PID to time proportional

Skidood · Oct 31, 2022

All this recent talk of PID brought me back to an issue I was looking at a few months ago.
My next project will involve cycling a valve to provide a PID controlled heating function.
In the screenshot below, we see an example provided by Rockwell in the SLC instruction set manual of how to convert a PID CV output to a "time proportional" digital output.
But they managed to obscure an address with an arrowhead. See the red circle. Can someone take a guess as to what is under that arrowhead?
Thank you kindly!

Steve Bailey · Oct 31, 2022

From the manual I have, which is publication 1747-6-15, dated January 1996, the DN bit is bit 13 of the first word of the control block, so in your case it is N7:2/bit 13.

drbitboy · Oct 31, 2022

It looks like Bit 13 of PID Control Word 0 is a one-shot that is 1 one the single scans when the PID updates.

Cf. this link, page 296, as noted by @Steve Bailey.

Peter Nachtwey · Nov 1, 2022

Now how do you do this without a timer but assume you are executing the PID every 10 ms using a timed interrupt. Think about it.

cardosocea · Nov 1, 2022

Peter Nachtwey said:
Now how do you do this without a timer but assume you are executing the PID every 10 ms using a timed interrupt. Think about it.

Count cycles instead? The issue there is more around the range of values available, I guess.

drbitboy · Nov 1, 2022

Peter Nachtwey said:
Now how do you do this without a timer but assume you are executing the PID every 10 ms using a timed interrupt. Think about it.

Trick question, because PLC programs are about time, and are never without a timer of some sort, including the scan cycle itself; the only question is the fidelity requirement vs. capability.

Peter Nachtwey · Nov 1, 2022

Assume you are using timed interrupts. Assume the CV is 0 to 100. The upper limit isn't that important but I will assume the CV range is 0 to 100.
The trick is to accumulate the CV or add the output of the CV to a CV_SUM. Now it the CV_SUM is >=100 then the output is on and 100 is subtracted from the CV_sum. Otherwise the output is off. The amount of code for this is small. You can see the upper limit of the output doesn't make much difference.
If using a real DtoA converter with a range of 0-4095 then the upper limit is 4096. There is no need to scale the output to a round number like 0 to 100 or 0 to 1000.
Simple.

Skidood · Nov 1, 2022

The Rockwell strategy above to convert the CV to "time proportional" didn't appeal to me, plus I can't use double divide, not supported on my PLC, so I came up with my own logic which was a lot simpler, in my mind.

The problem I'm having now that I can't get the PID instruction to execute.
I have read all the documentation.
Tried it in an STI...can't even get the STI to work.
Tried it a normal routine...didn't work
Im using an integer like 10000 as a setpoint, with no limits set on the CV
My PV is 1000 which is also a integer
Mode is auto and timed. Interval is 0.1 seconds.

Using RSEmulate and I can see the program running, no problem. The PID rung is green as it should be. I give up. I have enough problems in life, I don't need this silliness.

drbitboy · Nov 1, 2022

Skidood said:
The Rockwell strategy above to convert the CV to "time proportional" didn't appeal to me, plus I can't use double divide, not supported on my PLC, so I came up with my own logic which was a lot simpler, in my mind.

The problem I'm having now that I can't get the PID instruction to execute.
I have read all the documentation.
Tried it in an STI...can't even get the STI to work.
Tried it a normal routine...didn't work
Im using an integer like 10000 as a setpoint, with no limits set on the CV
My PV is 1000 which is also a integer
Mode is auto and timed. Interval is 0.1 seconds.

Using RSEmulate and I can see the program running, no problem. The PID rung is green as it should be. I give up. I have enough problems in life, I don't need this silliness.

RSEmulate does not emulate the PID instruction.

Cf. here and here.

Skidood · Nov 1, 2022

drbitboy said:
RSEmulate does not emulate the PID instruction.

Cf. here and here.

OK thanks Mr.

drbitboy · Nov 1, 2022

That said, it would not be difficult to code a Program File that emulates the SLC 5/xx PID instruction.

The size of the task depends how many features are to be included. E.g. P and I action, and Auto/Manual would take but a handful of lines. Any single timing mode should not be too tough' handling all timing modes would be a bit (pun intended) more involved.

Skidood · Nov 1, 2022

I would love to immerse myself in that but at some point a guy's gotta do things like eat and attend to personal hygiene. But yeah, I'm just using P and I....and Auto...haven't even looked at the equation yet. If I do, I'm in trouble. I just got a PO in for 90K for something unrelated to PLCs so my mind is going a little haywire right now. (as per usual).

drbitboy said:
That said, it would not be difficult to code a Program File that emulates the SLC 5/xx PID instruction.

The size of the task depends how many features are to be included. E.g. P and I action, and Auto/Manual would take but a handful of lines. Any single timing mode should not be too tough' handling all timing modes would be a bit (pun intended) more involved.

Peter Nachtwey · Nov 2, 2022

Writing a PID is easy. The SLC500 PID uses integers. I would use floats.
I don't know what drbitboy means by timing modes. The time between PID executions should be fixed and put in a timed interrupt so updates are done a relatively equal intervals. If the PID is to run in the main scan then it is necessary to read the system clock to compute the time since the last time. There should be NO need to enter a scan time. Most of the calculations could be done in a few compute blocks.

drbitboy · Nov 2, 2022

Peter Nachtwey said:
I don't know what drbitboy means by timing modes.

The SLC 5/xx PID instruction has two timing modes, both of which @Peter Nachtwey describes, even though he claims not to know what they are (there is nothing new under the sun

); which mode is active is determined by the state of a bit in the Control Block word 0 (TM, bit 0).

What I was referring to was whether a manually-coded PID would have such a configuration option and would model both timing modes of the SLC PID instruction.

drbitboy · Nov 3, 2022

Skidood said:
...
But they managed to obscure an address with an arrowhead. See the red circle. Can someone take a guess as to what is under that arrowhead? ...

The bit number is text in that PDF, so even though the arrow is covering it, it can be double-clicked to select the text (see image below), then copied, then pasted somewhere else to read it.

PID to time proportional

Skidood

Member

Steve Bailey

Lifetime Supporting Member + Moderator

drbitboy

Lifetime Supporting Member

Peter Nachtwey

Member

cardosocea

Member

drbitboy

Lifetime Supporting Member

Peter Nachtwey

Member

Skidood

Member

drbitboy

Lifetime Supporting Member

Skidood

Member

drbitboy

Lifetime Supporting Member

Skidood

Member

Peter Nachtwey

Member

drbitboy

Lifetime Supporting Member

drbitboy

Lifetime Supporting Member

Similar Topics