Understanding Tuning Parameters

[Potcarb]

I'm trying to understand how to tune a controller. The one I have is PI. I've searched some videos online where it talks about getting the Process Gain, as opposed to the Controller Gain. Is there a way for me to determine the Process Gain without messing with the Controller Gain? Or is there any advice on what I need to do to figure this out?

My confusion is coming from using graphical methods to determine the Process Gain, Dead Time, and Process Time Constant. Do I set the Controller Gain to 0 and see what happens, then get those values from there? Or do I attempt to modify the Controller Gain until I see constant oscillations and call that the Process Gain, then use a correlation for that to get my "true" Controller Gain? Any advice is appreciated, thanks!

 

[thebard3]

Some modern PID controllers have lots of tuning and configuration settings that might be included with yours. Could you post a link or list what controller you are using? Also what is the type of process you are controlling? Otherwise, I will harken back to my experience from 35 years ago of tuning processes. If it really is just a PI controller, then you need to only worry about Proportional band or gain(P), and integral(I). The integral term is often referred to as reset. Most controllers also include derivative action (D) but you do not mention that.
A small proportional band will yield a quick response to increase in the error signal (process value PV - setpoint SP), as will a large gain value. They function inversely but are actually derived differently. Gain is simply a multiplier like the volume knob on an amplifier, where proportional band is a range about SP where the controller output is proportional to the error within. As a rule, neither of those parameters will allow you to maintain a steady process at SP, particularly in a process that has a loss like heating, because an error must be present to generate an output. The integral component is used to periodically add to the controller output to always push the control value CV more toward the direction needed to achieve SP.
Depending on your process, it is usually pretty easy to tune using only PB/gain and Integral. You need to research your device to determine what the constants mean. Determine if the controller uses proportional band or gain. Usually the integral term is defined in seconds per reset but might be resets per minute or something else. Select a value for proportional band/ gain that achieves nearly the SP in an acceptable period of time. Depending on the process, I would normally like to start where the PV approaches SP quickly and maybe overshoots just a little, the settles down to a value somewhere near the SP. It's usually okay if there is a little bit of oscillation (ring)that should disappear after maybe 4 or 5 cycles. Then add the integral constant to bring your PV to SP.
To answer your questions directly;
1- I'd say you don't really need to worry about the process gain unless you are trying to devise tuning parameters using a formula that came with the controller. Process gain is a characteristic of the process that you won't change by tuning.
2- If you set gain to zero, nothing will happen as the controller output is derived in part by multiplying the error signal by the gain. If your controller works on proportional band, then setting that to zero might be akin to multiplying by infinity. There's a little more to it than that, so don't come back and tell me it didn't work exactly like that.

Brad Waybright

The more you know, the more you know you don't know.