing a little research a couple of months back.

I've got to look into it and get back to you on whatever I found.


In the meantime, thanks for the info and for your input... and let's not give up just yet.

-

[blk-majik]

You are probably thinking about the OBD2 stuff again. That is not part of the ECU, really, but we can read it =) Its a seporate system that just monitors the sensor signals, much like the piggy back does for its input. It is very useful for aiding while tuning, but it doesnt let you tune anything directly. Basically, its saves you from having to put 100 gauges on the car while tuning.

The ODB2 interface is made to be universal, though. That is, 1000 different cars with 1000 different ECU codes will all use the same OBD2 diagnostics tool. Its an effort to make troublshooting auto problems universal. So when youre CEL light goes on, any pepboys in the world can tell you buy, just by hooking up a $100 OBD2 scan tool, regardless of what kind of car you have =)

Think of it like a computer USB port. All PCs have them so that one mouse can work on any PC. Same concept

Its alot to take in, i know. I didnt know jack about the ECU or OBD2 about a year ago, but i read alot (if work only knew ). Just google around and you can find some really good reads that might help put the final picture together.

[PR_Mivec]

just put a microthech ecu and done.... let the stockone deal with the mivec

[DangerousDan]

Would an ECU chip replacement make more sense? Any major chip manufacturer can figure out a circuit board and design a new chip around the current board to control everything our ECU controls.
My friend has a BMW Z3 and he has a learning chip. When he is driving on the highway it operates at a more moderate level and when he begins to drive on the track the ECU learns the new driving style and adapts to more suit the power needs. I think this would be neat to have, since you wouldn't want to fly around the city all the time and waste gas
I agree that flashing the ECU would take lots of R&D but the replacement ECU would not take nearly as much time if I am not mistaken. Please correct me if I am. I even think one of the mods has a link to a partial wiring diagram of the ECU in his signature, so we know what wires do what. Now all we need to find out is what pin on the ecu receives inputs for each different path on the board and design a new chip. Not nearly as difficult as figuring out how someone elses chip interprets signals. I don't have the piggyback but I am assuming that it modifies the signals on the wiring side going to and from the ECU, which means RRM has already taken half the steps to creating a whole chip upgrade, they just started with the simplist way to go about it. This also means that you may not gain anything more than what the piggyback already provides. The only difference would be maybe a learning chip. Sorry to be so lengthy.

[WoRkZ]

-

I can't see how designing and manufacturing a brand new ECU from scratch can be commercially feasible for the RA. There's not a big enough market for that. So much so since instead of 250$ you'd probably have to pay much closer to 550-600$ for it... or else they wouldn't have put enough R&D in it to make it a good product.

Designing and manufacturing hardware is a long and costly process... and then, they would still have to come up with software for it.

Once again, if it becomes clear that there's no possible way of reflashing the existing harware (ECU), the potential market for this product will go from 500 units to 50 units... and it's not gonna be commercially viable. Plus, as I already stated, most guys are looking for a simple, risk free solution... and an affordable one to boot.

If blk-majik is correct (and I have no reason to doubt him, he's always made sense in all his posts to date) and this reflash involves costly reverse engineering, then all we have to hope for is for one company out there to first crack Mitsubishi's proprietary code and then come up with maps for most of their products on the road today (the second step being the easy part).

And blk-majik: Yeah, the picture is building in my mind quite fast now... but I still haven't given up on it just yet. Maybe the people wo can crack the code are few, but they are out there I'm sure... and I sure hope they drive a Mistsubishi.

Thanks for your implication guys! Feel free to add to what we already know... or to direct us to people you think might be able to help and/or provide more info.

-

[WoRkZ]

-

BTW: I'm kind of a marketing expert... my technical knowledge is limited (but my subsequent understanding of the basic principles is always fast and crisp). It appears that blk-majik has our back covered as far as software is concerned. That means we will also need guys with ties to corporate engineering firms (preferably in the car business) and maybe also guys with ties to the tuning scene (PR-Mivec may be our man on this) . Maybe a businessman would be of help too... afterall, we must know what constitutes an interesting product/market for a developer and for a retailer.

So let the people around you know of this... maybe someone out there is looking for a challenge.

-

[EM@WORKS]

We were the very first tuner to flash the EVO ECU in the spring of '03 and our software engineers have disassembled more maps than any others to gain the best understanding of the factory ECU logic.
We are currently working around the clock to provide a flash for the 2.4L MIVEC and 3.8L Mitsubishi family of ECUs.

This is necessary as we are building a special WORKS TR-Edition Eclipse for Mitsubishi Motors North America for the '05 SEMA show. We could not begin our list of serious engine modifications without complete control over the factory software.
Although we don't have an exact release date, stay tuned to our website for more information as the year progresses.
Thank you.

Unfortunately I won't be checking back on this thread but please call 415.641.WORKS (9675) or write info@worksmitsu.com if you have any questions.

[vonzipper]

o man i am all for this project. my friend has a 04 scooby and when he got his euctek re-flash i was with him the the shop and it is nothing short of amazing. the place is called grd and when you buy an ecu reflash the throw in a free dyno and tune for you. from first hand you can see huge gains. great in vestment.his cost 600 dollars but dont for get the is with an hour of fine tuning. his gains where 30hp and 34 ft. lbs. o man i would love to see those kinds of gain on a ralliart.