help in bulid digital PID controller.

[carrier]

autotunerpid and polimi

hi
i want to build the digital PID controller for control the valve in a close loop.
a 4-20mA from PLC is a set point and a 4-20mA from a valve is a feedback.
i use equation uk=u(k-1)+q0e(k)+q1e(k-1)+q2e(k-2) in my design but valve are ringing and fixed after very wide range.(about 30 times valve close-open to fix)
i use q0=8.5-q1=-13.5 and q2=6.5 based on elektor digital PID controller 1999.
if anyone know about realization of this circuit plz help me.
thanks

 

[carrier]

any one know????????????????????????
this article base my design.

 

[7rots51]

your problem is in your coeffiecints,you must find good values for your plant,good values for PB(proportional band in %),Ti,Td(both in sec).

You sould use autotuning method to find the best values ,you need to learn more to have a sense about autotuning.

For PID use standard ISA equation because that is very near to real world PIDs.

 

[carrier]

what is ISA equation?

 

[jonw0224]

carrier,

I don't know how much you know about control. In a PID, you have three components of your control.

(1) Proportional - the amount you correct is proportional to the error (you give the car more gas when you want it go faster, as you get close to your desired speed, you let off the pedal).

(2) Integral - the amount you correct is proportional to your cummulative error. You will likely over shoot once you use integral, but your steady state error is guarranteed to be zero if you use any integral control.

(3) Derivative - the amount you correct is proportional to the rate of change of your error.

One way to tune a PID is to set all coefficients to zero. Then increase q0 and decrease q1 proportionally until the valve moves (q0 > 0, q0=-q1) in this step. This will give you a proportional control, but won't insure a steady state error of zero. To insure a steady state error of zero, we'll add some integral control. You can do this by increasing q0 without decreasing q1. However, you may want to decrease the proportional control as you increase the integral control to make the valve react quickly AND have a steady state error of zero. Usually, in a control problem you don't need to use differential control BUT you can add differential control by decreasing q0 and q2 proportionally and increasing q1 twice as much as q0 or q2.

There is a whole field of engineering dedicated to figuring out your problem. Many ways to do it, most of them difficult.

 

[carrier]

the formula is correct?
what mean of u(k-1)?

 

[jonw0224]

Yes, the formula is correct, but your values for q0, q1, and q2 are not.

k is the sample number.

u is the control. u(k-1) is the control at the last sample. Similarly u(k) is the current control.

e is the error. e(k-1) is the error at the last sample. e(k-2) is the error two samples ago.

 

[jshroff]

Hi,

Anyone have the PIC code for the PID controller article that is posted here?

Thanks

 

[carrier]

thank you
i know that u(k-1) is last sample but what meaning physical or real world?
e(k) is a error in this sample and q0 coefficient of this. this mean that with large q0 the weight of error at this time is very important.
but what meaning of u(k-1)?

 

[Guest]

u(k) is the manipulated value at the present
u(k-1) is the manipulated value at the sample privious than present

 

[olemp]

i suggest you to read this article:
PID without a Phd.
**broken link removed**

 

[7rots51]

Hi


**broken link removed**