Have I found the ECU's maps?

vrg3 · Post by **vrg3** » Sat Oct 25, 2003 3:38 am

I'm trying to understand our ECUs better, with the ultimate goal of remapping them (since I believe that will be the safest and most cost-effective way to improve engine management). The trouble is, I have very limited experience dealing with fuel injection mapping, and none with ignition timing mapping.

Hopefully some of you guys can help fill that void. :)

I've been playing with a 1992 non-turbo ECU, part number 22611AA930F. So far I've found what appear to be three maps and one lookup table. It seems the maps are basically two tables specifying the sample points and a 2D array of values. I guessed that one table was for fuel and the other for ignition... I might have the scale wrong though. I'm not sure what the third is; it's also smaller (8x8 instead of 16x16). I vaguely recall seeing a graph like that of the unknown lookup table on a tuner's web site before, representing MAF voltage vs. airflow, so maybe the ECU uses that internally instead of taking the MAF voltage at face value.

Can anyone tell me if I'm on the right track with these?

http://www.graphics.cornell.edu/~v/pics/aa930f.gif

Post by **Legacy777** » Sun Oct 26, 2003 2:19 am

hmmm.....I know some people that may be able to help.....

You may want to try emailing some tuners, cobb for example, tell them what you're doing and see if they can provide some feedback

FG!! · Post by **FG!!** » Sun Oct 26, 2003 8:16 am

Did you make up the units on the graphs too?

Are you sure it uses Airflow and not Thottle Position?

It looks like it has the chareteristic midrange 3500rpm dip in the fuel at WOT.

Can you adjust the maps?

teaguespeed · Post by **teaguespeed** » Sun Oct 26, 2003 7:33 pm

whoa! thats cool, how do you get that stuff out of the ecu?

-K- · Post by **-K-** » Sun Oct 26, 2003 7:52 pm

You know there is a forum I can't remember the name of that's all about hacking ecu's and maping stand alone ecu's. They are all WRX guys I think, but I bet they would help. I think it had hack or hackers in the name and I found the link on nasioc.

vrg3 · Post by **vrg3** » Mon Oct 27, 2003 5:22 am

Josh - Thanks for the references... Yeah, I'll probably contact them at some point. I do want all my work here to be publicly available, though, so I want to make sure I'm not taking business away from people who help me.

Alan - I did make up the units for RPM; the other stuff I don't have units for.

I'm pretty sure it uses airflow (or some kind of corrected airflow number); it doesn't fuel based on the TPS. It may be some kind of computed "load" value... possibly something as simple as airflow divided by RPM.

You're right, it does seem to have that dip, so that's encouraging.

I do not yet have the ability to adjust the maps, but see below for more info on that.

teaguespeed - The Select Monitor works by asking the ECU for values out of its data memory. The Select Monitor basically says, "tell me the value of the byte at address 1234" and the ECU answers "10, 10, 10, 11, 12, 13, 14 ..." until told to stop. (see http://lillith.sk.tsukuba.ac.jp/~kashima/car/gc8-e.html for more info from another guy who hacked at his ECU) Well, it turns out that you can ask for the value at any address, including ROM. On this ECU it appears that there is a 64-kilobyte address space, with the first 32K being RAM and the last 32K being ROM. So I set up the ECU to run on the bench, hooked it up to my computer, and dumped all the data from the ECU and then viewed it graphically, since the human eye is really good at recognizing patterns.

K - http://www.wrxhackers.com/? I looked on there a long time ago; I should take another look. Thanks.

Okay, so a little more info:

The 1992 non-turbo ECU I've been playing with is made by JECS (Japan Engine Control Systems or something similar). The processor is a 16-bit Renesas 7790-series which is no longer made, and no longer supported by Renesas. Apparently they shredded all the documentation on it a few years ago or something. There's another big chip on there which may or may not be the stock ROM. There is a socket for a replacement ROM, which is a 27C1028. The 27C1028 is also no longer produced or supported by anyone, but it is a 1-megabit EPROM, which can either be addressed as 128K 8-bit bytes or 64K 16-bit words. That's really weird to me, since that's like twice as much space as the 7790 can address. Apparently it just doesn't use it all. People have apparently had success emulating the 27C1028 with some other more standard EPROMs.

