Tried out auto tune with TunerStudio MS

[Derek]

I tried it out yesterday and was quite impressed. It's under a tab called "VE Analyze Live". There doesn't seem to be a manual yet but it's pretty intuitive. I was running a little rich at idle. I reset my AFR table and turned on auto tune. Slowly but surely it dialed it right in to the AFR I requested. No fuss no muss.

 

Derek

[olie05]

how did it compare to the autotune feature in megatune?

[naviathan]

Unless it's improved any in the last year I've always heard the MS autotune was very unreliable.

[olie05]

I had a decent experience using it in the past. I think the problem is that it has too many constants that can ruin the stability of it... and maybe most people don't take the time to get those right.

 

Does tunerstudio have presets for different types of o2 sensors?

[Zmanco]

I've been using it for a while and except for the name, I'd say there is no similarity to the old MT autotune - it's that much better.

 

But, and there's always a "but", it only works well if it has accurate Lambda delay values. The defaults were way off for my setup (L6 with stock turbo exhaust manifold and t3/t4o3 turbo with O2 sensor a few inches away on the downpipe). Here's what I've settled on after a lot of trial and error.

 

The developer says he's going to have a way for it to identify events that should trigger AFR changes and measure the delay so the table will be populated with values based on real data, not guesses. I hope he can pull that one off because that is the missing piece in my mind.

 

I'm using it instead of MLV from now on because the lambda delays are so much more accurate than simply using 1 single value as MLV does. My advice is to NOT check the "update controller" button and run it with "cell change resistance" set to very hard and after half an hour or more of driving, look at what it's recommending before you accept it.

[Zmanco]

BTW, it's not a replacement for reviewing datalogs to see what's really going on - but it's a much better tuning aid than what we had before.

[Derek]

Hi Daniel

How did you determine your lag. By looking at the data log?

I sometimes think tuning this thing has been harder than building it!

Thanks

Derek

[Zmanco]

Hi Derek, I had the advantage of starting with a tune that was already pretty good. I tried measuring delays but that doesn't help much in the corners of the VE table as it's hard to cause an event in those regions. I knew that it should be a longer delay at lower revs and smaller loads, so played around with the lamda delay values so that the curves it recommended made sense. For example, when the values for low rpm and low load were too short it was recommending to increase those VE values much more than common sense said it should. For the high rpm high load corner I adjusted delays until what it recommended got to within a few tenths of the target AFR.

 

With MLV I had previously done a 2 stage pass with VE Analyzer where I used a delay of 8 for everything <= 2500 rpm and delay of 3 for everything >2500. It was a poor compromise and a hassle. FWIW, I think Phil the developer is going to add a lambda delay table to MLV for offline datalog analysis.

 

Actually, the tuning isn't so hard, it's tuning the tools that's time consuming

[cygnusx1]

I am not sure if this is fact but maps tuned with TS don't work properly with MT and vice versa. I ran into trouble doing this. Can anyone confirm or bust this myth?

[Zmanco]

Some of the newer features being added to 2.x and 3.x MS/Extra are only supported on TS and not on MT. But I haven't heard anything about the maps themselves being different between TS and MT. I'm having a hard time imaging how that could happen as that would imply that the firmware running on MS itself would behave differently when talking to MT vs. TS and I'm pretty sure that's not the case.