EJ20G "RobTune" Info

[Florin1]

Huh? You got me confused now... As far as i know, w/ the 20g ecu one needs to convert the ignition system to COP. Do you know something that i dont? (when i met up with Rob to get my ECU he said he was working on something so i can run my wasted spark)

[rob]

I know of two possible ways to deal with the ignition coversion;

1) rewire the stock ignition with two igniters and use the stock waste spark coil with the G ECU.
2) rewire and run direct fire COP with coils from a +02 WRX.

I think I figured out a third way to use the G ECU and the stock 22t ignition as-is by converting the 4-channel signal to a 2 channel signal. I need to finish wiring it and write up a quick how-to, then find a test monkey. I am coming up bupkis on purchasing my own test vehicle. There are too many subaru nerds around these parts.

-Rob

[93Leg-c]

May need some fixing but how about this one: http://bbs.legacycentral.org/viewtopic.php?f=15&t=43916

[Florin1]

Well i'm ur test Monkey. . Looking forward to the writeup so i can give it a shot.

Check parts shed, someone just posted a turbo wagon in Pulsbo for 850. Its looking pretty beat though. I sent them an email. I think i kinda wish i had a wagon too. lol.

[Florin1]

lol, what he said^

[rob]

Yeah talked to that guy, was first on his list, looking to get it delivered on Saturday. Ran a carfax and BAM, full collision rebuild and exempt mileage. No thank you. There was another nice wagon in portland

http://portland.craigslist.org/clc/cto/1726844034.html

I was going to pick it up on Sunday and BAM guy emails me back that it is now sold. Oh well.......

[dscoobydoo]

If you need a test mule, I have the car I saved from a swap in Auburn ( Sumner side)

Just tell me when and if you want to take a look-see

[MikeyMeyagi]

Rob,
Installed the permanent mount wideband into the car tonight and did some runs, turned the boost up to 17 peak tapering to 15, AFR's at wot were 11.5 to 11.8 or so, no knock. pulls hard. im thinking its time to add some timing!

[rob]

Cool Mike! Have you had a chance to log the MAF output voltage? I would like to know how close you are to maxing it out.

-Rob

[BSOD2600]

Have you had a chance to log the MAF output voltage? I would like to know how close you are to maxing it out.

I've done various ECU logging with different stages/components and posted results here. Now that I've "fixed" the knocking issue (knock sensor relocation), I'll be able to do more useful data logging again with the Revtronix stage 2 v3 chip... well once the fuel tables have completely learned.

[rob]

Have you had a chance to log the MAF output voltage? I would like to know how close you are to maxing it out.

I've done various ECU logging with different stages/components and posted results here. Now that I've "fixed" the knocking issue (knock sensor relocation), I'll be able to do more useful data logging again with the Revtronix stage 2 v3 chip... well once the fuel tables have completely learned.

Hmm lots of data with a lot of changing variables

Did Revtronix claim to have modified the MAF transfer table? What kind of live adjustments can you make with Revtronix software? From the little I have read, it sounds like you simply write to the closed loop short term tables?

Mike, I still would like to see direct voltage measurement to see where your MAF is at.

-Rob

[ericem]

They must have, since you need to use a JECS maf sensor.

[rob]

Given the amount of ecu "learning" that the revtronix required, I am very skeptical. Reading carefully it sounds like the sole purpose of him recommending the bigger MAF was simply to prevent maxing out the stock MAF signal. It would explain some of the odd reports of extremely lean idle and excessively high EGTs. I finally read through the entire revtronix thread and these thoughts were bouncing around in my head (as well as my eyes bleeding...)

Really it doesn't matter, it's more just a curiosity.

[Florin1]

I love going to the new post category and seeing this thread at the top again. Its like Christmas! I quickly jump to the last page to see the latest news. Lol
Fixed. It was late. I love Christmas...

[BSOD2600]

Hmm lots of data with a lot of changing variables

Did Revtronix claim to have modified the MAF transfer table? What kind of live adjustments can you make with Revtronix software? From the little I have read, it sounds like you simply write to the closed loop short term tables?

Heh, yup. Wanted to get a decent idea where things sat with various components, since no one else on the forum really had backed up their posts with data. A bunch of butt dynos or "it ran" stuff doesn't cut it for me.

