First time tuning an E38 from scratch

[mowton]

Matt,

Try this new config file. I set up the analog input to HPTuners to read 9.7 stoich from the LC-1 ....I think There is a new EIO (LC-1 E85 WB) and a new AFR error PID (LC-1 E85 AFR err). Log these as well to test both.

Matts rev A .cfg

Ed M.

[bikewiz]

Thanks! I'll try that out.

Taking a break from the Mustang and building the trans in my Cummins Ram this weekend. Will be back to tuning the race car next week most likely.

[bikewiz]

Ed,

I'm pretty sure I've already got a similar setup to what you suggested. I attached my config file. I started with the basic PIDs that the Greg Banish video has you set up to plot MAF lambda error, and then I tweaked a few things from there. My AFR reading from the LC1 is scaled down to e85 stoich in the PIDs. My MAF lambda error is historgram #10.

Also, is there a way to fast forward the VCM Scanner playback? I know you can go up to 10 frames per seconds, but when I'm trying to see what my average lambda error was at the end of a 8 min log, it takes forever to get there.

[bikewiz]

I'm looking over the last couple logs i did, and I'm noticing something strange. Typically, I load a tune, run the car (at idle for now) and log the MAF lambda error. Then I go back to the editor and multiply the cells in the MAF table by what my logs showed for that specific cell. For example, if I tuned the car and my MAF lambda error was 0.90 from 2550-3000 Hz, I go into the editor, select those cells in the MAF table, and multiple them by 0.9 (effectively removing 10% MAF from those cells.) Then, when I re-flash the new calibration into the car, I expect to fire the car up and see those cells at 1.0 in my scanner, or close to it.

The last time I did this, I had readings of about 0.88-0.91 for the MAF lambda error from 2550-3150 Hz. So, I went to the editor, multiplied those cells by 0.91 (pulling 9% MAF) and saved as a new file. Then I flashed the car with the tune and made another scanner log of the results. My changes appeared to have no effect. The log still showed I have a MAF lambda error of 0.9 in those cells.

It was the last flash of the night, so I haven't made any changes since then. Is it possible I was tired and made the mistake of selecting the wrong tune file and loading the same tune that was already in the car? Sure. I'm wondering, how can I use the MPVI pro to read the current calibration in the car and see what is loaded in the ECM? I know I could just spend 2 mins and flash it again, but I would like to learn if its possible to read the current calibration and how.

[HartRod]

Just go into the editor under "Flash" and do read entire.

[bikewiz]

OK cool! I tried doing a read entire once before and the editor locked up on me. Ill give it a try tonight.

[bikewiz]

Hope you all are staying warm! It's so cold here that the car is near impossible to start. The piston rings aren't even seated yet and these arctic starting attempts are filling the crankcase with fuel. Already had to change the oil once. Just going to hold off any further tuning until I get some consistent weather above freezing lol

[bikewiz]

So, I've had some warmer days to play with the car. I did some more tweaking with the MAF table and got pretty close to where I need to be. Today was warmer, so I tried SD tuning. Read plenty of the EQ<>VE Bluecat tutorials and some good write-ups on forums. I loaded in a tune file and logged my lamba error VE in a histogram. I then applied changes to my VE table in bluecat and loaded the new equations into HPT.

I noticed the car runs MUCH better on SD than MAF. The car used to break up when I would rev it around 2k and it would clear up when the rpm's got higher. I thought this was the nature of the cam. With SD, the engine runs so much smoother through the entire powerband up to 3500 rpm. I am just tuning with the car stationary now and free-revving to get close to where I need to be. I was doing the usual logging and adjusting the VE table by the results of my log. First I started reducing the whole table by 15%, then 20%, then another 15%. You could say it started out rich, lol. Then I got to a point where my logged error was so close that I was adjustngi groups of cell by less than 10%. I was doing real well. On my last adjustment, I even used the paste multiply feature in bluecat to multiply a section of my VE table by the cells from log histogram. Had the car running great. With the exception of one hiccup/misfire that occurred around 2k rpm, I had a very smooth log session. After the one misfire, the car still ran fine for the rest of the log session and the AFR numbers were good. I shut the car off to let it cool down for a few mins and grab a beer. I fired the car back up and all of a sudden I'm lean. No tuning was changed. It idles around 0.99-1.00 lambda error, but as soon as I give any throttle the wideband goes lean. I know at this point I was low on fuel in the fuel cell, so I ran to the gas station down the street and got 5 gals of E85 just to be safe. Filled the tank and fired it back up, still lean.

