Step response questions

[kpower]

Hello,

Wonderful software package, appreciate the free availability.  I would have included a link for you, but am getting the following error:

Failed to fetch existing project.
Error code: XMLHttpRequest failed: “Unable to connect to the Parse API”

I think it would be easy enough to explain my issue.

Modeling a 2nd order plant with the K,w,z values exactly like what PID tuner shows.  Our plant models are in complete agreement (I am using GNU Octave).

Octave can also use the standard model PID, so I am fairly certain our controllers are in agreement.  When I close the loop and run step tests, Octave and PIDtuner are providing the same output response on PI controllers.  So far so good.

However, when I begin to add significant amounts of D to the control, the step responses differ.  No D filters are modeled.  Is this a problem with my setup, with Octave, or with PIDtuner?

All pertinent output from Octave is shown below, and I have attached a screen shot of the step response output.  It would be simple to force PID tuner to these same values for the plant and PID.

 

K = 3.7600
w = 1.2078
z = 2.3056

Transfer function ‘Plant’ from input ‘u1’ to output …

5.485
y1: ———————
s^2 + 5.569 s + 1.459

Continuous-time model.
P = 30
Ti = 10
Td = 0.1000

Transfer function ‘Control’ from input ‘u1’ to output …

30 s^2 + 300 s + 30
y1: ——————-
10 s

Continuous-time model.

Transfer function ‘ClosedLoop’ from input ‘u1’ to output …

164.5 s^2 + 1645 s + 164.5
y1: ———————————–
10 s^3 + 220.2 s^2 + 1660 s + 164.5

Continuous-time model.

[pidtuner]

Hi,
Thanks for notifying the Error code. That needs fixing indeed.
Regarding the difference between simulations, maybe this other question is the cause:

PID Implementation when simulating

Basically, the PID implementation applied the D gain on the measurement, not on the error.

[kpower]

Perfect, that was it.

Modified the Octave model around for DoPV, and both tools are in agreement.

Thank you!

[Author]

Posts