autotune vs AMC vs MLV

[vinister]

Hey guys,

So I've setup my VE manually, getting it 'close'. Then I turned on autotune and drove it for about 6 hours all in all, trying to stay in the bins long enough for the VE's to be adjusted down until the correction was close to 100%. After that I turned on AMC, and drove for about a week, a couple hours a night. After all that, I took a nice long log and loaded it up in MegaLogViewer. I did the analyze, and it found a bunch of points with pretty solid hits, so I loaded it up.

I have done the above procedure a couple/few times now. It seems autotune, AMC, and MLV just can't agree on what the VE entries should be.

- autotune gets to a point in each cell where it will go up and down by 1%, back and forth. That means its hitting my target spot on.... right?

- MLV wants to richen up my main torque areas by a few points. I have tried using a delay of 1 and 2, but have left all the other options alone. There does not seem to be too much difference between delays 1 and 2.

- AMC... well, who knows if its even working? How do we know? There are no indications, and the algorithm used doesn't get talked about much.

I have reduced the EGO authority from 15, to 10, and now to 5, to reduce the 'surging' effect that is felt as the PW is corrected up and down. Is there a way to make AMC and/or autotune reduce its step size to less than 1%? I have a feeling that since it is always going from +1 to -1, back and forth, that the correct value must be at +0.5.

What is the best way to set it up as a semi-permanent system? EGO off, AMC on at 5%? I'd like to turn the EGO correction off to stop the surging, but I'd still like the computer to notice the mixture is off and adjust the VE for next time.

Do any of us ever actually stop tuning, and just drive?

[devastator]

It sounds like you've got it dialed pretty darned well. The only question I can answer is the last one:

Do any of us ever actually stop tuning, and just drive?

No, not ever.

[Buick pwr'd Brit]

vinister
I am at the same point as you. Take it for a drive, run data logs set to very hard. Sometimes it will want 1% richer, other times it will want 1% leaner or no change. What I do is look at the 02 data, it appears to be pretty darn close. So I just go with the setting that it runs best at, (usaully the richer #) and call it good!
I dont think you can change it by .5%. It would be cool if you could, as 1% change makes a noticable difference at the lower end of my table.
For the O2 correction surge, mine was most noticeable at very light throttle low speed cruise. This is the same area of the table that is sensitive to a 1% change, and where I just leave it at a number that the engine likes the best. To keep it from correction surging, I set it down to 3%.

Bill

[vinister]

Great post! thanks, makes me feel better. I guess I'll just leave AMC on with a low authority, and stop doing logs for now.

[SQLGUY]

For the autotune, there are a couple of things to consider:

1. There's about a one second time lag for a WB EGO
2. Autotune's just trying to meet your requested AFR

I don't know what MLV is basing its values on, but it might not agree with the AFR you've requested for a particular bin.

How does it feel, and how do the plugs look?

[vinister]

Really one second? that seems like a really long time considering its only a few inches from the cylinder.

All 3 methods are trying to reach my target AFR (Autotune, AMC, MLV), I'm just wondering why they don't seem to agree on a perfect number with each other.

And on a related note... since my O2 sensor is only attached to 2 of the cylinders (dual downpipe), how does it deal with the pulses it is getting, and is that why the sensor signal swings back and forth?

The car feels great, the only thing is my ignition map isn't very smooth, so under load I can feel the transitions as the advance changes. I find the advance is the hardest to tune because we don't get any feedback for it. I'll be hooking up the knock sensors eventually, but even that doesn't tell you much other than "too much!". Theres a couple other megasquirt dudes here in Vancouver, I'm thinking of organizing a dyno day so we can have some actual professional tuning sessions.

I don't know how the plugs look ATM because the manifold has to come off to access them (unfortunately).

[chevelle]

I know this is a pretty old post but wanted to share a finding of mine relating to "02 surge". I recently installed a wideband 02 sensor on my engine which I also recently converted from a 2 barrel TBI to a 4 barrel TBI. Not very long after installing the new wideband sensor I turned on EGO Control. I noticed that the engine had what seemed like "02 surge" especially at light loads where the AFR table was tuned pretty lean.

Upon further examination of the datalog however I found that the true cause of the surge was that I had the battery voltage correction set to .3 ms per volt. This worked fine on the 2 barrel but had way too much response in fueling with the 4 barrel. EGO Correction simply made the problem worse where the battery voltage change would change the pulse width, thus changing the AFR then EGO control trying to correct it. After setting BVC to .1 this problem is gone and the AFR seems much more stable.

Another symptom was that just after hot start before reving the engine it would run pig rich with the EGO control set at only 15% +/- being unable to correct it. This is because before reving the engine the alternator fields wouldn't be excited and the battery voltage would only be around 11.8. After reving the engine once the voltage would go up to 14 and the problem went away. Again after setting BVC to .1 that problem is gone.

I got to thinking this problem may also exist in some scenarios where BVC is set to low?

[142 guy]

Interesting observation on the effect of reduced voltage at start up. I assume you are talking about the sensitivity of the injector offset (opening time) to battery voltage? If your voltage sensitivity was too large, then yes it would be increasing the offset more than it needed to which would result in larger than required injector pulse widths. If you haven't tried this already, theoretically, it should be possible to verify this effect by logging voltage during a start. If you see the injector pulse width drop as the voltage recovers and the AFRs improve, you have a pretty clear link. In reality, might be pretty difficult to see this effect because of all the other stuff happening at the same time.

Be aware that the injector offset and voltage sensitivity of the injector offset is highly non linear. The average sensitivities of the particular injectors I am using are:

16 - 14 v .03 msec/volt
14 - 12 v .12 msec/volt
12 - 10 v .2 msec / volt

MS uses 13.2 volts as the base for correcting injector offset so I use a sensitivity of 0.12 msec/volt.