Yes, Mike@Rev claimed to adjust the MAF transfer table for the JECS vs stock MAF (and eliminated the CEL thrown from it w/o the need to ground the pin). The Revtronix RevScan also exposed a parameter called "airflow", which a few others were skeptical the ECU actually had. Upon further investigation and COM port sniffing, both MAF voltage and MAF airflow were requesting the same ECU memory location... thus I believe the airflow is a fake/calculated parameter. Mike@Rev would never talk about how revscan would retrieve or calculate data. Older versions of revscan there was a DAT file one could open and find parameter locations. Newer Revscan version, one could open with a binary editor and do the same. Now it's encrypted/compressed and no such luck. RevScan is free to download and runs w/o any cable attached. The EvoScan subaru cable works with revscan/evoscan.

RevScan allows one to:
-view ecu parameters.
-read/reset ECU codes
-adjust full throttle fuel trims (chip added parameter)
-change the octane factor (OF) (aka ignition advance multiplier).
-view/change the "long term fuel trims" in memory. Cells 1-3 aren't editable while 4-7, idle are.
-view/change knock retard table
-view/change knock count table.

The high amount of "learning" required is IMO an stock ECU limitation. For example, myself. Last year when I had my 94 TW and the same hardware, it was running great on the stage 2 chip (100% OF and target AFR while in boost) -- seem to recall several hundred miles of driving for this to occur. Now with the same hardware in my 94 SS, after "fixing" the knock sensor issue, the OF is back to 100%. This is where the slow learning comes in -- fuel tables. While on boost my AFR still dips down into the ~10 range. From before, I recall that suddenly one day it just became correct (aka ~11.5 AFR) and there is stayed.

While one can edit the fuel tables with revscan, they don't seem to stick properly. For example, before resetting my ECU for the KS issues, I recorded down all the fuel table numbers. Reset the ecu, drove around a bit, came back put back in the fuel table numbers, drove around more, checked the fuel table and the numbers (except idle) where all reset way out of range of what I manually typed in. This leads me to believe mike@rev doesn't have Revscan properly editing the fuel table fields -- or I guess the ECU simply doesn't allow it and must go through its slow ass learning cycles.

Excessively high EGT.. can't speak to that one since no sensor, but I use to smell a different kind of exhaust smell after a hard run while the ecu was learning/increasing the OF. Extremely lean idle IS a known bug with the stage 2 chip that only version 3 resolved -- the fuel trims were getting maxed out of their stock ranges (I recall -20/+14%). This phenomena is observable via data logging and I have it documented on the revtronix forum. Speaking of revtronix forum... there are several good posts with details from mike on the chip, etc... just have to sort through all the spam crap. I've actually contemplated going through all of his posts and making a manifesto of sorts of all the things he said.

[rob]

I love going to the new post category and seeing this thread at the top again. Its like Christnas! I quickly jump to the last page to see the latest news. Lol

Well Merry ChristNas Florin.

Some good news to report; Drew and I are working out details on using his car for testing. Things are looking good for having something to tinker with in my garage in a couple of weeks. My first set of experiments include extracting the stock ECU code, figuring out the hardware requirements for an add-in board and perhaps experimenting with a G ECU.

Second piece of good news is that I have been getting reports from Mike about his results with my regular G tune. So far is sounds like it's working very well. We are going to push up the timing a bit and see how it behaves. With some real results, I feel a lot better with the adaptation of the G ECU and the 22t motor.

-Rob

[Florin1]

I love going to the new post category and seeing this thread at the top again. Its like Christnas! I quickly jump to the last page to see the latest news. Lol

Well Merry ChristNas Florin.

Some good news to report; Drew and I are working out details on using his car for testing. Things are looking good for having something to tinker with in my garage in a couple of weeks. My first set of experiments include extracting the stock ECU code, figuring out the hardware requirements for an add-in board and perhaps experimenting with a G ECU.

Second piece of good news is that I have been getting reports from Mike about his results with my regular G tune. So far is sounds like it's working very well. We are going to push up the timing a bit and see how it behaves. With some real results, I feel a lot better with the adaptation of the G ECU and the 22t motor.

-Rob

Fixed. I suck for misspelling Christmas. Great news on your progress!

I picked up a vf22 turbo and some other goodies. Waiting for shipment. Than i'll be ready to throw in the jecs, injectors, and your ecu. Hopefully you will have more news about the COP/WS so i dont have to convert. I'm also looking into dynoing my car for a baseline. Just researching shops that are semi close so i can go back after.

[rob]

Yes, Mike@Rev claimed to adjust the MAF transfer table for the JECS vs stock MAF (and eliminated the CEL thrown from it w/o the need to ground the pin). The Revtronix RevScan also exposed a parameter called "airflow", which a few others were skeptical the ECU actually had. Upon further investigation and COM port sniffing, both MAF voltage and MAF airflow were requesting the same ECU memory location... thus I believe the airflow is a fake/calculated parameter. Mike@Rev would never talk about how revscan would retrieve or calculate data.

