How to find the engineering units of PID values?

Hi,

I'm working with an actuator using a PID controller. The controller is like a black box and nothing has been said in the manual on the SI units of proportional, Integral and Derivative gains. The range of PID values is 0 to 32767.

Could anyone help me on how to find their SI units?
My goal is to calculate the process variable mathematically and draw the curve for the pid controller.

Best

the Value you mention looks either like an output or input value.
nothing to do with the actual PID.
do you have any other information.
If this is just the actuator - The PID is done by it's controller

Proportional gains are generally unitless. The units on integral are typically resets per minute. The units for deriviative are usually seconds.

The engineering units on your input depend on the transmitter, with 0 being 0% of the transmitter range and 32767 being 100% of the range. That could be or to 100 kPa or 0-1 Newtons per hectare or 0-500 degrees C or whatever.

You need to get a manual for the controller, as different controllers can use quite different set up.

Proportional gain is dimensionless, but some controllers use proportional band, usually quoted as percent, which is the reciprocal of gain. Integral action is usually minutes per repeat, but might be repeats per second. The PID algorithm may also be different, with independent gains or not.

I already know that there are three different types for PID algorithms, Ideal, Parallel and Series.
Proportional (P) unit: Gain or band
Integral (I) unit: Seconds/Repeat, Repeats/Second, Minutes/Repeat or Repeats/Minute
Derivative (D) unit: Seconds or Minutes

How is it possible to go through all combinations?! There should be a way to find them by a procedure!

To AutomaticLeigh:
Yes, I can put the controller in the manual mode but it only gives the dead time, time constant and steady-state gain. Is there a way to find the SI-unit by using them?

It might help someone help you if you tell the forum the brand name, model number, catalog number of this controller. And the process it's controlling.

http://www.pidtuning.net/pid-algorithm.php

ramah said:

Hi,

I'm working with an actuator using a PID controller. The controller is like a black box and nothing has been said in the manual on the SI units of proportional, Integral and Derivative gains. The range of PID values is 0 to 32767.

Click to expand...

It is going to be difficult to help you without more information. There are many ways do express gains for a PID controller.

My goal is to calculate the process variable mathematically and draw the curve for the pid controller.

Click to expand...

Calculate the process variable or calculate the control variable? I don't under stand what you mean by a PID curve.

The brand name is SMAC actuator and the model number is LAL35-025-51-F. The controller is LAC-1. It is controlling the position.

I want to calculate the current position when the process is under the control that means the process variable.

ramah said:

The brand name is SMAC actuator and the model number is LAL35-025-51-F. The controller is LAC-1. It is controlling the position.

I want to calculate the current position when the process is under the control that means the process variable.

Click to expand...

PID settings are individually set based on how the process you are controlling, is to control the actuator valve.

You mentioned "The range of PID values is 0 to 32767", this to me sounds like your reading the input range of what ever sensor you are using. PID setting don't have SI units and you will be unable to calculate the position of the actuator valve/motor if this is your intention.

If i understand you correctly you want to calculate the position of the actuator valve/motor as it rotates (if its a rotating type) through its range.
If this is the case then the most reliable way instead of calculating is to use an actuator with an inbuilt FEEDBACK pot, this can be used to give an accurate position of the actuator valve/motor, not sure what you want to do with the information but normally i either use it to confirm the correct position of the actuator against what the process controller is saying or feed it back to a PLC.

Please supply more info and what your intentions are.

ramah said:

The brand name is SMAC actuator and the model number is LAL35-025-51-F. The controller is LAC-1. It is controlling the position.

I want to calculate the current position when the process is under the control that means the process variable.

Click to expand...

Here is a link to the manuals for those interested.

http://www.smac-mca.com/manuals.php

It looks like the LMAC-1 does all its math in counts. There are no SI units involved unless scaling is don't externally to convert the counts to SI units.
The proportional gain is unit less or more precisely counts output/counts of error
The integrator gain is counts output/(counts of error*seconds)
The derivative gain is counts output/(counts of error/seconds)
The velocity feed forward is counts output/(target counts/seconds)
The acceleration feed forward is counts output/(target counts/seconds^2)

Since this controller does everything in integer counts the control can be very coarse unless a high resolution encoder is used. Otherwise the control output will change in steps and the motor will always be humming. Also, if you want to move to 100mm there may not be a count that corresponds exactly to 100mm so the controller will probably round down to the next lowest count. You may needs a rounding algorithm to round up if the next count higher is closest to the set point.

The LAC-1 controller has a built-in feedback system for data capture. It can record the motion paths and servo output forces during a cycle. The linear encoder in this actuator is 1 micron.

I'm trying to calculate the current position mathematically and compare them with the data captured

I loaded up a simple test program for the purpose of this test I have the following:

Proportional Gain= 1 (SGn command in the manual and scaling to 0 <= n <= 32767)
Integral Gain=0 (SI command in the manual and scaling to 0 to 32767)
Derivative Gain=0 (SI command in the manual and scaling to 0 to 32767)

Initial position =0
Set point= 2000 (2mm) (MA2000, MAn command in the manual scaling to -2,147,483,647 <= n <= 2,147,483,647 )

Direction: Forward

For data capturing, I set the time interval at which each data sample is
taken to 1 ms and the file is attached.

So the output doesn't depend on the PID algorithms since for all the output is error * Kc

I tried to provide all , if I have forgotten something, tell me.

My goal is to calculate the current position mathematically with the above information regardless of the feedback.

It's somewhat puzzling why you don't want to use the actual feedback but since you've asked the question: You need to be able to separate the P, I and D components of the controller. The I and D components of the output bear no direct relationship to the actual position and will vary depending on the dynamics of the system i.e. Loading, Acceleration etc. The P part of the output will be directly proportional (within limits!) to the position error so the actual position could be determined by Demand position - (P component/P Gain). This method would only work if the P component is not limited.

Nick

I think Ramah is a student that wants to simulate the positions for his PID controller.

I have simulated (crudely) PID loops for software testing. To do this you need to simulate the effects of the PID output on a simulated feedback in real time. The feedback simulation also needs to include anything else that influences it i.e. feed forward/predicted speed. This is easy to do in a crude fashion but a bit more challenging to include the dynamics of the actuator in your feedback simulator.

Nick