Kevin Zhou Physics Olympiad Handouts

Electromagnetism VI: Circuits

AC circuits are covered in chapter 8 of Purcell, or chapter 10 of Wang and Ricardo, volume 2.
Transmission lines, filters, and resonant cavities are covered physically in chapters II-22 and II-23
of the Feynman lectures, which will also build intuition for the next unit. Also see Jaan Kalda’s
circuits handout, an excellent resource which covers nonlinear circuit elements and much more.
This problem set assumes knowledge about linear differential equations covered in M1 and M4,
but you can review the relevant material in chapter 4 of Morin. If you’d like to learn much more
about circuits, from the electrical engineering perspective, a nice book is Foundations of Analog
and Digital Electronic Circuits by Agarwal and Lang. There is a total of 84 points.

1 DC RLC Circuits
Idea 1
AC circuits correspond to driven damped oscillators by the analogies

Q ↔ x, I ↔ v, I˙ ↔ a, L ↔ m, R ↔ b, C ↔ 1/k, V0 ↔ F0 .

More precisely, Kirchoff’s loop equation in an AC circuit immediately becomes Newton’s

second law for a driven damped oscillator upon making these replacements.

Example 1

Consider a circuit with a battery of emf E, a resistor R, and an inductor L in series, with
zero initial current. Find the current I(t) and verify that energy is conserved.

Solution
Kirchoff’s loop equation is
dI
E =L + IR.
dt
To solve for the current, we can separate and integrate, giving
dt dI
=
L E − IR
which yields
E
I(t) = (1 − e−(R/L)t ).
R
At long times, the inductor has no effect, since the current stops changing. To verify energy
conservation we multiply Kirchoff’s loop equation by I, since power is emf times current,
dI
IE = LI + I 2 R.
dt
The left-hand side is the power output by the battery, and the two terms on the right-hand
side represent the rate of increase in energy LI 2 /2 stored in the inductor, and the power
dissipated in the resistor, so all power is accounted for.

1
Kevin Zhou Physics Olympiad Handouts

Example 2

An ideal battery of voltage E is suddenly connected to an ideal capacitor. After a short time,
the capacitor has energy U . How much energy has been released by the battery?

Solution
First let’s consider how the capacitor is charged up over time. Naively, since there’s no
resistance or inductance, the current in the circuit instantly becomes infinite, then instantly
shuts off. This isn’t realistic: to understand what’s actually going on, we have to account for
nonideal features of the circuit, such as resistance or self-inductance. For example, if the
resistance dominates (overdamping), the capacitor charges up monotonically, as in an RC
circuit. If the inductance dominates (underdamping), the capacitor voltage oscillates about
E, until eventually settling down due to the resistance.

At the end, the total energy on the capacitor is

E2
Z Z
VC 1
VC dQ = dVC = = EQ
C 2C 2
where Q is the total charge. But the work done by the battery is
Z
E dQ = EQ

so the battery has released energy 2U . Evidently, half of it is lost, no matter how close to
“ideal” the circuit is! If the tiny resistance dominates, it is lost to heat in the circuit. If
the tiny inductance dominates, then LC oscillations result, and energy can also be lost to
electromagnetic radiation, as covered in E7.

In fact, this is just another example of the nonadiabatic processes you saw in T1. Instantly
attaching a battery is the same kind of thing as instantly dropping a piston, and letting it
bounce until it comes to a stop. Just like those nonadiabatic processes, attaching the battery
in this way creates entropy; if the circuit and environment have temperature T , then
∆Q EQ
∆S = = .
T 2T
We can avoid wasting energy and producing entropy if we use an adjustable battery and
gradually turn its voltage up, slowly enough so that the circuit is always near equilibrium.
This is the electrical analogue of a smooth, adiabatic compression.

Remark
Suppose you wanted to account for the nonideal properties of a capacitor. In principle,
the only way to get the answer exactly is to treat all the fields with Maxwell’s equations
and all the charges with Newton’s laws. But we can often mimic nonideal effects by just
adding resistors to the circuit. This has the benefit of staying within the “lumped element
abstraction”, where we can solve for everything with Kirchoff’s laws, which are much simpler.