But, I found something a little more exciting... I pulled the ECU out of my car (a 1993 Turbo) to plug it into my ECU rom-dumping apparatus. At first I was bummed; nothing happened. The ECU didn't seem to respond to the commands at all. I shrugged it off and decided that I might as well pop the box open and re-examine the innards (the last time I looked inside it was soon after I first got the car so I didn't know what I was looking for).

Well, apparently this ECU is much nicer to deal with! It's made by Hitachi rather than JECS. It contains an 8-bit Hitachi processor that appears to be well documented. The external ROM socket is way weird, though; it's some kind of 30-pin male socket, bigger than the usual dual-inline-pin stuff. Bizarre. But, you know what? I don't care! Because underneath that socket is the stock ROM, an industry-standard 27C256 32-kilobyte EPROM! I can very easily desolder the EPROM and replace it with a socket so I can swap ROMs at will. My plan is to use a 28C256 in its place (that is an EEPROM, which means it's easier to reprogram, that almost perfectly swaps in).

I actually have experience doing almost exactly the same thing, since Subaru Select Monitor cartridges contain 27C256s and I built a copier for them that used a "blank" cartridge containing a 28C256.

I'm hypothesizing here, but: apparently the 90-91 ECUs are Hitachi and the 92-94 ECUs are JECS, but all the turbo ECUs are Hitachi?

Hopefully when I dump the ROM from my Hitachi ECU I'll find maps as easily as on the JECS one. And then I'll pretty much be ready to start remapping it, if I'm lucky.

This could be really cool, if I can easily remap Hitachi ECUs. Using a slightly modified stock ECU could be a reasonable option for people doing turbo swaps and stuff.

Of course, there's still a lot more to it than the maps... There are calibrations and unit conversions and rev/boost/speed limiters and all kinds of other stuff. But since the Hitachi stuff is better documented it should be possible to figure it all out.

I think I'll just concentrate more on the Hitachi ECUs for now, since they can be swapped in place of JECS ECUs as far as I can tell.

So nobody has any ideas about what this unidentified 8x8 map might be?

Thanks, guys.

FG!! · Post by **FG!!** » Mon Oct 27, 2003 7:38 am

let me just offer some things things that i'm thinking

accelerator pump function (so when you you mash the pedal it doesn't stall)?
ambient air temp correction?
injector timing (injector fires before tdc, but i think it's using a set value) ?
or some some combination?

I'm rusty, but at least i'm offering up some stuff to think about

Post by **Legacy777** » Mon Oct 27, 2003 2:54 pm

Yeah I believe all the 90-91 models had the 8-bit ECU. that's interesting the Turbo's didn't convert over to the 16-bit ECU as well.

What advantages do the 16-bit ECU's have?

vrg3 · Post by **vrg3** » Mon Oct 27, 2003 3:05 pm

Alan - There's definitely a bunch of correction factors... TPS acceleration enrichment, coolant temperature enrichment and advance, short- and long-term fuel trims and timing adjustment, barometric pressure correction, etc. There isn't an air temperature correction because there's no intake air temperature sensor. I don't think it varies injector pulse timing, but I really never thought about that at all. I have no idea how those are stored and whether or not they're adjustable at all. Maybe with the 8-bit ECU I'll be able to figure it out. You say you're rusty -- does that mean you have some experience with this stuff?

Josh - I don't know if there are advantages or not... It's possible that the 8-bit ECU has simpler maps or doesn't do as much extra calculation or something. I haven't determined the clock rate of the Hitachi processor yet, but the 16-bit Renesas processor runs at 8 Mhz, which is really pretty fast.

I do know that Subaru stuck with the 16-bit 7790-series processor for many years in foreign markets where they continued to make OBD-1 ECUs.

I've only disassembled one turbo ECU (mine), but since mine is the most recent of the series of five sequentially superceding part numbers I would presume the earlier ones are also 8-bit. It's possible that all the differences would be just in the programming. Hopefully the dead ECU that MK is giving me will be an earlier part number than mine... the EPROM on its board is probably still fine so I can compare its code to mine.

FG!! · Post by **FG!!** » Mon Oct 27, 2003 7:00 pm

I do have some experience, but it was all with aftermarket systems.

Playing with the fuel map at WOT is what your gonna want to play with first. It's almost guarenteed that you'll see improvement. Everything else your going to need a dyno for, unless you just want to pull the timing back for high boost.

