Everything there is to know about MegaTune Auto-Tune

Natedogg · Post by **Natedogg** » Wed Sep 21, 2005 6:17 pm

I got the 025x1 code loaded and I started using autotune. It works great. No lean/rich problems.

Does it keep running when you close the VE tuning window though?

Nate

Post by **efahl** » Thu Sep 22, 2005 7:58 am

From a PM. (I urge you guys to ask this stuff on the public forums, it's good info and virtually every PM I get is not a "dumb question," like this one, it's good useful stuff. Don't be shy!)

Eric, in the autotune parameters code, if the EGO correction step is set to 1%, controller authority is set to 15% and the proportional gain is set to .5 - will it try to adjust based on the step or the authority size from the changes i am seeing it adjusts on the step size. I also am not 100% sure it pays attention to the lumpiness as i have seen some points exceed 5% of their neighbors.

The EGO correction step doesn't come into play at all, that's only internal to the EGO algorithm itself and dictates how rapidly it moves when correcting. The auto-tune step is determined by the current EGO correction (which is limited by the EGO controller authority), and the proportional gain, so only those two values really play a role in computing how big the auto-tune step will be. So, things proceed like this:

1) The EGO algorithm uses whatever means necessary to come up with a number, this is limited to 100% +/- EGO authority with the familiar algorithms.

2) The auto-tune algorithm uses the EGO correction value from 1, multiplies its difference from 100% by the gain to get the adjustment proportion. Gain is 0.5, EGO correction is 110%, then the VE adjustment will be 5%.

Remember that lumpiness is dynamic, the current point looks at its neighbors each time an adjustment is made, so two adjacent points could "climb each other" if that's what the corrections are indicating... Could this explain what's going on?

Eric

McNugget · Post by **McNugget** » Thu Sep 22, 2005 8:07 am

Hi there,
I have the same software version installed and it works perfectly! I like autotune

The only thing I have to say is that I miss a solution for enableing EGO-correction at full throttle. I have no idea if it runs 50% to rich... or whatever. Ego correction runs until 111mi/h but above EGO correction disables and the engine stops talking to me. I have to guess whats right for the system.

Sorry for my english, I have no idea how to explain what I want so say

Any solutions for this problem?

Cheers,
Timo

Natedogg · Post by **Natedogg** » Thu Sep 22, 2005 9:45 am

I figured out my own question, so never mind.

It would be cool though to have an autocorrection feature like this, but it doesn't make permanent changes to the VE. It would use a base VE, but make adjustments as it goes for changes in conditions. When the power is cut, it goes back to the base VE. Then when the car is started next time it starts all over again. It would require a good base tune though.

This would be more of a feature in the MSnS code though, so I guess I should post it there.

Nate

whittlebeast · Post by **whittlebeast** » Thu Sep 22, 2005 9:48 am

This won't happen in Extra There is just not enough room for the code. MS2 has a version of this and lots more in the works.

Natedogg · Post by **Natedogg** » Thu Sep 22, 2005 9:57 am

Cool. I posted it in the MSnS forum anyway. I figured at some point you would run out of room, but like I've said, I'm no code guy so I don't know how much space it would take.

Nate

whittlebeast · Post by **whittlebeast** » Thu Sep 22, 2005 10:17 am

The amount of room left for new features in extra is ZERO They are now sharing bits for start up features with a feature in traction control as they will never happen at the same time. Thatss how full its getting.

MiseryQ · Post by **MiseryQ** » Thu Sep 22, 2005 3:41 pm

I don't know If I mentioned this before, I planned on doing more testing but I dropped my laptop.
I've notice a 28% in adjacent bins with a 5% lump setting.

Should've mentioned this was with VERY aggresive AutoTune settings though.
Normal "fine tuning" was great.

Post by **efahl** » Fri Sep 23, 2005 5:18 am

McNugget wrote:The only thing I have to say is that I miss a solution for enableing EGO-correction at full throttle. I have no idea if it runs 50% to rich... or whatever. Ego correction runs until 111mi/h but above EGO correction disables and the engine stops talking to me. I have to guess whats right for the system.

Timo,

What firmware are you running? Do you have the EGO correction parameters set to limit the RPM for EGO correction? Do you have a wideband lambda controller and sensor? (This is really necessary for doing any medium- to high-load auto-tuning.)

Eric

whittlebeast · Post by **whittlebeast** » Fri Sep 23, 2005 6:11 am

I was talking to Bruce yesterday and he has found that there is a fair amount of hystresis in many wide bands. Some are filtered to slow the response time.

AW