2
 Kevin Zhou Physics Olympiad Handouts

But where should we add the resistors? It depends on what nonideal effect we’re trying to
model. For example, if we want to account for the resistance of the wires, we should add a
small resistance in series with the capacitor. But if instead the capacitor is slightly leaky,
we should instead add a large resistance in parallel with the capacitor. If both effects are
important, we should add both. And if radiation is the main way energy is lost, this can’t
really be modeled as a resistor at all, because the amount of radiated power depends on
the rate of change of the current, not the current itself. In some cases, the lumped element
abstraction simply can’t be rescued.

[3] Problem 1 (Purcell 7.46). We have found that in an LR circuit the current changes on the timescale
L/R. In a large conducting body such as the metallic core of the Earth, the “circuit” is not easy
to identify. Nevertheless, we can estimate the decay time. Suppose the current flows in a solid
doughnut of square cross section, as shown, with conductivity σ.

The current is spread out in some way over the cross section.

(a) Make a rough estimate of the resistance and inductance. For the latter, it may be easiest to
estimate the magnetic field at the center of the donut first, then use that to estimate the total
magnetic field energy.

(b) With these results, show that τ ∼ µ0 a2 σ, which also follows from dimensional analysis.

(c) Given that the radius of the Earth is r ∼ 3000 km and σ ∼ 106 (Ω · m)−1 , estimate τ .

[2] Problem 2 (PPP 171). A circuit contains three identical lamps (modeled as resistors) and two
identical inductors, as shown.

The switch S is closed for a long time, then suddenly opened. Immediately afterward, what are the
relative brightnesses of the lamps?

[3] Problem 3 (Kalda). A capacitor C and resistance R are connected in series. Rectangular voltage
pulses are applied, as shown below.

3
Kevin Zhou Physics Olympiad Handouts

After a long time, find the average power dissipated on the resistor if (a) T  RC and (b) T  RC.

Remark
Out of all the analogies mentioned above, which is capacitance defined “backwards”, so
that C ∼ 1/k? I actually have no idea, but one possibility is that large quantities should
intuitively correspond to large objects. An object has to be physically large (and thereby
expensive) to have a high C or a high L, and you can easily see this on your circuit board.
(Of course, this doesn’t explain everything; the largest R you can get is just a break in the
circuit, which is neither large nor expensive.)

Another difference in the analogies is that for circuits we usually measure I(t), while for
mechanical oscillators we usually measure x(t).

We now consider some problems involving mutual inductance.

Example 3: Griffiths 7.57

Two coils are wrapped around a cylindrical form so that the same flux passes through every
turn of both coils, i.e. so that the mutual inductance is maximal. In practice this is achieved
by inserting an iron core through the cylinder, which has the effect of forcing the magnetic
flux to stay inside the cylinder.

The “primary” coil has N1 turns and the secondary has N2 . If the current I in the primary
is changing, show that the emf E2 in the secondary obeys
E2 N2
=
E1 N1
where E1 is the (back) emf of the primary.

4
 Kevin Zhou Physics Olympiad Handouts

Solution
Let Φ be the flux through a single loop of either coil due to the current in the primary. Then

Φ1 = N1 Φ, Φ2 = N2 Φ.

By Faraday’s law,
dΦ dΦ
E1 = −N1 , E2 = −N2
dt dt
which gives the desired result. This is a primitive transformer, a device for raising or lowering
the emf of an alternating current source. By choosing the appropriate number of turns, any
desired secondary emf can be obtained.

We can also solve this problem more formally using what we know about inductance, which
will also tell us what happens when both currents are nonzero. The emfs obey
dI1 dI2 dI2 dI1
E1 = −L1 −M , E2 = −L2 −M .
dt dt dt dt
We showed in E5 that Li = µ0 Ni2 πR2 /H for a cylindrical solenoid. (Here, H stands for the
length of the iron core, since this is the length over which the magnetic field exists.) As you’ll
show below, the maximum possible value of the mutual inductance, which is achieved by this
√
ideal transformer, is L1 L2 . Plugging in these results gives

µ0 πR2 µ0 πR2
     
2 dI1 dI2 2 dI2 dI1
E1 = − N1 + N1 N2 , E2 = − N2 + N1 N2 .
H dt dt H dt dt

This tells us the desired result holds for any values of the dIi /dt.

This result is not surprising from the standpoint of Faraday’s law. The flux change through
any cross-section of the iron core is the same, so the induced emf around any circle around
it is the same. Thus, the emf per turn is the same between the coils, E1 /N1 = E2 /N2 , which
again gives the desired result.

