PID Input Limits and other PID questions

[zed]

Hello,,

I'm using a 3 wire RTD as an input which has a range of -200C to 850C.  The integer value is x10 (1 decimal place) so for example 20C would be an integer of 200.  Should the Input limits be set to -200 to 850 or -2000 to 8500?

What is the effect and proper way of setting sample time?  

In the config struct there is a STATUS tag.  The help file decodes all the numeric values but my status tag indicates "2" which does not have a definition in the help file.  There is a -2 but not a 2.  So what does 2 mean?

Is there a formula for the Cp, Ci, Cd influences?  Just noticing the tags for each do not seem to add up to the CV.  Also wondering the impact to control for the integral setting.  For example if the integral is 5 seconds, what would be the impact if changing that to 10 seconds?  To 2 seconds? I guess that's why i'm asking if there is a formula because that would give me the answers.  I can of course google it but it depends on how the manufacturer applies it.  Proportional band vs. proportional gain or integral time vs integral rate.  

 

Lastly,  I had a PID loop freeze up on me and i have no idea why.  Its been in service for about a week working fine.  Then after the weekend it would not respond and was stuck at 100% output even though the the PV was well below the SV (cooling application).  Cp, Ci, and Cd values were not changing.  Integral error was zero (no wind up).  The only way i recovered control it was to run the AT again.  very odd, just wondering if anyone has encountered the same.

 

Thanks

[Pavel_setnicar]

Yes , also encounterd the same in my app with SAMBA.

For weeks it worked just great and then at some point output of the PID went 100% ad only autotune get rid off it.

I did not check the PID status since I was not on site ,but costumer lost confidence in unitronics SAMBA.

It is strange because it worked perfect for some time. Instalation is on wery remote site and I can not  get all information in time (hidroelectric power generator).

If windup would occour , should i get this symptome?

[Flex727]

21 minutes ago, Pavel_setnicar said:

I did not check the PID status since I was not on site

I always display the PID Status integer on screen, usually on the PID setup screen.

[kratmel]

In my applications that use a combination of PID and Samba, autotuning runs once or twice on startup. After that, the system works without problems.

In one system I found 100% output on the screen several times, but it is a water-to-air heat exchanger. Heated water from the cooling system of another machine is used. And several times the water from this machine came at a low temperature.

Therefore, in PID regulation systems, it is worth carefully studying the behavior of the energy source. Inadequate (lack of sufficient) amount of energy causes the transition of the PID regulator to the saturation state.

Carrying out auto-tuning, we suspend the normal operation of the system, and sometimes this is enough to restore the energy balance. Maybe a little too much theory.

But you say that auto-tuning helps.

Question: Does the result of auto-tuning repeat the result that was before it? I mean the obtained coefficients.

On my systems, multiple autotuning repeats the results with reasonably high accuracy.

 

 

[zed]

@kratmel, In my system AT does not produce the same results for coefficients nor would I expect them to given my system.  Its a heat exchanger system with a fairly long residence time and the application is in the food industry cooling various types of sauces used in meals all of which have different specific heat capacities, densities and flow rates.  I don't incorporate any sort of feed forward control for flow changes however in a given day flows are relatively consistent.  I find that being careful to tune for a laminar flow with a medium to high flow rate sku works adequately for all skus.  Not perfect but very hard to expect perfection with this application.

I do have some products where the capacity of the cooling system is not enough to cool to the setpoint and as such the output will be at 100% most of the day.  I guess i'm expecting that when the run PID block is disabled, any windup or saturation would be cleared.  Maybe i need to reset this manually (using logic) at the leading or trailing edge of Run PID?  Oddly, in the frozen at 100% case above, the Integral error was zero,  which I think indicates there was no windup present?

@Flex727 I also have the status integer displayed.  Problem is it shows "2" for which there is no definition in the help file.  Also during the freeze event I was able to go online and confirm that the Run PID block was indeed enabled.

[Pavel_setnicar]

Putting PID stus messages on screen is a good idea.

Autotuning did help but than situation repeated again so I installed new SAMBA with modified program to reset Integral error in case of windup or winddown (even if I do not know if it ever occured).I also display pid status.  Now Im waiting for next call from client.

[Flex727]

17 hours ago, zed said:

Problem is it shows "2" for which there is no definition in the help file. 

Sure there is:

[zed]

@Flex727 Well now that's odd.  I'm using Unilogic version 1.33.373 which is the latest, last I checked and when i "update help file" it tells me that it is up to date.

Here's my help file page:  No "2"....

 From your sheet 1,2,3 would seem to indicate the current number of autotune cycles which makes sense but I can also say that in my case autotune was not in progress at the time. Something is fishy.  I'll have to keep watch.

[zed]

Just after posting that I noticed the description for the autotune done bit.  "Reset by user" and thought, Oh I didn't do that, maybe that's why i'm having problems, but when I tried it won't let me reset that bit using either reset coil or reset numeric. or store function.  Seems to be read only.  

[Flex727]

13 hours ago, zed said:

Here's my help file page:  No "2"....

My apologies, I was still thinking about the previous comment mentioning Samba and posted the VisiLogic Help file.