How can I compute new pid coeff from old ones based on

[JulienDevillers]

Hello

I'm driving the pwm of an electric motor with a pid controler.
It works pretty well.

After some testings, I realized that the fact that the motor is powered by a battery
makes the system behave differently depending on the battery voltage.

I will change the system in order to make the pid system control the voltage provided to the motor instead of the pwm ; another
system will do the conversion V -> PWM based on the battery voltage.
Therefore, I expect the same behaviour whatever the battery voltage (if of course, the pid does not ask for a voltage
more than battery voltage)

Concretely, PID will not control values from 0 to 20k (pwm period) but 0 to 8400 (mV).

My question is:
Is there a mathematical way to compute the new PID coefficients from the old ones to get the same behaviour with the
new actuator without having to go again to a new pid tuning session?

Thanks.

[br3ttb]

So an output range of of 0-20k is now 0-8.4k, and the rest of the loop is equivalent? If so, then the controller is now 2.4x (20/8.4) more powerful. That is, an output change of 1 means 2.4x more process change than it used to. As a result, dividing the 3 coefficients by 2.4 should make the pid output with the same oomph as before the conversion

…

On Sun, Nov 10, 2024, 8:26 AM Julien Devillers ***@***.***> wrote: Hello I'm driving the pwm of an electric motor with a pid controler. It works pretty well. After some testings, I realized that the fact that the motor is powered by a battery makes the system behave differently depending on the battery voltage. I will change the system in order to make the pid system control the voltage provided to the motor instead of the pwm ; another system will do the conversion V -> PWM based on the battery voltage. Therefore, I expect the same behaviour whatever the battery voltage (if of course, the pid does not ask for a voltage more than battery voltage) Concretely, PID will not control values from 0 to 20k (pwm period) but 0 to 8400 (mV). My question is: Is there a mathematical way to compute the new PID coefficients from the old ones to get the same behaviour with the new actuator without having to go again to a new pid tuning session? Thanks. — Reply to this email directly, view it on GitHub <#147>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACYX4Q47M56QSH5TXQ6VDTZ75NHZAVCNFSM6AAAAABRQG4JYWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGY2DOMRSHEZTQNI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[JulienDevillers]

Hi Brett, Thanks for your answer, it was obvious for p, but a confirmation for i and d was useful! Le dim. 17 nov. 2024 à 13:35, br3ttb ***@***.***> a écrit :

…

So an output range of of 0-20k is now 0-8.4k, and the rest of the loop is equivalent? If so, then the controller is now 2.4x (20/8.4) more powerful. That is, an output change of 1 means 2.4x more process change than it used to. As a result, dividing the 3 coefficients by 2.4 should make the pid output with the same oomph as before the conversion On Sun, Nov 10, 2024, 8:26 AM Julien Devillers ***@***.***> wrote: > Hello > > I'm driving the pwm of an electric motor with a pid controler. > It works pretty well. > > After some testings, I realized that the fact that the motor is powered by > a battery > makes the system behave differently depending on the battery voltage. > > I will change the system in order to make the pid system control the > voltage provided to the motor instead of the pwm ; another > system will do the conversion V -> PWM based on the battery voltage. > Therefore, I expect the same behaviour whatever the battery voltage (if of > course, the pid does not ask for a voltage > more than battery voltage) > > Concretely, PID will not control values from 0 to 20k (pwm period) but 0 > to 8400 (mV). > > My question is: > Is there a mathematical way to compute the new PID coefficients from the > old ones to get the same behaviour with the > new actuator without having to go again to a new pid tuning session? > > Thanks. > > — > Reply to this email directly, view it on GitHub > <#147>, or unsubscribe > < https://github.com/notifications/unsubscribe-auth/AACYX4Q47M56QSH5TXQ6VDTZ75NHZAVCNFSM6AAAAABRQG4JYWVHI2DSMVQWIX3LMV43ASLTON2WKOZSGY2DOMRSHEZTQNI> > . > You are receiving this because you are subscribed to this thread.Message > ID: ***@***.***> > — Reply to this email directly, view it on GitHub <#147 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACUD2A7KY35VGIAKPLAZ5C32BCEQVAVCNFSM6AAAAABRQG4JYWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIOBRGI2DSNRSGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>