But like i said, I'm rusty. If you can find somebody who does this more often, take there advice over mine.

vrg3 · Post by **vrg3** » Mon Oct 27, 2003 8:27 pm

Yeah, my only tuning experience is a very small bit of playing with MegaSquirt, since that's what my brother is using on his Supra.

Are you experienced dealing with fuel maps on turbocharged engines?

I was about to say "the fuel map at WOT would probably just be the portion of the map that I've labeled as highest airflow" but I just realized that on a turbocharged engine you can easily have widely different airflows at any given engine speed at WOT...

What kind of changes do most people do to get more power without actually modifying the engine mechanically? Is it that stock engine management runs a bit too rich and with timing a bit too retarded?

Don't worry about liability with what you say -- it's not like I'm gonna go lean out all my settings until my engine goes poof just because you said that stock mixtures are rich or anything like that. I just want to know more about what I'm getting into.

Thanks.

FG!! · Post by **FG!!** » Mon Oct 27, 2003 10:34 pm

You usually want to add fuel to level out that dip at 3500 and then add a smidge over the entire range. Contrary to what some people say, getting closer to stoich does not make more power. Extra fuel makes more power becuase it cools the combustion chamber. Too much though and it'll probably have some harmful effects on your turbo and your cats if you still have them.

I heard they make it lean in that dip for EPA testing.

I would leave the timing alone if you have a turbo. Possibly retard it more if you want high boost. Take it too a dyno if your NA and want to advance it.

I don't have experience with mapping turbo cars, but I read the Cliffs Notes

georryan · Post by **georryan** » Tue Oct 28, 2003 12:50 am

I wonder.....

I am thinking of doing an awic setup in my car, and I was thinking about the different ecu's. I think the 2.0 ecu actually has takes the reading from a temp sensor and turns the pump on for the intercooler on an as needed basis. I wonder if our ecu's have that capability or if that just isn't setup into our ecu.

-Ryan

vrg3 · Post by **vrg3** » Tue Oct 28, 2003 2:42 am

I don't think it has a temperature sensor; I think it just controls the motor based on other parameters, primarily manifold pressure.

I'm quite curious about these potential "hidden features" in our ECU... Maybe they do contain code to control an intercooler pump, or to do 3-, 4-, or 6-channel ignition, and maybe 6 injectors. (I believe at least early SVXes had very similar ECUs.)

I think after installing an intercooler or changing the turbo it may be advantageous to alter the timing map. Since the ECU has no intake air temperature sensor it must just have a map to predict good timing advance based on how hot intake air is at various levels of boost. A more efficient turbo or an intercooler would decrease the intake air temperature, allowing for more advance. The ECU does some learning about timing but it's unclear how much.

...maybe we can figure that out too. :)

vrg3 · Post by **vrg3** » Fri Oct 31, 2003 10:52 pm

Okay, after playing with the Select Monitor some more, I think I'm fairly certain that the maps are indexed by "load" and RPM. The ECU computes a load value based on a bunch of sensors. There's a strong correlation between manifold pressure (the traditional definition of "load") and the ECU's computed load, but it's certainly not that simple (especially given that most BC/BFs don't have manifold pressure sensors). It also appears to be more complex than just airflow divided by engine speed because at WOT it seemed to hold fairly steady on my car even when airflow did not.

This is similar to mapping by throttle position as Alan was talking about, but more precise.

On the non-turbo model whose maps I have linked to above, the highest load position corresponds to WOT, and the lowest correspond to coasting (idle is just above coasting).

On turbo models I would presume the map goes much higher allowing for the higher loads of boost, so WOT could be anywhere from the highest number on non-turbo maps to the highest number on the turbo map.

I still plan to dump my turbo ECU's memory and show you guys what I find; I just need a little time and a few components that I don't have on hand.

Still no clue about the unknown map. The "unknown table" appears very similar to the airflow calibration chart shown at EcuTek's web site:

http://www.ecutek.com/tuning/induction/

So I'm pretty sure that's what it is. The outliers must be errors in transfer; I'll have to transfer the ROM more carefully in the future.

vrg3 · Post by **vrg3** » Sun Nov 02, 2003 5:10 am

I was just looking at this data again, and I think I have the axes wrong or something.

