Announcement

Collapse
No announcement yet.

PID Low Pass Filter / Calcualtion

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • PID Low Pass Filter / Calcualtion

    Hello All,

    Is there any documentation in regards to the the math used under the hood of the PID function? I am mainly looking to find out if a LPF is applied to the derivative, I am dealing with a bit of a noisy signal and looking do some filtering to increase the loop stability.

  • #2
    To continue with a related issue. I am slowly heating up a liquid by adding steam, which is controlled by the PID. At the end of the process my output is 650, when the cycle is done I disable the pid. When I restart the cycle for a new batch and enable the PID the initial output starts at 650 and climbs down to a reasonable number which is about 100. How do I reset the output between runs?

    Comment


    • #3
      Set the Bias Term in the Advance Tab of the PID to zero before enabling. Also try the "Error Squared" in the Advanced Tab.

      Setting this bit will Square the Error. This can be useful for many processes by effectively diminishing the control effect on smaller Errors while maintaining the control effect on larger Errors. This can also be useful for reducing the effect of low frequency electrical noise in your process.

      Comment


      • #4
        I've had some success with a couple of the smoothing filters described here:
        https://en.wikipedia.org/wiki/Savitz...93Golay_filter

        Don't be put-off by the math they throw out. Go down to the bottom and look at the Appendix. There are tables here showing coefficients to use in the basic formulas just above the tables. What is neat is you can smooth the data points and get the 1st, 2nd, and 3rd derivatives with the coefficients in these tables. I've used the 1st derivative smoothing filter to provide the D term for a PID loop. You can set the D term in the PID instruction to zero and then add the filtered derivative to the PI output. I've also used this smoothed derivative for feedforward terms.

        You can also use the AVERAGE instruction to smooth data. It's basically a weighted, moving average filter and works quite well. It doesn't give you a derivative, of course, but you can calculate that from the data over several samples and then use the AVERAGE instruction. The ARRAY STATISTICS function might also be helpful.

        Comment


        • #5
          Thank you very much for the link, I will give this a good read. I didn't consider doing the derivative separately but that is a path I may follow.

          Comment

          Working...
          X