Page 1 of 1

First Time With Autotune--Oscillations

Posted: Wed Aug 09, 2006 9:13 pm
by scowman
So I got AutoTune to run by changing the parameters in the config.ini file.

And I also got the O2 sensor to work properly by getting the software bug out (thanks to Lance:))

Now when it idles, it oscillates between rich and lean. That is, the O2 sensor pegs between + and -. It also moves between squares on the VE table. I cannot get it to sit still.

When I turn auto turn on, it will tune one of the squares, but not the other square it is oscillating to. Very strange.

Any ideas?

How do I attach a datalog?

Thanks

Posted: Sun Aug 13, 2006 11:29 am
by scowman
Anyone?

Posted: Sun Aug 13, 2006 11:47 am
by Bernard Fife
scowman,

If this was mine, to reduce the autotune oscillations I'd reduce the proprtional gain and reduce the 'lumpiness' in the autotune settings. I'd also make sure the step size in the EGO settings is 1%, and make the ignition events per step at least twice the number of cylinders.

There's more on autotune here:
http://www.megamanual.com/megatune.htm#autotune

However, a narrow band sensor will not normally 'sit still' - the outuput curve slope is so steep near stoichiometric that a very small change in mixture (which is happening all the time) creates a large change in output, and the output swings back and forth. So for a narrow band, there are three modes, in effect:

steady and low = lean

oscillating = stoichiometric (or potentially some problems, like a misfire, so oscillating means stoich only in the absence of other problems)

steady and high = rich

Lance.

Posted: Sun Aug 13, 2006 4:05 pm
by scowman
Lance wrote:scowman,

If this was mine, to reduce the autotune oscillations I'd reduce the proprtional gain and reduce the 'lumpiness' in the autotune settings. I'd also make sure the step size in the EGO settings is 1%, and make the ignition events per step at least twice the number of cylinders.

There's more on autotune here:
http://www.megamanual.com/megatune.htm#autotune

However, a narrow band sensor will not normally 'sit still' - the outuput curve slope is so steep near stoichiometric that a very small change in mixture (which is happening all the time) creates a large change in output, and the output swings back and forth. So for a narrow band, there are three modes, in effect:

steady and low = lean

oscillating = stoichiometric (or potentially some problems, like a misfire, so oscillating means stoich only in the absence of other problems)

steady and high = rich

Lance.
Thanks for the reply Lance!

Maybe I have a misfire. I will change all the plugs on that side and see if it gets any better.

I guess a wide band is the answer.