[3] Problem 4. Consider two inductors Li , with mutual inductance M .

(a) Show that if the inductors have currents Ii , the total stored energy is
1 1
U = L1 I12 + L2 I22 + M I1 I2 .
2 2
√
Use this result to show that |M | ≤ L1 L2 .

(b) Suppose these two inductors are in series. Find their combined effective inductance.

(c) Suppose these two inductors are in parallel. Find their combined effective inductance.

[3] Problem 5 (Kalda). An electrical transformer is connected as shown.

5
Kevin Zhou Physics Olympiad Handouts

Both windings of the transformer have the same number of loops and the self-inductance of both
coils is equal to L. There is no leakage of the magnetic field lines from the core, so that the mutual
inductance is also equal to L.
(a) Find the current in both loops immediately after the switch is closed.
(b) Find the currents as a function of time.

2 AC RLC Circuits and Impedance

Idea 2: Impedance

Current and voltage can be promoted to complex quantities,

V (t) = V0 cos(ωt + φ), Ṽ (t) = Ṽ0 eiωt , Ṽ0 = V0 eiφ

where the physical quantity is the real part. This is useful because we can relate Ṽ and I˜ in
˜ where Z is the impedance, and
all cases by Ṽ = IZ
1
ZR = R, ZC = , ZL = iωL
iωC
for the three common circuit elements. Impedance is extremely useful for finding the steady
state response of a circuit. If you’re interested in the transients, you can find them by
applying the techniques of M4 to the Kirchoff’s loop rule equation.

Idea 3: Power
Turning parameters complex and taking the real part works because we’re dealing with
linear equations. As a result, it doesn’t work for energy or power, which are quadratic.

For instance, the power is not simply the real part of I˜Ṽ , but rather
˜ Re(Ṽ ) = I0 V0 cos(ωt) cos(ωt + φ)
P = IV = Re(I)
where φ is the phase angle of Z. To compute the average power, note that
V02
P = cos(ωt)(cos(ωt) cos(φ) − sin(ωt) sin(φ)).
|Z|
The second term averages to zero, while cos2 (ωt) averages to 1/2 as usual, so
1 V02 1
P = cos(φ) = I0 V0 cos(φ).
2 |Z| 2

6
Kevin Zhou Physics Olympiad Handouts

We can decompose a general impedance as Z = R + iX, in which case cos φ = R/|Z|, and

1 I0 V 0 R 1
P = = I02 R.
2 |Z| 2
2
It’s conventional to define Irms = I02 /2 to be the average value of I 2 , giving
2
Vrms
2
P = Irms R= .
R

Example 4

Find the magnitude of the current through a series RLC circuit with AC voltage source
V0 cos ωt.

Solution
We promote the voltage and current to complex numbers,

V (t) = V0 eiωt .

Kirchoff’s loop rule (subject to the caveats in E5) is

Q
LI˙ + IR + = V0 eiωt .
C
This is quite similar to a damped driven harmonic oscillator, except that we want to get I(t),
rather than Q(t). To get the steady state behavior, we guess

I(t) = I0 eiωt .

Then we have
˙ = (iω)I0 eiωt , 1
I(t) Q(t) = I0 eiωt .
iω
Plugging this in, we find  
1
iωL + R + I0 = V 0 .
iωC
Solving for the magnitude of the current gives

|V0 | |V0 |
|I0 | = =p
|iωL + R + 1/iωC| R2 + (ωL − 1/ωC)2
√
which is maximized when ω = 1/ LC, as we saw in M4. We could also have gotten straight
to this last step by just using complex impedances.

7
 Kevin Zhou Physics Olympiad Handouts

Example 5

An imperfect voltage source consists of an ideal AC voltage source in series with an impedance
ZS . It is attached to a load of impedance ZL . What value of ZL maximizes the power
transferred to the load?

Solution
Write the impedances as ZL = RL + iXL . When the impedances are purely real, it’s a
familiar fact that the optimum is at RS = RL . We consider the case of general impedance
here to illustrate how to work with power. First, the current has amplitude

|V |
I0 = .
|ZS + ZL |

The average power dissipated in the load is

1 RL RL
P = I02 RL ∝ 2
= .
2 |ZS + ZL | (RS + RL ) + (XS + XL )2
2

The denominator is minimized when XS + XL = 0, so the optimal real part is RL = RS by

the same logic as the purely real case. Thus, the highest power is achieved for ZL = ZS∗ .

