ControlLogix PID

[marshg]

I'm trying to tune a temperature loop using a ControlLogix.
I've used the closed and open loop methods in other PLCs . The other PLCs have given me a procedure and formula to figure out the integral and derivative time constants for their particular PID algorithm. I've looked, but have not seen a formula for the ControlLogix. Would anyone have an answer for this specifically for the ControlLogix?

I've also used the method of increasing the integral until the process variable oscillates and then increase the integral to smooth out the oscillations. What method is best to use and why? (open, closed, increase gain until oscillations occur).

Thanks,

 

[controlsdude]

I believe you can just download a PID tuner online. Explanations are a plenty. If that does not work you can dig out your controls books from when you went to college.

 

[cdevine]

I haven't got my RSLogix in front of me at the moment so I'll try and rememeber some of the settings that I normally use. I generally follow a Ziegler Nichols type approach.

1. Equation Type - Dependent Variables
2. Start with K' = 0.5, I' = 1000, D = 0
3. Increse K until oscillations are seen in response to minor SP changes. You might even want to use OP clamps to limit OP swings. At this point set K = 0.5 K'. Essentially K' is approximately the natural frequency of the system.
4. Decrease I until you see oscillation response to minor SP changes. At this point set I = 0.5 I'.
5. Same agin for D

Having said all this temperature loops generally have large time constants. In this case a combination of dominant P and D terms with low contributionfrom I works a lot better.

There are other more complex (and in some cases better) methods but you should be able to get a reasonable hand tune using the above.

Chris Devine

http://www.icande.com.au,

http://www.redbusbar.com

 

[PNachtwey]

I would follow the methods on

http://www.controlguru.com.

The idea is to excite the system and determine the gain, time constant(s) and dead time. Given these it is easy to use the IMC methods to calculate the controller gain Kc and the time constants for Ti and Td if required. ZN doesn't provide very good tuning by comparison.

See this thread

http://www.plctalk.net/qanda/showthread.php?t=27057&page=1&pp=15&highlight=Advanced+Control