PID Control: Derivative Action

Proportional, Integral, Derivative (PID) controllers are a common Single Input, Single Output (SISO) method to regulate a process. This discussion delves into the derivative term in additional detail.

[saeedmardani]

Dear John, I don't know if you see this post or not, but, I got a job after watching your videos related to control. I am very thank ful to you because of your great experience in the field of control and your experience in teaching. Thank you very much! :)

[dashypookk]

I teach process engineering and this video has been great for the process control part of my process design lectures. Thank you!

[chriscordingley]

Thanks. Taking the set-point out of the derivative term to reduce those spikes was new to me. (Never too old to learn) I'll give that a shot.

[pnachtwey]

Measurement noise is often not noise at all but quantizing error due to lack of resolution and sample jitter where the samples are not taken at regular intervals.

[dhiaelhakrouabah]

My only problem ..the only thing that i didnt understand about derivitive is how Mathematically its acting on our output signal (damping the oscilations) given that its assessing the rate of change of error?
Btw you provided very fruiful insight and clearer explanation that ive ever seen by far ..im thankful for that .

[HarshitSharma-gcjb]

Great explanation as usual.
I have a question related to setting up 'D' in a PID controller.
When we follow direct synthesis process with manual bump test, I understand that 'P' & 'I' are related to 'Process Gain' and 'Process Settling time' respectively.
How do we decide values for 'D'?

[pnachtwey]

There are a few things I would like to add.
1. It is possible to have the proportional gain act on changes of the PV too. I call this a I-PD because the proportional and derivative gain only act on the feed back, not the error.
2. Obviously, the derivative kick is not desired but there is more to the problem than the derivative kick. The gains in the forwards path, those that act on error instead of the PV changes, add zeros to the closed loop transfer function. This can be good or bad depending on where the closed loop zeros are. If one is very clever, one can place both the poles and zeros to get the optimal response. It is is also possible to place the zeros to get the effect of a notch filter. This could be good if you know where the notch filter is placed. It can be bad if one is not aware of where the notch frequency is.
3. Not all open loop systems require a derivative gain. A closed loop system system should have one gain for every open loop pole. The derivative gain does not count because it has it;s own pole.

[abdullahannaijjar]

Please what’s the difference between PID and QFT controllers? They all control the same Electro-hydrostatic Actuator position!
Please any answer!