PID Scaling Inquiry

In Allen-Bradley world (Controllogix Rev 30 controller) is PID scaling ALWAYS set to match the input card / sensor scaling?

I am simulating a process in a test lab, where I attempt to maintain a steady pressure reading using a PID while introducing changes to the system.

My Transducer is 4 to 20 mA = -10.0 psi to 450psi.

My logic is in kPa(because the system we are simulating is programmed this way). So my sensor range is ~-69 to 3103 kPa

The process I am simulating has a min pressure of 0 kpa and a max pressure of 80 kPa. So, you can see that I am only using a tiny portion of the sensor's range.

When I set my PID Scale tab to match the sensor scaling (unscaled and engineering min/max), I have to use huge P gains to get the process under control. Huge being P gain of 175 or so. It seems to control it fine when I do this, but I'm not accustomed to seeing values this high.

I make the PID scale max and min something more realistic to the process. Say min 0 and max 310 kPa (I did this because it was 10% of sensor range and easy math) my gains come down into the real world. But it is my understanding that this is incorrect.

So, are large gain numbers just a part of having a sensor with a kPa range that large, or should I adjust my scale to bring them down, or do I have something else going on?

Thanks for your time!

Do unit analysis.
A gain of 175 is meaningless. A gain should have units of % control output per unit of error. In this case are the units KPa or Pa? That will make a huge difference in the gains.

Peter Nachtwey said:

Do unit analysis.
A gain of 175 is meaningless. A gain should have units of % control output per unit of error. In this case are the units KPa or Pa? That will make a huge difference in the gains.

Click to expand...

All units are kPA.

I suppose it makes sense that to achieve the same % of move would take a larger number in the gain when you increase the overall range of the scale.

Thanks!

dladd29l said:

All units are kPA.

I suppose it makes sense that to achieve the same % of move would take a larger number in the gain when you increase the overall range of the scale.

Thanks!

Click to expand...

Yes, Ok, but a kPa is how many AtoD counts?
How many DtoA counts is 100%
Rockwell PIDs normally have unitless gains of error_counts to output_counts.
To figure the real gain one must use all 3 ratios.

Whats the complete catalog number of your pressure transmitter?

Some /most can be reranged.

Peter Nachtwey said:

Yes, Ok, but a kPa is how many AtoD counts?
How many DtoA counts is 100%
Rockwell PIDs normally have unitless gains of error_counts to output_counts.
To figure the real gain one must use all 3 ratios.

Click to expand...

I probably should have prefaced this with I'm fairly new to PLC's and I learned what I know through a couple online courses and programming this test stand, so there are probably gaping holes in my knowledge, haha.

So, having said that, I'm not sure how to determine the analog to digital counts and vice versa. I'm willing to learn if there's a thread on here explaining the process, i've spent the last half hour or so trying to locate one on here.
All tags pertaining to pressure are reals/floats to match the existing system we are mimicking. (I've attached a couple screen shots if that helps)

Mickey said:

Whats the complete catalog number of your pressure transmitter?

Some /most can be reranged.

Click to expand...

It's an STS ATM/IS 119055 - When trying to get more specific and to find my exact pressure range, i'm not seeing an option online so I might fall under their "custom" range option.

So, It's an STS 119055 ---> 1756-IF8/A-------> 1756-L71

dladd29l said:

So, having said that, I'm not sure how to determine the analog to digital counts and vice versa. I'm willing to learn if there's a thread on here explaining the process, i've spent the last half hour or so trying to locate one on here.
All tags pertaining to pressure are reals/floats to match the existing system we are mimicking. (I've attached a couple screen shots if that helps)

Click to expand...

Having read the i/o manual (imagine that) it appears that since I'm using reals/floats I don't have a count per se but instead the value is represented in % of my engineering units.

How does this effect calculating my gains?

dladd29l said:

Having read the i/o manual (imagine that) it appears that since I'm using reals/floats I don't have a count per se but instead the value is represented in % of my engineering units.

How does this effect calculating my gains?

Click to expand...

The ControlLogix (and PLC-5) PID instructions convert your PV and SP to percent of full scale before computing error. So the proportional gain is considered percent-to-percent in calculating output at each update. So if your tuning method generated gains based on engineering units, you will need to back out that scaling. This may also apply to integral and derivative if using the independent gains equation.

Mispeld said:

The ControlLogix (and PLC-5) PID instructions convert your PV and SP to percent of full scale before computing error. So the proportional gain is considered percent-to-percent in calculating output at each update. So if your tuning method generated gains based on engineering units, you will need to back out that scaling. This may also apply to integral and derivative if using the independent gains equation.

Click to expand...

This is helpful, thank you.