PID Explanation and Tuning

[Ben-S]

Agree about the duty limit being too low (really no reason it shouldn't be your solenoids max allowable) and the D being to high but you can achieve more accurate control than that with just duty cycle control which would lead me to believe not enough time was spent on tuning the "normal position" table it's always best to tune boost based 100% on duty cycle then add your PID parameters as there will be less "wind up"

[Brunel Racing]

Is there a way of limiting the Duty Cycle output? I'm trying to electronically control the fuel pressure/fuel pump using function 138 and an M800, but the electronics I've got (OEM stuff to convert the PWM into a variable voltage) doesn't seem to recognise more than ~80% duty.

Cheers,

James

[Scott@FP]

Looks like you may have a bit too much D once you get to 5th. You can see the duty cycle drop right off which gives you the drop in boost pressure. You then see the duty cycle go back up as the boost undershoots but the boost takes ages to recover, the duty looks like it has flat lined at the max allowable (about 65-ish%). You could need to have a bit more allowable duty cycle in fifth gear.

If you are using version 3 M800 software you can set everything as a table. Perhaps you need a gear position input into the ECU so you can tweak everything based on gear.

What do you think of a Derivative table using Boost Aim Error and EMAP as axes?

Or for that matter P,I or D and BAEr on X? When aim error is low shouldn't the factors used to correct it also be lower values?

[jaytourist]

MarkMc, thank's for the info. il check your attached document for my knowledge. tnx & keep up the good work. :)

[mr2andy]

Absolutely enjoy the reading! Great explanation on PID! Thanks Mark~

[jjb]

Hi - I just joined the group and I'm excited to start using my M880 & SDL on my Ducati 1098R superbike.

Anyway, in researching PID controllers, I came across the following two references. In addition to Mark's excellent pragmatic explanation, these give some background from control theory & mathematical perspectives. I hope they help some...

http://en.wikipedia.org/wiki/PID_controller
http://www.engin.umich.edu/group/ctm/PID/PID.html

[MarkMc]

I find that the best way to deal with PID is to have the basics, like in my document, and then just do it, practical application is the best way to learn.

For anything the best thing to do is get your Feed Forward (Normal Position, etc.) table as spot on as you can, for boost this may mean Gear position as an axis, or even aim boost. I really don't thin kthe boost aim error is the best thing to do because you are basically saying to change the gain based on the error when all your PID outputs are based on that anyway.