[1] Problem 6. Consider the cube of resistances R, capacitances C, and inductances L shown below.

Compute the impedance between the terminals.

[3] Problem 7. Consider an RLC circuit with a driving V (t) = V0 eiωt .
(a) Suppose the resistor, inductor, and capacitor are connected in parallel. Sketch the current
|I0 | through the driver as a function of ω, and compare it to the result for a standard series
RLC circuit. Can you give a qualitative explanation for the difference?

(b) As we saw in M4, the quality factor for an oscillator quantifies how fast the energy in an
undriven oscillator decays away. Specifically,
average energy stored in the oscillator
Q= .
average energy dissipated per radian
Find the quality factor for a series RLC circuit, and confirm your answer has correct dimensions.

(c) What is the condition for a series RLC circuit to be overdamped?

(d) Find the quality factor for a parallel RLC circuit. You should find that the quality factor
increases as R is increased – why does this make sense?

8
 Kevin Zhou Physics Olympiad Handouts

Example 6: NBPhO 2018.7

Consider the following AC circuit.

The voltage difference between D and E has amplitude VDE = 7 V. Similarly, VDF = 15 V
and VEF = 20 V. What is the magnitude of V0 ?

Solution
Treat the voltages as phasors, and let VA = 0, so VB = V0 . Then VAF is perpendicular to
VF B , which means that VF lies on the circle centered at V0 /2 with radius V0 /2. The exact
same logic applies for VE and VD . Therefore, we know that VDE , VDF and VEF form the
three sides of a triangle, where the desired answer is the diameter of its circumcircle.

For a triangle with side lengths a, b, and c, and a circumcircle of radius R, Heron’s formula
states that the area is
p 1
A= p(p − a)(p − b)(p − c), p = (a + b + c).
2
The area is also given by
abc
A= .
4R
Solving for the diameter, we have
abc
2R = p = 25 V.
2 p(p − a)(p − b)(p − c)

These geometric steps aren’t that important; the key idea is thinking in terms of phasors.

[2] Problem 8 (Purcell 8.26). The four curves shown below are plots, in some order, of the applied
voltage and the voltages across the resistor, inductor, and capacitor of a series RLC circuit.

9
 Kevin Zhou Physics Olympiad Handouts

Which is which? Whose impedance is larger, the inductor’s or the capacitor’s?

[2] Problem 9 (PPP 170). Consider each of the following circuits.

In each √
case, find the amplitude of the current drawn from the source as a function of ω/ω0 , where
ω0 = 1/ LC.

[3] Problem 10 (BAUPC). Consider the following RLC circuit.

The capacitors
p have capacitance C, the inductors have inductance
√ L, and the resistors have resistance
R = L/C. Furthermore, the driving frequency is ω = 1/ LC. Find the amplitude of the total

01W
current through the circuit.

[3] Problem 11. USAPhO 2002, problem A1.

[3] Problem 12. 01mƒ USAPhO 2011, problem B1.

10
 Kevin Zhou Physics Olympiad Handouts

3 Electrical Engineering
These next problems are about using RLC circuits for practical purposes. They don’t require
anything not already introduced in the previous section, but they represent a different way of
thinking that it’s crucial to get comfortable with.

[2] Problem 13 (Feynman). In electronic circuits it is often desired to provide a sinusoidal voltage of
constant amplitude but variable phase. A circuit which accomplishes this is called a phase-shifting
network. One example is shown below.

Show that the voltage measured between terminals A and B has half the amplitude of the input
voltage, and a phase which may be adjusted by changing the resistance R0 .

[3] Problem 14 (Kalda). The figure below shows a Maxwell’s bridge, which is used for measuring the
inductance L and resistance R of a coil.

To do this, the frequency ω is fixed and the known parameters R1 , R2 , RC , and C are adjusted
until the voltmeter reads zero. Once this is done, find R and L in terms of the other parameters.

[5] Problem 15. An alternating voltage V0 cos ωt is applied to the terminals at A. The terminals at
B are connected to an audio amplifier of very high input impedance. (That is, current flow into the
amplifier is negligible.)

This circuit is the most primitive of “low-pass” filters.

(a) Calculate the “gain” ratio |Ṽ1 |/V0 in this filter. Show that for sufficiently high frequencies,
the signal power is reduced by a factor of 4 for every doubling of the frequency.

