Problems with Auto VE Tuning

[thunder550]

Hey guys, got a question, looking for tips.

I started with my Nelson tune that I had and have been doing Auto VE tuning with my LC-1 wideband. I am running into problems. It's making my VE table jump from rich to lean on alternating tunes, and when I have something set right (like idle, had it perfect), the next round of AutoVE tuning will richen it up a lot, then the next one will lean it way out (goes from 8:1 to 20:1 on alternating tunes). It is doing this all across the board. I have run about 15 progressive tunes through, and the longer I go the worse the tables look. Any ideas?

Original tune (Nelson)
http://www.public.asu.edu/~cwessel/original.JPG

After 15 passes
http://www.public.asu.edu/~cwessel/tuned.JPG

[Blacky]

Is your wide band O2 sensor working correctly?
Is it calibrated?
Have you checked it against a known good wide band O2 sensor/controller?
Assumuing your narrow band O2 sensors are workig correctly, a "quick and dirty test" to check if your wideband is calibrated:

Log the pids: {GM.HO2S11} and {GM.HO2S12} and {GM.AFR} and your wide band O2 sensor AFR.

When the O2 PIDs start oscillating from rich to lean (high to low voltage), that means the engine is running as close to stoich as possible. Stoich is around 14.63 to 14.7 so if your wideband is also reading between 14.63 to 14.7 when your O2 sensors are oscillating then the wideband is probably ok.

Regards
Paul

[SinisterSS]

Are you truly in SD mode? Fuel trims still trying to do their job would cause havoc with VE tuning.

[TAQuickness]

Also, it looks like you may be smoothing the VE table. For now, don't smooth.

[thunder550]

Blacky,

I did not calibrate the WBO2 when I got it because it said it had already been calibrated. I will calibrate it as soon as I get a chance then check it with your "quick and dirty" method.

Sinister,

I am definitely in SD mode...I deliberately set my commanded AFR's to 14.7, and verified that stoich is set to 14.63. My STFT's are all 0 when I log them.


TAQuickness,

I have been smoothing....I will stop. Can it cause the AFR's to go that crazy?

One other thing I was thinking about doing is copying the LC-1 Base Efficiency table into Excel and limiting the amount of change to +- 2% per tune....will this help at all? That way I am not adding or subtracting some of the huge percentages that I have been (have seen up to 45% change per tune). Good idea or should I just let it do it's job? I am thinking that because some of the logged values are out of range of the LC-1 (below 8.xx AFR and above 21.xx AFR at times) it maxes out on the % change. Thoughts?

[TAQuickness]

Calibrating the WB is one of the easiest variables to eliminate right now.

While you are roughing the table in, smoothing may have you chasing your tail. Don't worry if your VE table has peaks and valleys. The BEN's will smooth the table out as you apply your changes

Also make sure you are applying the filters and hiding low cell counts as described in the AutoVE tutorial. You don't want to make tuning changes with throttle transients or other invalid data.

[caver]

I have observed this problem as well.
It is worse if you tune over a couple of days and the IAT varies a lot.
Try and keep the IAT at a reasonable level and complete the tune in one go.
I also keep an eye on my Ben values and dont change cells that were perfect but are now suddenly way to lean or to rich.

[SinisterSS]

Carver, I think you may be on to something there especially in thunder's application with his supercharged truck - intake air temps will vary from just over ambient to 130+ *F.

[thunder550]

That's a good point, I haven't been good about making sure I am logging after heat soaking the engine (IAT's around 120-130) and logging after cold start (I do wait for the ECT to reach at least 150, but IAT's take awhile to rise). I will make sure that I have the motor completely heat-soaked before I do anything as well. Thanks for the tips so far, keep em coming! I'm gonna take a long tuning trip tomorrow and try and do everything in one pass.

IAT's on my full heat-soaked motor are 120-140 while cruising, and I have seen them as high as 205 after a good WOT blast. I have a meth kit to cool things down, but I don't want to throw that into the mix until I get my base tune pretty close.

[Black02SS]

Sinister,

I am definitely in SD mode...I deliberately set my commanded AFR's to 14.7, and verified that stoich is set to 14.63. My STFT's are all 0 when I log them.



That sounds to me like you are in Open Loop and not SD. Are you using your MAF?

[thunder550]

MAF is disabled. I have it unplugged completely. I will go through the SD setup tutorial in the manual again though, just in case I missed something.

[thunder550]

Wow well I guess it helps to have a properly calibrated O2 sensor. I did a free-air cal this afternoon and got one really good AutoVE session in and one short one, table looks 100x better now, I'm pretty sure that was the only problem.

http://www.public.asu.edu/~cwessel/tuned1.jpg

[joecar]

MAF is disabled. I have it unplugged completely. I will go through the SD setup tutorial in the manual again though, just in case I missed something.
Make sure that it throws DTC P0103, that is a sign that it failed over to the VE table.

[TAQuickness]

Make sure that it throws DTC P0103, that is a sign that it failed over to the VE table.

good point. P0101, P0102, or P0103 will get you into SD mode for sure.

[thunder550]

OK I will re-enable those codes on the next flash and check to make sure that the PCM throws them.

[DrX]

OK I will re-enable those codes on the next flash and check to make sure that the PCM throws them.

They should already be set to non-emissions or you aren't in SD. Just leave them like that. If you are set up properly they will only show up if you use the scan tool to read the DTCs. They will not trigger the SES light.

I have a Dynojet WBC unit which has no recalibration feature. They all use the same sensors which are factory calibrated. Dynojet tells me that recal. is just compensating for a failing sensor and "not the best option". But that sensors "should last a long time" unless they are misused. In other words recalibration is unecessary??:???:

[thunder550]

I jsut went through the setup again and I have everything correct. I don't know why the O2 sensor wasn't calbrated but ever since I did that everything has looked much better. My charts are starting to look very normal. Thanks for the help everyone.

FWIW....my LC-1 manual said that I needed to recal the O2 sensor every couple of months I think.

[DrX]

Now I really don't know if I should trust my wideband considering that I also have had issues with AutoVE and other manufacturers are recommending regular recals!