Implementing a PID controller for motor speed control
Whelzorn
Posts: 256
Hello everyone,
I'm trying to implement a PID controller in spin to use for keeping a motor at a constant RPM. The equation I'm using can be found here: http://bestune.50megs.com/typeABC.htm
My main PID routine is as follows:
where the plant ( long[output_addr] ) is the PWM duty cycle %, the variable at SP_addr is the desired motor RPM and the variable at PV_addr is the current motor RPM (measured by another cog watching the encoder).
Does this appear to be a proper implementation of a PID controller? I cannot seem to get a set of Ks that don't result in an unstable system.
perhaps someone has a better implementation? I looked at the object exchange but I wanted to try this for myself.
Thanks,
Justin
I'm trying to implement a PID controller in spin to use for keeping a motor at a constant RPM. The equation I'm using can be found here: http://bestune.50megs.com/typeABC.htm
My main PID routine is as follows:
PUB MotorController( SP_addr, PV_addr, output_addr ) | e, SP, PV, PV_D1, PV_D2, Kp, Ki, Kd, T, time, CO, plant
'init variables
SP := 0
PV := 0
e := 0
PV_D1 := 0
PV_D2 := 0
CO := 0
plant := 0
'set PID gains
Kp := 1
Ki := 1
Kd := 1
'set T (in milliseconds)
T := 1000 / UPDATE_FREQ
time := cnt
repeat
SP := long[SP_addr]
PV := long[PV_addr]
'calculate the PID output
e := SP - PV
CO := CO - ( Kp * ( PV - PV_D1 ) ) + ( Ki * T * e ) - ( ( Kd / T ) * ( PV - ( 2 * PV_D1 ) + PV_D2 ) )
PV_D2 := PV_D1
PV_D1 := PV
'Limit the output for anti-windup
CO <#= 10
CO #>= -10
plant += CO
plant <#= 100
plant #>= 0
long[output_addr] := plant
'delay for T milliseconds
waitcnt( time += ( clkfreq / UPDATE_FREQ ) )
where the plant ( long[output_addr] ) is the PWM duty cycle %, the variable at SP_addr is the desired motor RPM and the variable at PV_addr is the current motor RPM (measured by another cog watching the encoder).
Does this appear to be a proper implementation of a PID controller? I cannot seem to get a set of Ks that don't result in an unstable system.
perhaps someone has a better implementation? I looked at the object exchange but I wanted to try this for myself.
Thanks,
Justin
Comments
I think you mean you can't find a set of K's for which the system is stable.
Sometimes tuning a system for PID can be a little tricky, involving some trial and error. The Parallax book "Process Control" has some info on PID. Free download too. Check out chapter 8.
http://www.parallax.com/Portals/0/Downloads/docs/prod/sic/Web-PC-v1.0.pdf
Where've you been, kid? (:
[Long time, no appearance.]
Oh man, it's been over 3 years since I've posted?! I suppose I have college to blame, although I would have to blame this community partially for inspiring me to go for an electrical engineering degree in the first place!
I'm glad to see everyone is still here and that I'm remembered!