(b) Design a low-pass filter without using a capacitor.

(c) Design a high-pass filter.

11
 Kevin Zhou Physics Olympiad Handouts

(d) Design a stronger low-pass filter, i.e. one which reduces the signal power by a greater factor
for every doubling of the frequency.

(e) Design a band-pass filter, which suppresses both low and high frequencies, but has a constant
gain for a wide range of medium frequencies. (It’s okay if the constant gain is less than 1, as
we can just pass the output through an amplifier.)

(f) Design a notch filter, which suppresses a very small range of frequencies, while letting all other

01@
frequencies through.

[4] Problem 16. IPhO 1984, problem 3. A nice, short problem on filters.
[2] Problem 17. Consider the same setup as problem 15, but with the resistor and capacitor switched.
(a) Assuming that V1  V0 , show that the output voltage is proportional to the derivative of the
input voltage. Hence the circuit is a differentiator. (Can you relate this to the kind of filtering
such a setup does?)

(b) Design a circuit whose output is proportional to the integral of V0 , again assuming V1  V0 .
[3] Problem 18. A resonant cavity of the form illustrated below is an essential part of many microwave
oscillators. It can be regarded as a simple LC circuit.

(a) Assuming that s  a, b, h, estimate the lowest resonant frequency of the cavity by treating it
as an LC circuit. It may be helpful to sketch the magnetic and electric fields.

(b) One of the most common types of cavity is a cylindrical cavity, i.e. a hollow cylinder. (It
corresponds to taking s = h in the above setup.) Assuming that h ≈ b, find a reasonable
estimate of the lowest resonant frequency ω.

Remark
In E3, we saw that for DC circuits, any system of resistors and ideal batteries with two
ports is equivalent, from the perspective of anything connected across the ports, to either
a single resistor and ideal battery in series (the Thevenin equivalent), or a single resistor
and ideal current source in parallel (the Norton equivalent). From the ideas covered in this
problem set, we also know that any system of resistors, inductors, and capacitors with two

12
Kevin Zhou Physics Olympiad Handouts

ports is equivalent, at a fixed frequency ω, to a single lumped element with impedance Zeq .
This in turn could be constructed out of a single resistor and inductor or capacitor in series.

This naturally leads to a more general question: it is possible to construct a simple

“equivalent” circuit that has exactly the same Zeq (ω), for all ω? The answer is yes. For
example, consider the simple case of a circuit of only inductors and capacitors. Here’s the
rough idea: in this case, the equivalent impedance is always a pure imaginary, rational
function of ω, meaning a ratio of two polynomials in ω. But rational functions can always
be expanded in partial fractions. Assuming no multiple roots for simplicity, each term in the
partial fraction decomposition can be mimicked with an LC circuit, and we get the sum by
placing these circuits in series.

In electrical engineering, the general task of constructing a circuit with a prescribed Z(ω) is
called network synthesis; the above example is called Foster’s synthesis. These techniques
can be used to construct filters more elaborate than the ones you explored in problem 15.

Remark
Power companies often transmit electricity with “three-phase power”. This means that there
are three “hot” electrical lines, carrying voltages

V1 (t) = V0 cos(ωt), V2 (t) = V0 cos(ωt + 2π/3), V3 (t) = V0 cos(ωt + 4π/3).

Depending on your home, you might be able to connect to this three-phase power with
special outlets, to use power tools. There are several advantages to three-phase power, but
one is that it supplies a constant power, as V12 + V22 + V32 is constant.

This shouldn’t be confused with the three holes in an ordinary American wall outlet. In
these outlets, one of the eyes is the “hot” one, with voltage V1 (t), while the other eye and
the mouth are both grounded. Appliances are powered by the voltage difference between
the eyes. Appliances that use significant power and have metal exteriors have three-prong
plugs. Here, the grounded “mouth” hole is connected directly to the exterior of the appliance,
ensuring that it can’t shock you, even if something goes wrong inside.

4 Normal Modes
Idea 4
A circuit with n independent loops has n normal modes. If we ignore resistances, the normal
modes are pure sinusoids, though in all real circuits they exponentially damp over time. Just
as in mechanics, the general solution for the behavior of a driven circuit is a superposition of
normal mode currents and the response to the driving.

There are many ways to find the normal mode frequencies.