If one were to follow the graph of ignition timing as I plotted it above, it would appear that advance is greatest when the engine is under load. But the opposite should be true, right? Usually timing is advanced when the load is light.

I probably have the axes switched; the axis marked "airflow" is probably "RPM" and the axis marked "RPM" is probably "Load."

Does that sound right to you guys?

Would the fuel map still look "right" if its axes were switched, too? I would expect both maps to be stored in the same manner.

THAWA · Post by **THAWA** » Sun Nov 02, 2003 6:51 am

I don't quite understand the whole advance/retard thing but if you look at the highest air flow the advance is almost at 40 at idle/low rpm, but at 6k its around 30. Same with the lowest airflow at idle its around 20 but then after 3k its around 0.

FG!! · Post by **FG!!** » Sun Nov 02, 2003 9:24 am

no, it looks pretty good as you have it. don't quote me on that.

You might want to take a look at some sample maps. Download the DTA software, it comes with a variety of sample maps: http://www.dtafast.co.uk/dtawin.zip
The turbo sample maps for fuel are interesting, they're very constant over TP, but then they have some correction factors for boost though.

And I would be really careful if you want to play around too much with advancing the timing. Do you have and EGT gauge?

vrg3 · Post by **vrg3** » Sun Nov 02, 2003 4:16 pm

THAWA - Well, like I said above, it's not airflow. It's really engine load. You're right that advance appears to decrease with increased engine speed... I don't know anything about whether or not that is common.

Alan - Thanks for the pointer. I was looking on the web for some sample maps and having a hard time finding any, especially for non-turbos. DTA tunes n-alpha (RPM vs. TPS)? That seems poorly suited to turbos since load can vary widely at any given throttle position. I'll look through all the stuff, though; thanks.

At the moment I'm just trying to learn how it all works. I'm not remapping yet (this ECU doesn't even work on my car). I'm not going to start until I have my EGT gauge and wideband oxygen sensor set up at the least.

Thanks for the help again, guys.

vrg3 · Post by **vrg3** » Sun Nov 02, 2003 4:36 pm

I looked through DTA's sample maps, and they seem to support the "retard with load, advance with speed" notion, which would indicate that I have the axes backwards for timing.

The turbo fuel maps were just as you described. They do use n-alpha with pressure correction factors, so I guess it makes sense that they'd just ignore throttle position on turbocharged cars.

The non-turbo fuel maps look quite a bit like the one I've got here, though, which is weird... since that means the axes are likely switched on one of the graphs but not the other.

teaguespeed · Post by **teaguespeed** » Sun Nov 02, 2003 5:57 pm

but aren't you playing with a non-turbo ecu?

mTk · Post by **mTk** » Sun Nov 02, 2003 6:41 pm

teaguespeed wrote:but aren't you playing with a non-turbo ecu?

If he is, he won't be for long. I just have to get off my ass and send vrg3 my toasted turbo ecu (Sorry vrg3

)

MK

teaguespeed · Post by **teaguespeed** » Sun Nov 02, 2003 7:52 pm

MK, i was refering to this:

vrg3 wrote:The non-turbo fuel maps look quite a bit like the one I've got here, though, which is weird... since that means the axes are likely switched on one of the graphs but not the other.

vrg3 · Post by **vrg3** » Sun Nov 02, 2003 8:32 pm

teaguespeed - Yes, I am playing with a non-turbo ECU for the moment. In that quote I was referring to the non-turbo maps that came with DTA's software. Sorry if I was unclear.

MK - It's ok, I can wait until you have time to get it shipped out. You're being very generous to give me the dead ECU gratis. :)

360ed · Post by **360ed** » Tue Nov 04, 2003 10:59 am

I'm watching this thread with greeaaat interest.... I've worked with these types of controllers before, but not in a vehicle. I'm changing careers, wandering away from the bench, and I think tinkering with ecu's could be a cool hobby. I have an old eprom reader lying around somewhere, a nice iron and an idle PC...

Anyways,

Renesas had a library of spec sheets for the 7900s

http://www.renesas.com/eng/products/mpu ... index.html

And a front page for tools and general info

http://www.renesas.com/eng/products/mpu ... 7900_e.htm

legacycentral bbs

A forum for 89-94 BC-BF(BJ) Legacy Owners and Fans

Have I found the ECU's maps?

Have I found the ECU's maps?