Did he actually explicitly spell that out, or is it just an assumption? Right, the early ECUs (at least from experience) don't deal with g/sec. There is a transfer function for voltage versus a unitless value. Unless he spent the time on a flow bench, then it's likely a meaningless value.

Older versions of revscan there was a DAT file one could open and find parameter locations. Newer Revscan version, one could open with a binary editor and do the same. Now it's encrypted/compressed and no such luck. RevScan is free to download and runs w/o any cable attached. The EvoScan subaru cable works with revscan/evoscan.

If you can put those params in a list, that would be helpful.

The high amount of "learning" required is IMO an stock ECU limitation. For example, myself. Last year when I had my 94 TW and the same hardware, it was running great on the stage 2 chip (100% OF and target AFR while in boost) -- seem to recall several hundred miles of driving for this to occur. Now with the same hardware in my 94 SS, after "fixing" the knock sensor issue, the OF is back to 100%. This is where the slow learning comes in -- fuel tables. While on boost my AFR still dips down into the ~10 range. From before, I recall that suddenly one day it just became correct (aka ~11.5 AFR) and there is stayed.

The ECU has these learning tables (knock correction igntion tables and O2 feedback fuel trims) as minor adjustments over time. My opinion is that if the ECU needs large adjustments to these tables over a long period of time to run the engine properly, there is a tuning issue. Modifying the main parameters like the MAF table, or the injector constants is complicated stuff, and if you get it wrong, these sort of issues pop up.

While one can edit the fuel tables with revscan, they don't seem to stick properly. For example, before resetting my ECU for the KS issues, I recorded down all the fuel table numbers. Reset the ecu, drove around a bit, came back put back in the fuel table numbers, drove around more, checked the fuel table and the numbers (except idle) where all reset way out of range of what I manually typed in. This leads me to believe mike@rev doesn't have Revscan properly editing the fuel table fields -- or I guess the ECU simply doesn't allow it and must go through its slow ass learning cycles.

My question why would you expect them to stick? If they really the fuel trim RAM locations, then all you are really doing is pre-loading the values, but the ECU will overwrite them through it's continous learning process.

Excessively high EGT.. can't speak to that one since no sensor, but I use to smell a different kind of exhaust smell after a hard run while the ecu was learning/increasing the OF. Extremely lean idle IS a known bug with the stage 2 chip that only version 3 resolved -- the fuel trims were getting maxed out of their stock ranges (I recall -20/+14%). This phenomena is observable via data logging and I have it documented on the revtronix forum. Speaking of revtronix forum... there are several good posts with details from mike on the chip, etc... just have to sort through all the spam crap. I've actually contemplated going through all of his posts and making a manifesto of sorts of all the things he said.

Some of the reported values were 1600+ which would scare the crap out of me. My guess would be the MAF table was not properly set which would really monkey up the fuel calculations. It would be great if you could datamine the old revtronix forum. I have looked at it a couple of times and walked away feeling like I picked up an STD...yeesh

-Rob

[BSOD2600]

Yes, Mike@Rev claimed to adjust the MAF transfer table for the JECS vs stock MAF (and eliminated the CEL thrown from it w/o the need to ground the pin). The Revtronix RevScan also exposed a parameter called "airflow", which a few others were skeptical the ECU actually had. Upon further investigation and COM port sniffing, both MAF voltage and MAF airflow were requesting the same ECU memory location... thus I believe the airflow is a fake/calculated parameter. Mike@Rev would never talk about how revscan would retrieve or calculate data.

Did he actually explicitly spell that out, or is it just an assumption? Right, the early ECUs (at least from experience) don't deal with g/sec. There is a transfer function for voltage versus a unitless value. Unless he spent the time on a flow bench, then it's likely a meaningless value.

On a forum post he stated, "Recalibrated for higher flow JECS MAF sensor". He did in several places also state the following... which I hoped is from flow testing, "Stock MAF limit is around 160 g/s, JECS MAF limit is around 215 g/s."

Older versions of revscan there was a DAT file one could open and find parameter locations. Newer Revscan version, one could open with a binary editor and do the same. Now it's encrypted/compressed and no such luck. RevScan is free to download and runs w/o any cable attached. The EvoScan subaru cable works with revscan/evoscan.

If you can put those params in a list, that would be helpful.

