Hi,
I am controlling a process using PID (with a low pass filter) to supress noise. can anyone tell me, why the response is not smooth (erratic trajectory) compared to given by Ideal PID (without the low pass filter?)

Thank you

 

Its possible I've misunderstood your question, but is the process PID the same algorithm as the Ideal PID? To transfer parameters directly between controllers requires them both to be the same type i.e. ideal, series or parallel. Otherwise the parameters need recalculating (which is no big deal and easy to do - but frequently forgotten!).

Regards
John Greene
Process Control Consultant
Contek Systems Ltd, Aberdeen
http://www.contek-systems.co.uk

 

I'm guessing that the trajectory is not really "erratic" but rather "oscillatory". If the PID feedback uses a derivative term, that term is inherently high-bandwidth (as compared to proportional and integral terms). If the derivative term is really needed (which is not very common) then adding a low-pass filter reduces the effect of the derivative term, which could cause instability.

Robert Scott
Embedded Systems Consulting

 

The Greek Philosophers struggled with the concept of the "ideal".

I don't have to struggleas they did, since I'm not a philosopher and I just have to deal with reality. Reality is, I have no ideal PID loops.

I would have to ask rhetorically in response to your question as to why you think your process should act as an ideal process?

If it doesn't, re-tune.

 

Hi,

Maybe I should rephrase the question. I m comparing the responses given by 'ideal' PID - Kc(1+1/Tis+Tds) with the one with filter in the derivative, i.e. Tds/(aTds+1). As the filter in derivative is meant to reduce noise, I'd expect smoother response (and control signals) given by PID with filter, but the results I got is the other way round.

Thank you for the reply.

 

I suggest that a low pass filter (appropriate eg. 5 sec) may be set on the PV to smoothen the noise content in the signal, especially for temperature signals from thermocouple. After filtering, the derivative action will be more effective since it will not be misdirected by high frequency noise which has been already filtered out.