What is the relay amplitude parameter for the PID Autotuning addon VI?

[BCho]

Hey Varun,

Unfortunately, I am not sure of the theroetical method of finding d from your listed equation. Could you attempt to use trial and error for d?

[>>]

Dear all,

when we go for relay autotuning , what should be the sampling period.I mean how often we have to execute (calculate) the pid subroutine ? Does it depends on the ultimate period?

hopefully ,

varun

[BCho]

Hey Varun,

Are you asking, how many times should you call the Autotune vi to calculate the PID gains in your overall system. I.E. How often should you adjust the PID gains? If that is the cause, you wouldn't need to recalculate these gains too often. I would image that the appropriate gains for you system wouldn't change that frequently. So, I would think you should recalculate the gains each time the system changes in some way. If you would just looking for a generic rate to run this Autotune VI, I would think once or twice a second would be more than enough.

[>>]

dear Ben,

        i used describing function anlysis to find kp,ki,kd. But the system takes lot of time(3hrs in my case) to maintain at set temperature. What makes it even bad is I am executing my pid routine during every 1ms.So can you give me any suggestion so that I can work on my settling time ?

hopefully,

 --varun

[BCho]

Hey Varun,

I am not extremely familiar with the method you used to find your gains. But, are you saying that with your current gains, it is taking almost 3 hours to see your system settle, even after calling the PID vi every 1ms? If this is the case, I would think that you might want to take another look at your gains.

[>>]

dear all,

how can i reduce overshoot for my pid controller ..??which is the most prefered method?

[BCho]

Hey >>

To reduce overshoot of your system, you will want to address the derivitive gain (Kd). Check out this PID Control and PID Theory Explained articles for a little more information.

[cmengnr]

Hello all

I am playing with PID Autotuning for temperature Control and seems Relay based feedback apporach is best and less time consuming for autotune process as compare to step input or zieglers oscillation where there is need to continously change P to find symmetric oscillations.

I have a confusion in Relay Amplitude parameter in Relay based Approach. As in above posts, I/p to the system is varied b/w Set point+Relay Amplitude and Set point-Relay Amplitude.

Suppose if set point is 50 degree and input to the system is 230 Volt AC, 10 Amp. How to relate this to find relay amplitude.

Also if input power is duty cycle that is applied to obtained symmetric oscillations when overshoot and undershoot accurs. Suppose duty cycle applied in that duration is 20 %. Then according to above Relay Amplitude = duty cycle- set point ???

Please clarify it. If we have to make hit and trials to find this values, anybody please suggest other accurate method through which we can make system fully automatic.

[PatJamSim]

Hello cmengnr,

After reading through the documents linked earlier in this thread describing the relay-based approach, my understanding of this concept is that it is designed to simulate the expected oscillations to the system, allowing you to autotune based on a high and low threshold.  The relay amplitude should then be similar in magnitude to the expected oscillations to your system.  Whether that magnitude is the total expected possible oscillation or is more related to a single increment up or down is less clear to me.  What values have you been using, and what sorts of results have you been getting?

Here is a good resource for more information on the relay amplitude.   It does appear that trial and error is the best way to determine the relay parameters, as these values are extremely dependent on your application and the performance you desire.

[cmengnr]

Thanks for your reply.

I also searched so many documents and find it can be find through trial and error to get sufficient amplitude oscillations.

But if we have to use trial and error then i think there is no benefit of autotune function.

Is there any other method through which we can fully automate the system.

I also studied ultimate gain method. But in this we have to make hit and trials on value of kp to achieve required oscillations.

But in this what should be the increment factor. It is in .1,.01.001 or 1.