In time, yes. Otherwise, I reverse engineered most of them and updated the EvoScan subaru data.xml file with them. That can be found on this post

While one can edit the fuel tables with revscan, they don't seem to stick properly. For example, before resetting my ECU for the KS issues, I recorded down all the fuel table numbers. Reset the ecu, drove around a bit, came back put back in the fuel table numbers, drove around more, checked the fuel table and the numbers (except idle) where all reset way out of range of what I manually typed in. This leads me to believe mike@rev doesn't have Revscan properly editing the fuel table fields -- or I guess the ECU simply doesn't allow it and must go through its slow ass learning cycles.

My question why would you expect them to stick? If they really the fuel trim RAM locations, then all you are really doing is pre-loading the values, but the ECU will overwrite them through it's continous learning process.

Yes, I do/did. The ECU had the fuel tables in a decent state before the reset. where the AFR was a nice 11.5 during boost (I'm ignoring my OF = 0% issue). Sure I understand that the fuel trims are constantly learning, changing, adapting, etc. But I figure if I gave it a good starting point, where it was previously, than the changes it would be making would not be all that large. This how ever is not the case.

Excessively high EGT.. can't speak to that one since no sensor, but I use to smell a different kind of exhaust smell after a hard run while the ecu was learning/increasing the OF. Extremely lean idle IS a known bug with the stage 2 chip that only version 3 resolved -- the fuel trims were getting maxed out of their stock ranges (I recall -20/+14%). This phenomena is observable via data logging and I have it documented on the revtronix forum. Speaking of revtronix forum... there are several good posts with details from mike on the chip, etc... just have to sort through all the spam crap. I've actually contemplated going through all of his posts and making a manifesto of sorts of all the things he said.

Some of the reported values were 1600+ which would scare the crap out of me. My guess would be the MAF table was not properly set which would really monkey up the fuel calculations. It would be great if you could datamine the old revtronix forum. I have looked at it a couple of times and walked away feeling like I picked up an STD...yeesh

I found a post from Mike@Rev stating, "We wanted to limit exhaust temp at the turbo to 900C (1650F) so we tuned for ~900C max at the start of the up-pipe. This leaves a safety margin because the exhaust will cool off by the time it reaches the turbo. We measured EGT at the end of the passenger side manifold right before the up-pipe." so the reports you've heard sound correct.

Also, I've created a new thread, The legacy of Revtronix to sum up the revtronix knowledge.

[rob]

Thanks for the info and the revtronix forum archive post. It looks he did take into account the MAF table changes. Sorry if I missed this in the other post, but do you have the conversion value from MAF volts to g/s for the JECS MAF?

I personally get a little nervous with EGTs that high on the exhaust port exit. Just FYI, on my tune, 19psi with the TD05 the max I ever see is ~1400F on WOT pulls. On uphill cruise it can be up to 1500 or so. Also I read about your knock sensor relocation. That makes me nervous too. If you move it away from the stock location, I hope you aren't surprised that you do not see any knock correction

-Rob

[Legacy777]

I'd like to throw in my two cents about the Revtronix chip. This is not meant to be negative or deragotory, but just my comments. There seems to be a lot of variability between users as to one chip/setup working and another not working. Or having to let the ecu "learn" the new parameters to work properly. I am with Rob on this one. If the learning parametrs are way off, that signifies a tuning issue.

Because of that, personally, I would want to verify the information provided from revtronix to ensure it's accuracy and/or make any adjustments I felt needed to be made.

[93forestpearl]

I personally get a little nervous with EGTs that high on the exhaust port exit. Just FYI, on my tune, 19psi with the TD05 the max I ever see is ~1400F on WOT pulls. On uphill cruise it can be up to 1500 or so.

-Rob

EGT's are going to be higher right before the uppipe than right off the exhaust port. The turbulence right there is much higher from the collector right before that location, and you read higher temps. I read up to 1700 degrees in that spot and I never melted a piston in two years of running my previous setup. (GT2871R @ 23psi) I wouldn't want to see those temps right near the exhaust port though.

For example, watch the exhaust manifold on this Honda F1 engine. http://www.youtube.com/watch?v=pdWlnLM8Rmw

[rob]

Hey Dan,

Not to wander too far off topic, but if you start here and go down a couple post, you will see 1600-1700 was measured at the port. Personally, I wouldn't be comfortable with those temps.

Sorry for the divergence, now back to the topic at hand.....

[93forestpearl]

My bad. I went off the post in this thread saying it was before the uppipe.

Carry on.

[brand]

So if I'm following this correctly, you're looking into modifying the EJ22T ECU so you can put your tune on it, and if that runs into snags then the options is to wire the car to run an EJ20G ECU?