Can AC ripple cause oscillation in a control loop?

[goingtoparis]

Hello,

I have a controller receiving signals from 3 componets: a warwe level transmitter, a steam flow transmitter, and a water flow transmitter. The controller controls ther water level control valves (i.e water flow). The level transmitter indicator and the steam transmitter indicator are steady. The water flow indicator and the water flow valves are oscillating. We measured some high AC ripple in the water flow loop. Could that be causing the oscillation? Does anyone have experience with that (BTW I am a mechanical engineer with limited knowledge about electronics)

 

[IRstuff]

Can't you compare the oscillation with the noise?

In any case, would that be surprising? How would your controller know the difference between noise on your imput versus actual data?

TTFN

 

[VisiGoth]

By oscillations I assume you mean the generic term of a repeating fluctuation pattern, maybe even sinusoidal looking.

In the control loop world, and in EE circuits there are intentional oscillator devices that have specific design criteria and the oscillations are intended.

IRstuff gave a question that leads to determining if the oscillations are from sort of feedthru based on an external stimulus, or finding out if somehow your control loop is oscillating. I do not suspect the latter simply because I assume the loop was properly designed and at some time had worked. However, the latter could be true if some parameter inside your control loop has changed since the time it had been working. Oscillations can come from an increase in gain in the forward path, the feedback path or the detector sensitivity. Also, an excessive increase in delay in the thru path or the feedback path have the possibility of causing an oscillation.

One way to do this, assuming it is practical is to break the feedback for some amount of time. Even if it is practical it might not be advisable since open lop gains could be quite high and damage can happen, so I do not want to advise that. So IRstuff's question is mort pertinent. Can you correlate the input and the output signals? i.e. can you take the sinusodial component of each and see what the phase shift is? Or if there is any similarity.

 

[goingtoparis]

Thanks for the response. To answer your question regarding correlating the output with the AC noise at input, I have actually found that the frequency of the noise is much higher than the oscillation. However, there are 3 inputs to this controller and 2 of the inputs have AC noise where the input is supposed to be DC only. Could the AC noise from the 2 inputs combine with the DC inputs to create a different pattern of oscillations?

 

[VisiGoth]

If the control laws in your controller are linear, then no, there would be no frequency conversions. Except ion the unlikely case that your high frequency noise is not noise but a stable tone that is somehow subsampled synchronously in your controller and folded down into a lower Nyquist band. But this is imaginary postulations, so we should stick with likelihoods.

Your controller may ignore (filter) high frequency noise. Low frequency oscillations could be a control loop problem. If the AC is 50 or 60 or 400 Hz, then it is most likely not an oscillation but a feedthrough.

If you can provide a data set of the sampled input and output and some info on the controller I could look further. Have you made sure that the actuator is out of spec? An erratic or sticky valve might cause the loop to over react then over react again to compensate for the overshoot. This pattern would repeat and show up as a low frequency oscillation.