PID (Basic Issue)

[Esther.Lopez]

Hello you all
I have to develop an application that reads a value from a DAQ card, compares it with a desired value, and then generates an output to try to bring the read value to the desired value.
I have thought on using a PID control, but I have many doubts on it. can somebody help me??
Thanks!!

Message Edited by Esther.Lopez on 04-20-2005 08:50 AM

[tst]

This is the basic use of PID. What kind of doubts are you having?
There should be some PID examples with LV (assuming you already have it) and there are probably some articles on the site on the topic.

[Esther.Lopez]

the problem is that I can manage to find correct pid parameters for my application. and I also have doubt with % and EGU units... It's not my best day today 😞

[Jorge M.]

Hello Esther,

taking into account that the P (proportional) factor is the most important (should be the higher value in "normal" systems), the values of the I and D factors are normally set after several trial and error tests with the system(this is usually faster than doing a mathematical study or simulation of the real system unless you have a very deep knowledge of it).

I would take a look at these links, but basically I would play with different PID values, first with a simulation (LabVIEW has examples), so you can visually see it on a graph and understand how the PID values affect the response of the system (overshoot, undershoot, response time, etc...). After that, experiment with your real system until you are happy with its response.

http://zone.ni.com/devzone/conceptd.nsf/webmain/F0A8687D6706719686256D6D005F9C56
http://zone.ni.com/devzone/conceptd.nsf/webmain/D211223AE3DE7E74862568110052F62F
http://zone.ni.com/devzone/devzone.nsf/webcategories/29533F539B47A15B8625696B00744C42

There are also more advanced techniques (autolearning, expert systems, etc...), so if you are interested, I can give you some references, though I feel it's not what you really need.


Regards,

Jorge M.

[Poormanphysics]

Hello Jeorge:

The links you provided are key fundamental principles in PID controls indeed. Since some lab project of mine is running along these lines now I thought I make a visit to the Forum and look around. Now some of my comments.

Many of the theories seemed to pre-suppose that the output variable or the controlled variable (call it process value PV) is sampled "simultaneously" with the input or controlling variable (call it set-point SP) and these two variables present themselves "simultaneously" to some comparator or error amplifier for calculating the difference before passing that to the P-I-D blocks. Now here is where we have to make certain engineering judgement: There is a "quantum leap" price diference for simultaneous sampling DAQ cards (like the S-series)from general purpose multiplexed inputs DAQ cards (like the M-series)and we know that we need two independent physical analogue input channels for real time PID (like what you can buy off the shelf small square caged controllers.

Questions:
1.If you can't afford a DAQ card which is simultaneous sampling, would a very fast muxed analog card do for PID?
Remembering that the channels has some skew of submicron or millisecond when going from one channel to the other?
2.From question #1 does this mean a "sampled input" "sampled output" would out-synch each other in the process when doing PID?
3.How do you exactly synch that from question #2?
4.Would there be stability issues? Since in effect there is presence of time delay in the closed loop when sampling the set-point and process value?

Regards,

Berns B.