I think I have a good understanding of differential equations, so I would like to first try to understand my system in terms of differential equations before transforming my system into the more standard Laplace-transform, looking-for-poles-type analysis that's standard in control theory.

Specifically I'm interested in finding the differential equation that governs what is known as a "Pound-Derver-Hall (PDH) lock" in optics.

In this locking scheme, an optical system is controlled with a piezo, and the goal is to have to use the piezo to correct for changes in the optical system. The system is set up like this:

enter image description here

A voltage is sent to the piezo in the mechanical-optical system. An error voltage is generated from the output of this system, and is sent to the PID controller. The error voltage looks something like this:

enter image description here

So, if you asked me, my differential equation describing this would be the form:

$\frac{dVin}{dt} = k_p V_{error} + k_i \int V_{error} + k_d\frac{d}{dt}V_{error}$

so in the area that my error function is strictly linear:

$V_{error} = g V_{in}$

then this becomes:

$\frac{dVin}{dt} = g k_p V_{in} + g k_i \int V_{in} + g k_d\frac{d}{dt}V_{in}$

Now what's confusing to me is that when I consider only proportional gain (ie $k_i = k_d = 0$)

Then I can solve my diff.eq. exactly:

$V_{in} = V_0 e^{g k_p t}$

which has a "sink" at zero as t approaches infinity.

Now I know from experience that "position only" controllers typically have lots of oscillations and don't necessarily converge to the solution but oscillate around it. Now if I had an "inertia" term to my differential equation that added a second derivative to the differential equation, then I would get a sinusoidal solution that I recognize as the typical "oscillations" in position-only control.

But generally speaking, is there a way to infer what this "inertia term" is?

Am I making some kind of mistake in my model?


1 Answer 1


is there a way to infer what this "inertia term" is?

Regarding that part of the question, (although you already are outlining it in your question) I would suggest that the highest derivative is usually assosiated with the inertia term. So in your case the $k_d$ which is conviniently set to zero for the P-controller.

The other two examples that spring to mind regarding inertia terms are:

  • mass damper spring (Harmonic oscillator): the mass in the system which is associated with $\frac{d^2 }{dt^2}x(t)$
  • RLC system: coil inductance L which is associated with $\frac{d^2 }{dt^2}i(t)$
  • $\begingroup$ So you think that the inertia term is specifically encoded in the error function? There's no extra terms in the differential equation? $\endgroup$ Commented Sep 20, 2020 at 17:08
  • $\begingroup$ Exactly. If you think about it, inertia is about a system's resistance to change its condition. The $k_d$ term is resisting changes... $\endgroup$
    – NMech
    Commented Sep 20, 2020 at 19:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.