MATLAB lost its brain...



  • So, I had to take a test in my college course (in Control Systems) earlier today, and part of that test involved using MATLAB to evaluate step responses on systems. So, I punch the system into MATLAB, step() it to get my step response graph, and get something like:

    back -- which clearly isn't anywhere close to a sane step response, as plotting it with 105 seconds on the X-axis doesn't change the shape of the plot one whit!

    TRWTF is MATLAB, amirite?



  • It's been about a million years since I took control systems, and almost as long since I used MATLAB, but it looks to me like the system is (intentionally or otherwise) an integrator. The step response of an integrator, after the initial rise time, will go linearly toward infinity. (Of course, a real system will eventually hit some sort of limit — power supply rail, run out of fuel, relativistic effects, depending on the nature of the system — and flatten out.)

    Maybe you made a mistake "punch[ing] the system into MATLAB," or maybe the system is supposed to be an integrator. Can't say without knowing more about the system.



  • @HardwareGeek said:

    It's been about a million years since I took control systems, and almost as long since I used MATLAB, but it looks to me like the system is (intentionally or otherwise) an integrator. The step response of an integrator, after the initial rise time, will go linearly toward infinity. (Of course, a real system will eventually hit some sort of limit — power supply rail, run out of fuel, relativistic effects, depending on the nature of the system — and flatten out.)

    Maybe you made a mistake "punch[ing] the system into MATLAB," or maybe the system is supposed to be an integrator. Can't say without knowing more about the system.

    Hrm. The forward transfer function we were given is indeed integrating -- do you have to punch the closed-loop TF into step() to get where you want to go?



  • I suspect so, but that's really a WAG. My experience with MATLAB is very limited, and a long time ago, so I'm the wrong person to ask.



  • I think that the graph in op is where you want to go. Step response means “the response of a system to a step input”, so the output won't necessarily be a step function. The plot you are seeing is a reasonable response for an integrating system that has been given a step function as input. That's my own wild-ass guess, anyway.



  • this title assumes a lot.



  • I found it for you.



  • It's interesting how many organizations — at least two state governments, multiple news outlets, addiction recovery centers and hazardous material cleanup companies — I stopped after 6 pages of GIS results — use exactly the same stock photo (minus the brain and Waldo hat) of a meth lab.



  • someone lost brain in a meth lab - I found it.

    where is my reward?



  • @tarunik said:

    as plotting it with 105 ( ⬅ Oh, look, a bug! :( ) seconds on the X-axis doesn't change the shape of the plot one whit!

    Bigger WAG than H-Geek's:
    Plot it for 1 or 1/2 second.... I think your 0=>1 transition is happening at t=0 (or as close as it can manage). And then (assuming your system is linear) responding appropriately to a continuous input....

    Goodness, I took control 30-some freaking years ago... 30.



  • @ijij said:

    I took control 30-some freaking years ago... 30.

    Wow, you're even older than me!1 :) I think it's probably been somewhere between 25 and 30 for me.

    1 Not necessarily; I didn't start university until I was in my 20s, so we are likely about the same age.



  • I was busy being born 30 years ago.


  • kills Dumbledore

    I was a gleam in the milkman's eye 30 years ago


Log in to reply