I'm scratching my head again guys Look over the log and tune files I've attached and tell me what I fucked up. It almost feels like I'm trying to hit a moving target, as if something is contributing the lean condition outside of what I did. I can't imagine anything mechanical would be a factor seeing that it just happened out of the blue with no change in running condition. I did verify that I am holding 58 psi solid at my FPR.

I am attaching quite a few tune files and each log to go with it, so it will take 2 posts.

[bikewiz]

oh and sorry about the incorrect dates on the tune files.

[bikewiz]

last log after filling tank and no change

[mowton]

Matt,

I forgot, what do you have the LC-1 output set to? If your tune is stating stoich is 9.7 then the LC-1 should be set the same.

With E85, I really would rather tune in Lambda. You don't always know what the fuel/alcohol mixture will be and lambda tuning puts all the unknowns to bed....stoich is 1 no matter what. The O2 sensors really only operate in Lambda anyways, the LC-1 and other instruments merely give you a means for representing it in an AFR value which you arbitrarily set based on the fuel you are using. Unless you know the specifics of the fuel, you are really just guessing.

Also as "belts and suspenders", I set all three P0101, 102 and 103 to fail first time and enable the ses so I can verify the PCM truly went into SD.

Will look more when I get home tonight

Ed M

[bikewiz]

Ed,

I'm pretty sure I've already got a similar setup to what you suggested. I attached my config file. I started with the basic PIDs that the Greg Banish video has you set up to plot MAF lambda error, and then I tweaked a few things from there. My AFR reading from the LC1 is scaled down to e85 stoich in the PIDs. My MAF lambda error is historgram #10.

Also, is there a way to fast forward the VCM Scanner playback? I know you can go up to 10 frames per seconds, but when I'm trying to see what my average lambda error was at the end of a 8 min log, it takes forever to get there.

I'll double check my MAF DTC settings tonight.

[bikewiz]

Ed,

I switched over to the config you made for me a while back and warmed the car up in the first log. I then change the tune to have P0101,102,103 to SES Enable and 0-MIL on first error. Fired the car up with the new tune and only ran it real quick because it was lean.

I'm not really sure what the issue is with the way my config is set up. The LC1 PID in HPT displays the reading in gasoline 14.7 scale (lambda*14.7). I have my PID set up to take the LC1 PID and divide by 14.7 which brings you back to lambda. Then, it multiplies by 9.7 to put me on the E85 scale. Whats wrong with this? I took a screenshot of my config on my computer at the end of playback from the last log from yesterday. You can see LC1 AFR and my calculated E85 AFR on the gauges.

[mowton]

Matt,

Personally, as far as E85, I would just go full Lambda. Just get the LC-1 to report Lambda and generate the error from your commanded Lambda PID.

Ed M

[one bad '010]

Hey bikewiz, I looked at your tune and noticed you still have your speed density portion failed.
Setting is at 400 rpm disable and 390 reenable. Up those numbers so the ecm knows to use speed density rather than maf. I did see maf failed to zero also. Hope this helps.

[bikewiz]

Matt,

Personally, as far as E85, I would just go full Lambda. Just get the LC-1 to report Lambda and generate the error from your commanded Lambda PID.

Ed M

Isn't that what I'm already doing in my config? On the lower left of the table display I have commanded lambda and lambda error listed above the LC-1 PID. I plot that lambda error in my histograms and use those error values as my factor for editing my VE table.

IIRC, we have been over this LC-1 debate already and came to the conclusion that changing the output values in the LC-1 only works for Innovate's logworks software. You can't alter what the LC-1 tells HPT. If you are running at lambda, HPT is going to display 14.7. You need to make custom PIDs (which I have) to alter that reading into whatever format you choose.

[mowton]

With the MAF failed and P0101 - 103 set to fail on 1st error, the Tune will only use VE (SD) for fueling. Matt is tuning the VE at this time.

Ed M

[bikewiz]

Hey bikewiz, I looked at your tune and noticed you still have your speed density portion failed.
Setting is at 400 rpm disable and 390 reenable. Up those numbers so the ecm knows to use speed density rather than maf. I did see maf failed to zero also. Hope this helps.

I'll look into that. I thought that just controlled when the ECM switched from blended MAF/SD to MAF only. My thinking was these values would be irrelevant if the MAF was failed in HPT.

[bikewiz]

Ed, you beat me to it! LOL

[bikewiz]

So any thought on why I went lean? My guess was if anything in the tune was commanding a lean condition, this would be apparent in my commanded lambda value.

I'm thinking maybe just press on with my VE table adjustments and increase the cells as needed to get my error to 1.00.