Auto Tune, a better way?
 

Intrigued by recent posts claiming great success with the autotune feature (in Boost), I decided to give it another try. I started a new project using my previous one as a base. My map in the 100 kpa and lower is pretty spot on. So I wanted to let autotune help me with the boost side of the fuel map.
Two hour later, roasted brakes and half a tank of gas later, I had a shit map. Nothing useful IMO. Just wacko numbers that were unbelievable and drive like poo.
My method :
1. Find a long steep grade to increase load and decrease brake abuse.
2. Start autotune (easy) while in 3rd gear, 2000 rpm use the rake and throttle to sweep through the manifold pressures from 100 up to 250.
3. Repeat for 2500, 3000, 3500, 4000, 4500 and 5000 rpm, saving the tune after every sweep.

Results= garbage. I can get results much faster and more accurately with 20 runs and logs, making adjustment manually.

What are you guys doing different? Do I haz the dumb again?

This method has worked fine for me.
Obviously I do back and do some smoothing and adjusting manually. Adjust cells in funny areas of the map by hand too.

MS2 or MS3? Is your AFR signal clean?

I've never done anything special with auto tune, just let it run while I drive making sure to slowly ramp to different throttle positions, and it's yielded nothing but good results. As Dillon said, it requires some smoothing afterwards, but should be a usable map before smoothing.

Works fine for me too, just to rough it in and get the general shape and close to target values.
Too far off and I manually help it. Real close and I start smoothing it by hand. It's good but it's still not magical, needs a lot of assistance.

Afr signal is clean and believable. Maybe I was just expecting too much. I don't know. I just done belive that 15 psi boost would require less or the same fuel as 90 kpa. I have a huge block of numbers I don't believe. Not just a couple sporadic ones.

Do u you guys use the afr target safety at the same time? If so what was the allowable differential? I was at .2 but couldn't even get into boost area long enough. So I changed it to .5

Did you turn off AE? If you did it was a mistake. Autotune doesn't make changes until AE has tapered off so it doesn't interfere with the tune. But turning AE off can do weird things to the autotuning.

All on.

So I guess it sounds like I was doing it right, I mean it's not too complicated. Just isn't netting results. Going back to logs and manual tune process.

For the under 100kpa cells (still N/A....) I can bang out a useable table in an hour or so with autotune. Not perfect, but ok. Duno what is up with yours.

I was wondering if that was the case or not, since it has worked well in the non boosted cells for me with AE on.

I do this on the regular. Sooner or later I'm going to go to jail if I don't sort it out.

Under 100 kpa works fine. Same thing. Got that roughed out in an hour. Dialed in perfect in a few drives. Turbo says no.

I still have a good map, all up to 250 kpa. I was just trying the autotune for boost, just to see. Looks nothing like my known good map.

boosting is the easiest for it to tune, since the load is constant through an rpm range. do 4th gear pulls from as low of an rpm as you can.

I take it for what it's worth, just another tool to utilize in the process of manual tuning. It's VE Analyze Live. It still needs help in taking it's suggested numbers and making them copacetic with the rest of the map. It's far from magic or auto-tune in my opinion. I've gotten a driveable map from VEAL, but never a great/solid map without manual smoothing and corrections.

VEAL helps make tuning a lazy affair. Just slowly increased boost and communte to work / do a canyon run and let VEAL get the map in the ballpark. In doing so I keep an eye peeled to my AFR's and made sure they weren't waaaaaay off. This gave me a decent shape of the graph and I manually went in, smoothed and tweaked a bit, did some logs to confirm that the map works properly. Allows me to more or less just drive the car normally as opposed to doing specific logged runs.

^^^^^this. Maybe I need to start at low boost, and then keep raising it, rather than trying to get it all at once. I don't know. I feel I haz the dumb on this. It's just wierd. I auto tuned and then manually adjusted my 0-100 kpa, and I can stay within .2 of afr target at almost zero ergo control. Like I said, I'm not dead in the water, this was just an experiment.

What do you have your cell change resistance set to. If you have it set to very easy it will make changes that are way too big.

Another possible root cause.

I agree with Brain. Don't try to run columns, run rows. I don't know why, I only know that's how I did it and it worked well