pid update rate effect

[kamenges]

Is this a PID function you are writing yourself or is this a controller-resident PID function?

Keith

 

[sthompson]

I think you might have wanted to reply to Kieth.

If you are using the Independent Equation (AB) in a PLC5/40.

Kp = 0.01
Ki = 0.001/secs
kd = 0.01 secs

Stu...

 

[michaeli]

Stu - furnace using a AB plc 5/40e and independent equation having Ki of 1/secs and Kd of secs

 

[sthompson]

Are you by chance using a PD File for the Control Block?

The one I looked at in one of my Programs was using an N File for the Control Block. I just put one in with PD Control Block, and it matches what you were using.

Kp = Gain
Ki = Integral / Seconds
Kd - Derivative in Seconds

If that is the case, sorry for the confusion.

Stu....

 

[michaeli]

Stu - we use PLC5 Independent with "PD block".

 

[kamenges]

So, after going through these posts, do you still have a specific question? At this point you may be better off asking any question you have again so we know specifically what you want to know. We have covered enough ground at this point that I'm not sure exactly what you want to know anymore.

Keith

 

[michaeli]

thank you for all response. very helpful. have read about open and closed loop tuning methods. also Ziegler Nichols method. will this all work for getting Kp, Ki, and Kd values for the independent equation using PV within derivative term and with Allen Bradley PD blocks setup? with the open loop tuning using dead time and time constant generated values. for temperature control on a electric oven/furnace.

 

[Ron Beaufort]

if you'll post your ENTIRE project file (RSP extension) - and the results of an "open loop" test similar to the one shown in the link below, I'll be happy to calculate some suggested values for P, I, and D ...

http://www.plctalk.net/qanda/showthread.php?p=27858&postcount=10

if you decide to go this route, PLEASE change the background color of your graph from black to white ... and choose pen colors with enough contrast to be readable ...

 

[Peter Nachtwey]

I still have my "autotuning" program I did for Ron's Hot Rods
http://deltamotion.com/peter/Scilab/AutoTune/
In the downloads/misc section there is a hotrod.zip but the code is for SLC5/X and MicroLogix.

 

[michaeli]

will work on getting rsp file for open loop test on equipment. attaching file related to tuning furnace using Z/N method. Calculated I and D values would need adjusted as PID loop update and PLC scan times are not same but still Kd value appears extreme.

 

[Ron Beaufort]

will work on getting rsp file for open loop test on equipment.

we just need the RSP file – not necessarily the file while the test is being run ... just save the file – then zip it and attach it to a post right here in this thread ... NOTE: some forum members (including myself) will NOT use a "dropbox" type address to recover your file ...

here's the reason why we want the RSP file:

90% of PIDs are set up wrong ... the scaling and the timing/triggering are the biggest culprits ... until we are sure that the PID is set up correctly – trying to get the "tuning" part right is pretty much just a waste of time ...

attaching file related to tuning furnace using Z/N method.

we don't want to see any tuning at all ... in the open-loop test that we're requesting, the PID is out of the picture (or at least, running in a MANUAL or SET OUTPUT mode) ...

basically you crank in a value for the output (say 10%) and let the system settle (repeat SETTLE) out to a steady state ...

then you crank in a new higher value for the output (say 80%) and let the system settle (repeat SETTLE) again ...

record the initial – and the ending – values for CV and PV ... and make sure that your graph shows the time scaling too ...

examples:
an initial setting of 10% CV gives 115 DegsF for the PV ...
an initial setting of 80% CV gives 375 DegsF for the PV ...
total graphing time is 22 minutes ...
(again - these are EXAMPLES - your values will vary) ...

you can see the types of measurements that we plan to take from the graph in this post:

http://www.plctalk.net/qanda/showthread.php?p=27701&postcount=6

if the open loop test is properly done, it will give us a LOT of information about the characteristics of your system ... then we can base our "ball park" suggestions for your tuning values on the results of the test ...

personally I'm having zero luck opening the PowerPoint picture that you posted ...

it would be better to run a simple TREND in RSLogix5 and then do a screen shot of that ... then attach it to a post as a PNG graphics file ... (ask if you don't know how) ... the graph attached to my first post (#23 above) would be an excellent example to follow ...

 

[michaeli]

will work on conducting open loop test. also for getting rsp file. attaching doc file as replacement for zip file for Z/N tuning study.

 

[michaeli]

have open loop test results. will attach for review. thank you

 

[michaeli]

attached are results for open loop testing. looking for starting Kp, Ki and Kd values for plc 5 using ab independent equation having derivative of PV and with error = sp-pv. thank you for any suggestions

 

[michaeli]

Stu
as you suggested i am working on putting timer in plc so pid Loop update and plc scan time match. for now if i have a Ki = 0.001 with Loop update of 1 sec and a plc scan time of 0.035 sec then the actual Ki value is 1/0.035 x 0.001 = 0.028. also with Kd of 0.025 the actual Kd value is 1/0.035 x 0.025 = 0.71. do i have correct? thanks