3.4DOHC VE AutoTuner

[ThirtyTwo]

Ok I already posted this on fiero.com but I figured I would post it here too.

I know a little C++ and I started making a program that takes a log file from freescan and fine tunes the main VE table in a user specified .bin file. Its able to tune the table to a user specified BLM value(default 12 and it also gives out statistics and the strength of the log file.

I started making it mainly for myself but I was wondering if anybody else would have any interest in it? If anybody is good with C++ and wants to help that would be cool, I have the code part done but I need to create the GUI part and I dont know much on how to do that. But its about 90% done so far and everything works with the reading of the log file and editing of the .bin file.

Also does anybody have any ideas for additional features?

[sil]

this sounds like a good idea, i think people will gobble this up lol
i would be intrested in testing this out for you, not proficient in c++
i have c++ tools and stuff but i need to learn lol

[Fierobsessed]

You want to write a program that would really kickass... You would have it re program the ECM to really learn the VE "feel" of the entire engines range. The BLM table is 4 X 4 and the VE table is 14 X 17 or so... there is a couple of things you can do. You could write a program that learns the VE somehow. Or you can have the program change the BLM table's cell limits, to learn different areas of the VE table. As my quad 4's program is set from the factory, The BLM table ONLY focuses on what happens between 1200-2400 at high vaccum levels between 15 and 40 KPA. Yes, it does adjust the entire VE range, but it uses the same BLM cell to compensate frrom 2400 RPM's to redline, and 35 KPA to WOT. Not exactly what I call "fine tuned".

[ThirtyTwo]

I thought it had more cells than that, what if I used the average Integrator value? The BLM is probably an average of the INT anyway so it seems like it would work. And the INT is always changing

I think I will eventually make it compatable with wideband.

[ThirtyTwo]

I just took a look at the stock 3.4 .bin and the BLM updates between every .5seconds to 4seconds, so even though there are only 4 cells I think it will still update fast enough to represent each running condition.

What im going to do is make the measuring value user adjustable between BLM values, INT values, 02 Volts, or Wideband volts

edit:after testing with 02 volts as a measuring value it made changes that were way off, I think you would need a very strong logfile to be able to use this method

Here are the changes it did to my stock .bin after I took a couple mile log. These changes are based on BLM values


Here are the changes the same log did with INT values

[napatiger]

I wish I understood what the hell you were talking about.....

[ThirtyTwo]

I wish I understood what the hell you were talking about.....

what do you want to know? My first post explains the basics, the program takes a log that you make while driving your car around and then fine tunes the fuel curve on your computer

[ThirtyTwo]

Time for a screenshot, I should be done in a couple days

[SappySE107]

If you are making it for the DF code, that is also 93 3.1 code:P

There are 3 cells, unlike the F bodies that have 16. You will need to add the base VE and the 3D VE table to get a true VE % for each RPM/MAP reading range.

Don't use the BLM, use the INT. If possible, have 2 modes for your program, rough and fine. BLM is rough, INT with BLM locked at 128 is fine. You could have it automatically lock the BLM for a fine tune mode and then revert back to unlocked if you want. I leave mine locked after I get it tuned and it runs nice and consistant. 5 speed cars are much tougher to deal with for locked BLM though so if you want to make it idiot proof we can make your program a lot more complicated to setup:P

[GTP_MuNkY]

I wish I understood what the hell you were talking about.....

I'll second that... you all might as well be talking in binary...

What year range of 3.4s would this work on??
Would you need C++ to run the program??

Jamie

[SappySE107]

I would assume 91-93. The 94-95 code would be nice but I am not aware of a 100% accurate/complete disassembly of the code. Some people can read that code and decipher what its doing. Im not one of those people.

[ThirtyTwo]

If you are making it for the DF code, that is also 93 3.1 code:P

There are 3 cells, unlike the F bodies that have 16. You will need to add the base VE and the 3D VE table to get a true VE % for each RPM/MAP reading range.

Don't use the BLM, use the INT. If possible, have 2 modes for your program, rough and fine. BLM is rough, INT with BLM locked at 128 is fine. You could have it automatically lock the BLM for a fine tune mode and then revert back to unlocked if you want. I leave mine locked after I get it tuned and it runs nice and consistant. 5 speed cars are much tougher to deal with for locked BLM though so if you want to make it idiot proof we can make your program a lot more complicated to setup:P

Crap I didnt even think about the base VE table, so the base values are just added to the Main VE values to get the final VE?

I could put the rough and fine tune options in, but I dont get what you mean by "locking in the BLM". I have noticed that the INT values sometimes spike and you really need a solid logfile to get good results so maybe I could just have the program throw out the outliers but the user would have to make a strong logfile.

I think the program will need a lot of testing to get it dialed in and see what works best but we can get it there eventually. It would be great if you could input on the best tuning methods since I am slightly new to tuning these engines.

[ThirtyTwo]

I'll second that... you all might as well be talking in binary...

What year range of 3.4s would this work on??
Would you need C++ to run the program??

Jamie

As sappy said its for 91-93 TDCs

Its a standalone program like notepad or paint. So you dont need anything else. It doesnt have to install or anything though

[bszopi]

I am far from a tuning expert, but the few times I messed with it (or when I was watching Ben), there is a way to set the BLM to 128 so that the ECM will always see it as 128, and it will adjust the INT value as required to maintain the curves. Ben can probably explain it better though...

[SappySE107]

Yup, its simple. You set the max and min BLM values to 128 so it uses the INT only for tuning.

For the VE, you add the 1200 RPM base VE to every value in the 3D lookup table. I had to map out everything when I was learning how to tune. Once you adust for the base you will see that graph for the 3D table isn't what you thought it was. The 91-93 DOHC has the same base VE for quite a few RPM ranges though.

Honestly, you won't be able to make an "auto tuner" but you can make something that is easy to use for newbies and those with near stock setups. I would have it setup to take a log with the stock chip, check for BLM less than 120 or greater than 136. If true, then use rough tuning, which would keep the BLM code stock and do the normal tuning. I wouldn't do the spark tuning until the fueling is dialed in a little better. Only do spark retard correction after the fine tune. Since this will be for near stock it won't be so bad but put a disclaimer. Hell, you may as well have a book with it:P

It would be trick if you could convert to MPG from the injector load vs rpm vs injector constant. Then you could have it auto tune spark/fuel for economy.