optimal EGO correction PID and cycles?

[AVP]

Has anyone tried what it is the optimal EGO PID (speed) and cycles of engine per correction, to avoid oscillations?

I am currently using 34 and 13 respectively, as instructed by the help files, but even with the IDLE settings off and adjusting the VE table as close to ideal as possible, there is still some oscillation, sometimes one way, sometimes the other.

What am i aiming to see while trying out figures on those 2 areas in order to get the finest result?

thanks

[tweak]

I used 40 and 14.. Step size 1 and 20% upper and lower limit. Think even less than 40 would be ok.
Number of engine cycles depends on how far downstream your sensor is. Obviously the closer to the engine, the better. I figured it out by looking carefully at the log. Just count the time after fuel cut overrun event, from start feeding fuel to when burned gas get visible on the λ readings.

As i mentioned in the WBo2 calibration topic, i'm not 100% sure about the WBo2 settings itself. I got less oscillation after playing with those settings, but i don't know if this is 'right'. I'm still in the progress of trying this.

[AVP]

thanks
let me know if you get anywhere with it.

vas

[tweak]

Hi AVP. Perhaps I should mention i went back to values close to factory default for the WBo2 itself. When you look at the theory for WBo2 sensor you will see that it's very sensitive around 1.00 lambda with sharply falling off over 1,3 and under 0,65. (Like a S shaped curve) Thats why you easily will get "jumping" at idle with slower gas stream and values close to 1,00.

Actually the round display firmware has a feature which handles this better than the ECU. It has a confidence filter, that takes out spikes. As far as i can see the ECU software does not have this feature.

[AVP]

why should a round 'display' have anything as a feature to control lamda better than the actual ECU.

that doesnt make sense! i dont use a display as i have the digital multi display from vems, but i would assume that lamda is carefully managed by the wideband from the ecu itself.

[tweak]

I have no good answer to this, but i see that the lambda values on the round display fluctuates less than on the VemsTune screen. Perhaps its a display issue only, and the ECU itself handles this without any problems. I might be real data you know

[AVP]

Or refresh rate of the display.

[tweak]

Might be, but next time i connect to the display i will take a printscreen and point out the feature i'm talking abt.
I just gave up worrying about this really, but it would be interesting to know more of course. Controlling a petrol engine is a rather complex affair and i'm not sure the VEMS guys has thought of it all yet

[GintsK]

Hi AVP. Perhaps I should mention i went back to values close to factory default for the WBo2 itself. When you look at the theory for WBo2 sensor you will see that it's very sensitive around 1.00 lambda with sharply falling off over 1,3 and under 0,65. (Like a S shaped curve) Thats why you easily will get "jumping" at idle with slower gas stream and values close to 1,00.

Actually the round display firmware has a feature which handles this better than the ECU. It has a confidence filter, that takes out spikes. As far as i can see the ECU software does not have this feature.

Can you give the reference to this info about S-shape curve. I can't rememberer anything alike WBO2 related.
Yes, sensor basement is narrowband sensor, but it is not used as output. Output from sensor is current necessary for pump. And current graph is pretty flat IIRC.

[tweak]

Hm you made me unsecure there. But i think i have the old type  LSU4 which has more limited bandwith. I also never seen lambda over 1,29 or under 0,65 on my setup. Why you will se jumping signal? A small fluctuation will give a more amplified change at 1,00 lambda than in the outskirts of sensor range. This depends of course how the signal is treated inside the ECU, and if this theory actually applies to our sensor. My thought is that you might need other filter set for the 1,00 area than out in the tails.

02 sensor theory:
http://performancetrends.com/wide_band_uego.htm

About the LSU4. Bandwith accuracy depends on which version you have.
http://www.ecotrons.com/Accurate_Lambda_Meter.html

Perhaps an upgrade to the 4.9 LSU is what we need.

[AVP]

apparently the 4.9LSU version seems faster and better response! maybe VEMS should incorporate it!!

[GintsK]

02 sensor theory:
http://performancetrends.com/wide_band_uego.htm

There is just some pieces of theory. Here is wide band oxygen sensor curve what relates to output. It is pump current neccesary to keep mixture stoich inside sensor. At same time it is input signal to sensor. And most important is how stable ecu handle it.

As you see nothing sharp at Lambda=1.