A radio, which we feel obligated to characterize as "an old-fashioned"
radio, receives a broadcast signal which can then be heard by sending it
to a speaker system to produce sound. There are, of course, many
signals being broadcast and which are received by the radio, and which
signal is used by the radio determined by the tuner in the radio,
which selects the carrier wave for the signal (or station) one is trying
to listen to. A very simple model of such a tuner is the RLC circuit
shown below.
Here the input voltage \(E(t)\) is the signal that is being received by
the tuner. \(R\), \(L\) and \(C\) are the resistance, inductance and
capacitance in the circuit. We can adjust the capacitance \(C\) to
select one frequency in the input—that is, by changing \(C\) we
can make the (long-term) behavior of the system mirror only one
of the frequencies in the input function \(E(t)\).
Let \(I\) be the current in the circuit and \(Q\) the charge on the capacitor. Then Kirchoff's second law (that the sum of the voltage drops in the circuit must equal the input voltage) and the elementary laws of electricity give \[ L\,\frac{dI}{dt} + R\,I + \frac1{C}\,Q = E(t), \] so that \[ L\,I'' + R\,I' + \frac{1}{C}\,I = E'(t) \] (see, for example, [1] or [2] for a fairly complete and fairly succinct derivation of this equation, respectively).
Following [1], we might take \(L = 1\)H (Henry, the unit of inductance) and \(R = 0.1\Omega\) (Ohms, the unit of resistance). Then, we may suppose that the input voltage is a combination of several carrier signals with different frequencies; for example, we might take \(E(t) = -\cos(t) - \frac45\cos(5t)\). For initial conditions, if there is initially no current, \(I(0) = 0\) and \(I'(0) = (E(0) - q_0/C)/L\). The question for the tuner is how to pick \(C\) so that \(I(t)\) looks like only one of the terms in the input \(E(t)\).
Of course, this requires that the current actually look like the input (or part of the input). The Method of Undetermined Coefficients argues that this is the case, and in these demonstrations we consider that correspondence. To do this initially it may be instructive to consider a further simplification of the model, in which we take a single input function, \(E(t) = -\cos(\omega t)\), and consider what the resulting current \(I(t)\) looks like. Because we know that this will be composed of homogeneous and particular solutions, we look at both of these. And because we are initially interested in considering how the Method of Undetermined Coefficients works, we may consider other input functions than oscillatory ones. We will return to the question of tuning the circuit to one input frequency in a subsequent demonstration.
A collection of Matlab demos for this:
Some questions that may be worth considering: