Lamda P/I/D parameters per RPM

[Reverant]

I've been tasked to handle several drag cars that have 1600-2200cc injectors.

One of the requirements is to keep them in closed loop even during boost (a sensor failure doesn't worry us here because reasons).

The problem is that with such big injectors, I need to keep the P/I/D values low for idling/very low load, somewhat increase them during medium loads, and really crank them up during boost.

Apparently even the ancient AEM Series 1 has this.

The same is true for boost.

This requires a table so that we can set the P/I/D values per RPM. 8 cells would be great.

[muythaibxr]

Have you attempted to use the lambda delay table to slow down/speed up response based on RPM?

Just because other EMSs do something a certain way doesn't make their way correct, it just makes it their way. "Other people do it" isn't a good reason.

Also, people have enough trouble tuning PID without having to tune a curve for each gain.

[Reverant]

I'm going to try it for the sake of trying, but I'm not too optimistic about it.

What would the equivalent be for boost PID?

[muythaibxr]

So far for boost I've not seen anyone who couldn't tune the bias tables to make PID just need to hold boost.

Do you have logs showing that doesn't work?

Ken

[Reverant]

Tried the lamda delay table last night, it didn't work for idle. As soon as I reduced the PID values everything went smooth in literally 2 seconds. So yeah, I could use PID values per RPM.

[muythaibxr]

Tried the lamda delay table last night, it didn't work for idle. As soon as I reduced the PID values everything went smooth in literally 2 seconds. So yeah, I could use PID values per RPM.

Logs? settings? etc...

[muythaibxr]

Still looking for these.

[Reverant]

Patch for this attached. It replaces the standard P/I/D boxes (but doesn't remove them from the menus) with an additional window that has 8 RPM bins, so you can set each P/I/D gain per RPM. It also addes the calculated PID values to the outpc, I've added three gauges for these (under Calculations 3) and logging for them as well.

It made EGO control much better on my car (I have ID1000 injectors), as it allows a slower ramp at idle and doesn't cause oscillations, but a faster one once in boost at higher RPMs.

James, Ken, can we add this to the new firmware versions?

[muythaibxr]

No. You never provided logs or settings showing it was necessary. I am not going to add 24 more settings to the firmware without proof that they are needed.

[Reverant]

Please define "necessary". I don't want to go out doing two hours of datalogging boosted runs in closed loop, only to be turned down by saying "I don't think that this is much of a problem".

[Matt Cramer]

Back to back logs with and without your patch, and their corresponding MSQs, would make a very good starting point and should go a long way towards demonstrating your case.

[Reverant]

I understand what is being asked of me, I just don't want to go out bashing a car only to get turned down because of a subjective result, hence I would like an answer on what kind of a result the developers expect to see in a log, in order to approve this.