AB PLC5 PIDs

Hello Ron,

Thanks for your interest! Answers to your questions:

(a) Yes the program works to the customers satisfaction, but not mine on the 5/30.

(b) No the program does not work to the customers satisfaction on the 5/40.

(c) No the 5/30 will stay until the customer replaces with Modicon shortly.

The PIDs use PD control block, independent equation and the gains are P10, I0.1 and D0.

There is alot of STX. This PLC controls 4 similar areas. Each area has about ten common STXs; most just average signals, etc but three for each area are quite invloved. FOR-NEXTs and IF-THEN-ELSES nested deep.

The eight PIDs that the customer noticed changes on were for temperature control. Four others that control ventilation had no complaints.

The PIDs are triggered every eight seconds, I was mistaken in post 4. The triggers are consistent. The configured time constants are either 1 or 4 seconds.

When the process is started or taken out of a hold state the SP is set to PV. STX logic ramps the SP to the target over ten minutes while the PID is in auto.

I do not fel comfortable posting the code as the original author might see it and be offended. I will email it.

The system is working to the customers satisfaction. The 5/30 had had 420 words of free memory when I found it. When adding logic to control a VFD, the 5/30 became full. I swapped in a 5/40 and experiencced the problems. Examined 5/30 and found unused data files that I deleted to make room for new logic. System now works like it used to on 5/30 with new logic.

I believe the problems were due to PIDs running in a strange state. The change from 5/30 series C to 5/40 series E merely brought out the fact that the PIDs are non-standard to put it nicely. I could have just re-tuned the PIDs, but felt that I should not have had to because I was adding very little logic to the system. The last and max scan times between the systems changed very little; last ~ 40, max ~ 100 ms.

I will not have a chance to do further testing.

the end of the line? ...

Greetings Derek,

you said:

No the 5/30 will stay until the customer replaces with Modicon shortly.

Click to expand...

now that we have that out of the way, I’d say that finding the answers to all of the PID issues becomes a purely academic exercise ... personally I’d sort of like to know exactly what caused the problem but I doubt that we’ll ever find out for sure ... I’ll always have a hunch that the PIDs were not being “triggered” on a consistently timed cycle and that the PLC-5/40’s scan rate was significantly faster or slower than the PLC-5/30’s ... that scenario would “make sense” and it’s good enough for me ...

anyway ... please let us know if you find out anything new ... but I guess that it’s time to let this one drift off to the archives ...

PLC 5 PID instruction

Gents,

There is a bug in the PLC 5 ladder PID instructions (& it is not documented anywhere as far as i know). Its output tends to wind up when the limits are dynamically adjusted. Switching to MANUAL didn't help ... I have experinced the same problems with CLx ladder PID instruction (I guess is legacy from PLC 5).
The way around to this problem I found was to write my own PID instruction (it works rather well if the process it is not too demanding). Else you can always use CLx FBD PIDE instruction ... I understand that this instruction is not related to the original PLC 5 (They bought the algorithm (rights) from Bailey Controls).

cheers,
LJAM

now THIS sounds interesting ...

Greetings LJAM,

you said:

There is a bug in the PLC 5 ladder PID instructions (& it is not documented anywhere as far as i know). Its output tends to wind up when the limits are dynamically adjusted.

Click to expand...

I certainly do NOT doubt that there is an “undocumented bug” as you say ... but could you please give us a few more details so that we can try to duplicate the problem? ... personally I’ve spent a LOT of time working through the Allen-Bradley PID and I’ve never noticed this particular problem ... I’d sure like to know more about it ... a few specific questions:

1. when you said the “limits are ... adjusted” are you talking about the CV limits? ...

2. by “dynamically” do you mean while the PID is in automatic operation? ... or something else? ...

3. does this happen with both N-type and PD-type control blocks? ... or just with one but not with the other?

4. does this happen with both the “ISA/Dependent Gains” equation and with the “AB/Independent Gains” equation? ... or just with one but not with the other?

please share some more details about this ... I’ll try to run some experiments along these lines tonight but anything that you can add would certainly be welcome ... again, I’m really not doubting your claim but if there’s a bug in there that I haven’t run across, I’d certainly like to know more about it ... thanks for your input ...

PLC 5 ladder PID bug

G'day Ron,

By dynamically I meant during PID automatic execution (also meant CV limits). The problem is also quite visible in applications that require a discontinue operation of the PID instruction (i.e. batch recipe applications) …
Few years ago, I used the ladder PID instruction to control the ramp down braking of conveyors in a power station … since the PID wasn’t used during normal operation I did switch the controller to MAN (& handle its output to suit) … when braking was required, I switched the controller to AUTO and generated a ramp down set point to stop in xx seconds … at times the result was the expected but some other times it took too long.

I was using a hydraulic power pack with relief proportional valve (reverse action). So I had to preset my output to 100% before switch back to AUTO. Trending the variables I released that the controller output at times stayed few seconds at 100% before start chasing the set point.

Another example was during the commissioning of a fully automatic bucket wheel reclaimer. Since the slew is proportional to the reclaiming rate a PID is used to control the slew speed using the power drawn by the bucket wheel drive as PV. Long tale short, output limiting and wind up reset was required at the slew turn around angles (there again the problem resurfaced).

I made some enquires to AB without much success till a local AB expert confirmed (off record) that there was a problem with their algorithm …. Personally I rather use the independent form of the algorithm but I do not think it does have anything to do with it.

Cheers,
LJAM

Thanks for the extra information, LJAM ... I’ll try to play around with this over the next week or two and post any results that I come up with ...

maybe i shouldn't ask this type of question...

how can i tune PID using ladder diagram?

Greetings adzlan,

it’s ok to ask the question ... but you need to put more detail into it ... what exactly are you asking? ...

if you want to know WHERE in the ladder logic program you have to go to tune the loop, then the answer is: “inside the PID instruction’s setup screen” – at least with Allen-Bradley ... depending on how you’ve set up the PID – and on what model of processor you’re using – then the locations of the tuning parameters (P and I and D) will have slightly different names and different units of measurement ... but basically, the adjustments to the tuning values will be made inside the PID’s setup screen ... here’s an example using one of the most common setups ...



on the other hand, if you’re asking how to write something like an “auto-tuning” routine in ladder logic, then you’re getting into some pretty deep voodoo ... before we head down that particular road, I’d have the following question for you ... how good are you at PID tuning already? ... the basic idea is that if you can’t do the procedure manually, then what approach would you use to program the procedure to happen automatically? ...

and incidentally ... if you ARE asking about writing your own autotune procedure in ladder logic, then my friendly advice is to find another way to spend your time ...

thanx Ron,

you are very good in PID.
i've read some of your thread ...how you do that?

can you recommend me any software that good and cheap?


adzlan