Tuning PID values

[radhika]

Hi everybody,

I am controlling temperature process using simple PID program that comes with Labview 7.1

I have attached the response i obtained .

I am trying to find out the critical gain and the time period by setting various P values using closed loop step test method.

My question is whatever P value i set am not able to get steady oscillations.............

so how can i findout the critical gain value and the time period????

Any help is highly appreciated.

Thanks in advance

radhika

[Barp]

radhita,

I noticed that your measurement signal is presenting a low resolution, defined by the values.

To be able to use this technique, you have to increase the resolution of your measurement signal until the point that resolution doesn't affect it. 

If you can't change that, there is no point in trying to improve your controller, since you can never make a control system being more accurate than your measurement system. Does it make sense?

Anyway, since you are using LabVIEW, you can tune the controller online, starting with a Proportional Gain and you can increase until the control signal starts to oscillate. It is doesn't oscillate, you can just use a P controller and you probably will have the error very small.

If is oscillate, just reduce the gain and try to modify the Integral Control and see if you can zero the error.

After that, try to change the setpoint in several operating points to be sure that the parameters that you have is good and retune accordingly.

Hope this helps. Let me know if you have questions.

[radhika]

Barp,

Thankyou for the inputs.

I am using NI6023E DAQ board and pyrometer as a temperature sensor.

The data output from my sensor is of 1mV resolution which accounts to about 0.25 degrees , which means that at room temperature my reading might vary +/- 0.25 degrees.

But as of now , the room temperature reading is varying by 3 to 5 degrees.

How can i increase the resolution???

Thanks

radhika

[GiGe]

Hi radhika,

If you look at page 38 of the E series User Manual there is a list of input ranges and the corresponding resolution.  You can set the input range in the Measurement Automation Explorer when you set up your task.   Basically the smaller your input signal range is the better your resolution will be.  If you try this are you getting the expected resolution?  I hope this helps.

GG