miatauser884

iTrader: (11)

I decided to revisit some of my older settings. One of the oldest that I set was the EGO PID settings. I'm using:

P: 1
I: 110
D: 0

AEM linear gauge type

It seems like it works alright, but I wondered how this compares to what others are running?

miatauser884

iTrader: (11)

Ok, if no one wants to post there settings, then maybe chime in on the following. These PID settings were given to me by DIYautotune. All of my other settings are custom to my setup, and now I want to play with the ego PID settings.

My thought is to enable the ego correction at idle and watch it oscillate around the AFR target. Increase the I-term until it oscillates very tightly around the AFR target, then increase the p-term and see if it gets better. Ideally this would be done at cruise with a passenger tuning the laptop, but I am solo. Does this sound like reasonable logic applied to tuning this setting?

Braineack

iTrader: (62)

I havent tried EGO PID yet.

miatauser884

iTrader: (11)

Really??? I've never not used it. huh. I guess I will try my tuning method and see where I end up. Then again, I guess if you have a well tuned VE table the ego correction isn't that important. I've been autotuning recently, and my table usually goes back and forth 1 point each session.

y8s

iTrader: (8)

if your I is causing oscillation, it's too high.

basically:
I makes sure you hit target. the longer you're away from the target, the stronger it wants to hit the target.

P is how strongly attracted to the target you are. the bigger the deviation from target, the stronger P gets

D is like the bouncer. when a big P drives the value toward the target really fast, D puts on the brakes and slows it down as it approaches the target.

So in short, to tune EGO (which is a pretty slow system in my opinion), you should run as much I as you can without going beyond the stability limit... so if you see it oscillate, back off 5%. Then crank up P until the correction overshoots from a transient event (blip of the pedal eg) and back it down so there is almost no overshoot.

leave D at zero because it's too hard to tune. D is generally used to allow a higher P value and improve response time. With a well tuned PI and D system, you can tune P to a value that would overshoot and use D to slam on the brakes at the last second. You reach your target faster (though maybe with minor overshoot and oscillation).

Ben

iTrader: (33)

The values I usually use have a large I, small P, and 0 D with 5% authority. A well tuned fuel map needs little correction.

miatauser884

iTrader: (11)

I ended up with I=95 and P=1 using the above mentioned tuning method. I'm sure tuning the ego PID at idle is not the best method, but I did notice increased stability.

While at idle if you can create a minor instability with headlights, clutch in, etc. The afr will start to oscillate. The ego kicks in to try and counter the oscillations, and shows up as a counter oscillation. I adjusted the I-term until the opposing amplitudes of the ego and afr were at a minimum. Ideal would be two flat lines.

In my opinion this means that the algorithm reacted quickly with the correct amount of compensation. Hopefully tuning this at idle will translate to a good tune when the AFR is changing more rapidly. One can only hope.

muythaibxr

I do have some improvements that should in theory allow for a larger P term on the way. They are ms3-only however. I am essentially going to add a looptime table so you can use observed sensor response delay to slow down/speed up correction.

y8s: since you're looking, looks like single cam VVT will be able to make it into 1.1 after all. I was pushing for 2.0 without a 1.1 but James went on a coding spree and did the part of VVT I didn't want to do along with a bunch of other stuff, so no good reason not to do it for controlling a single cam now. I will post a full revised roadmap at msextra at some point soon.

Ken

Braineack

iTrader: (62)

eh. work on stabilizing the timing advance and the oscillation will go away. I found that ego at idle will ruin a perfectly stable idle for no good reason because it fights with everything else.

miatauser884

iTrader: (11)

You are absolutely right. EGO correction is a horible idea for idle. This may only be true for an untuend ego PID algorithm. I must say that my idle was back to its normal stable self after I tuned the PID settings.

muythaibxr:

How much more MS3 stuff will be back ported to MS2? I feel like I made a mistake getting the MS2 because (unknowingly at the time) it was at the end of its progression. It would be nice to have the MS3 idle control table.