please help! damn PID control

[sparkytex]

So were commissioning all the PID loops I had programmed to work with the new wonderware HMI system that was installed. Got all but 1 tuned pretty damn good. This last one is a FIC-7012 which is a flow transmitter ranged 0-150" (DP cell) which is square rooted inside the instrument then sends the 4-20 mA signal to the PLC 5 processor. This damn flow transmitter is bouncing quite a bit at the low end and is causing the PID to chase a lot, and due to the fact the PID runs at the lower end it makes this difficult. I know there's a way I can smooth out the signal so its more of an average value so the PID can handle it easier. How do I got about averaging this value, its scaled 0-4095 raw data in I/O config (N22:9)

2.) I need to change the PID enable scan time to .250 seconds, I know I can use a timer on the .01 time base and a pre-set of 25. But how do I change it in the PID update time, do I just type in .250, I've faulted processors in the past typing in wrong update times for the PID control so I'm double checking?

this flow control is the only thing slowing preventing us to run as of now and I'd like to get this smoothing figured out as well as the PID update time. Any help is much appreciated right now!!

cheers, Tex

 

[Peter Nachtwey]

You need a low pass filter on the PV. There are plenty of threads on how to implement a low pass filter. You should use floating point.
PV = PV(n-1)+K*(NewPV-PV(n-1))
PV will be the filtered PV.
K should be a float between 0 and 1 but start out at 0.1. Make smaller for more filtering.
A compute block would work well for this.

 

[sparkytex]

awesome! got it any hints on the update time? thanks a bunch

 

[Peter Nachtwey]

It depends on how fast you expect the flow to change. Every 0.1 to 0.25 seconds seems reasonable. The big thing is that the time interval needs to be consistent. Don't use timers if you can use interrupts. Interrupts are much more consistent.

It is also best to configure the PID to act like a I-PD when possible. This means that only the I term will act on the error. The P and D terms act only on the changes in the PV. This effectively turns the I-PD into a low pass filter too.