Hi All, Saw Gary's reply that PI controller design is done experimentally. But, shouldn't there be a starting point? Actually, I've a PM DC Motor, 180V, 1500rpm. I want to run it using single phase IGBT converters. the open loop is done. for closed loop, i've got a tachogenerator attached. after the feedback summation, there should be a Speed Controller (PI). but, where do i start? i mean, what would be the values of ki and kp?
See my article in Embedded Systems Programming magazine. You can get to it from
Unless you want to go into a whole slew of math you just need to experiment. If you're using velocity feed back then the amount of proportional gain required will be a function of the motor characteristics, PWM scaling, and the load on the motor.
You could work up a set of differential equations for the motor behavior, translate them into difference equations and solve using information in
Unless you measure your system response I _strongly_ advise that you stick to a very conservative set of tuning values (like, do the PID article with the proportional backed of 4x, then keep the integral backed off 4x from what works with that). If you don't do this you may find your system oscillating at unexpected (and embarrassing) times.
If you _do_ want to measure your system response you can use
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Posting from Google? See http://cfaj.freeshell.org/google/
Forgot to mention -- starting values for a PID controller can also be had from the Ziegler-Nichols method, or Astrom-Hagglund. I believe that both of these want to give you PID tunings, not PI tunings, though.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Posting from Google? See http://cfaj.freeshell.org/google/
The derivative term is set last - so if it can be skipped, a PI controller is there.
There must be very good grounds to take the I term but leave the D term out, as the controller will probably like to have instabilities. The steady-state error obtained without the I term may be a smaller nuisance than the oscillation tendencies.
-- Tauno Voipio tauno voipio (at) iki fi
Ah. I did a quick check on the web -- you can get just PI tunings from the Z-N method. I don't mess with it much; I much prefer to start with a good mathematical model of the plant and a conservative tuning, then take measured frequency response data and tune from there.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Posting from Google? See http://cfaj.freeshell.org/google/
Yes, but ZN suggests that ID are strictly related to one another (D = I/6, IIRC). ZN effectively yields a gain and a single time constant.
Steve
The problem is that if you take the numbers from a Z-N PID tuning your proportional will be high, based on the assumption of some derivative action. If you are planning from the outset _not_ to use a derivative term then you need to back off on the proportional gain.
For the purposes of getting a rough order of magnitude, though, an open-loop Z-N process will get you there, regardless.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Posting from Google? See http://cfaj.freeshell.org/google/
Yes, find the information listed in the link below or find some system identification software.
We get asked that question all the time. There is no way that can be answered without the information needed to create a simple model.. Your system will be simple.
I never had any luck with ZN and even if I did, the customers would not put up with the overshoot. I have known better for a long time.
I think all the PID gains should be set at the same time as a function of the desired response.
Doesn't the correct solution depend of the model and the needs of the application? A velocity ( type 0 ) system will not reach the desired velocity unless the controller adds an integrator. So what determines if the D gain is required?
It would be interesting to find out what PID gains you would calculate for this motor example:
Peter Nachtwey
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.