13
Kevin Zhou Physics Olympiad Handouts

• One way is to pick any two points not directly connected by wires. We may imagine that across
these points we have attached a current source I˜ which is doing nothing, I˜ = 0. If a normal
mode is present at frequency ω, then we can have Ṽ =6 0, even though I˜ = 0 because current is
merely sloshing around inside the circuit. Thus, the equivalent impedance Z(ω) between these
points is infinite.

• Another way is to pick two points directly connected by wires. We may imagine this wire is
actually a voltage source Ṽ which is doing nothing, Ṽ = 0. If a normal mode is present at
frequency ω, then we can have a current I˜ 6= 0 through the wire even though Ṽ = 0, so the
equivalent impedance Z(ω) between these points is zero.

• Some LC circuits can be mapped to sets of masses and springs using the analogies in idea 1,
which can help with guessing the normal modes.

• Finally, one may simply write down all of Kirchoff’s loop equations, plug in eiωt time dependence,
and look for a solution. This boils down to solving a system of n equations, or equivalently
evaluating the determinant of an n × n matrix. This is rarely the best approach on an Olympiad.

• Not every problem benefits from using normal modes; for relatively simple circuits with special
initial conditions, it may be better to solve Kirchoff’s loop equations directly.

Example 7: Kalda 89

Find the normal mode frequencies of the circuit below.

Solution
There are four independent Kirchoff’s loop equations, so we expect four normal modes.
One normal mode consists of current simply flowing uniformly along the outside, along the
inductors. Since the capacitors aren’t involved, this normal mode has ω0 = 0.

Now we apply the first technique listed above: we pick two points not directly connected
with wires, and set the impedance to infinity. By symmetry, it’s best to pick A and D.
By symmetry, if any voltage is applied between A and D, the points B and E will be at
the same voltage. Furthermore, this point will be at the same voltage as O, because the

14
 Kevin Zhou Physics Olympiad Handouts

remaining circuit forms a balanced Wheatstone bridge, as introduced in E3. Identifying B,

E, and O straightforwardly gives a simple LC √circuit with Leff = (3/2)L and Ceff = (2/3)C,
√
and resonant frequency ω1 = 1/ Leff Ceff = 1/ LC.

This procedure only gave one of the three remaining normal modes, so we must have missed
the other two because they have zero voltage difference between A and D. Therefore, to find
the other two, we can join A and D, leading to the simpler equivalent circuit below.

We now apply the same procedure between points B and E. This circuit is again a balanced
Wheatstone bridge, so O and A are at the same
p voltage. We then have a simple LC circuit
with Leff = (4/3)L and Ceff = C, giving ω2 = 3/4LC.

Again, we have missed a normal mode, so that remaining mode must have zero voltage
difference between B and E. Joining them together leads to the final equivalent circuit
below.

This is now a simple LCpcircuit with Leff = (1/3)L and Ceff = (12/7)C, giving the final
resonant frequency ω3 = 7/4LC.

[2] Problem 19 (Kalda). Consider the LC circuit below.

15
 Kevin Zhou Physics Olympiad Handouts

√ √
Show that the normal mode frequencies are ω = ( 5 ± 1)/2 LC.

[3] Problem 20 (IPhO 2014). Initially, the switch S is open in the circuit shown below.

The capacitor with capacitance 2C is given a charge q0 , and immediately begins to discharge. At
the moment when the current through the inductors reaches its maximum value, the switch S is
closed. Find the maximum current through the switch thereafter.

[5] Problem 21 (Physics Cup 2012). Find the frequencies of the normal modes of the circuit below,
where C1  C2 and L1  L2 .

You may give all of your answers to lowest order in C1 /C2 and L1 /L2 .

5 Nonlinear Circuit Elements

In this section we’ll introduce nonlinear circuit elements, focusing on diodes. More exotic circuit
elements will be covered in E7.
Idea 5
Many nonlinear circuit elements can be described by a current-voltage characteristic I(V ).
Such circuit elements have trivial time dependence, just like resistors, and working with
them basically amounts to using Kirchoff’s laws as usual, plugging in I(V ) where necessary.

Since the implementation details of such elements can be very complicated, and many draw
power from external sources, it generally isn’t productive to think of them “physically”; they
are more like miniature computers than physical objects. One just has to take I(V ) as given
and work directly with it. Some simple examples are:

16
Kevin Zhou Physics Olympiad Handouts

• An ideal diode acts like a wire in one direction and a break in the other, so it has
(
∞ V > 0,
I(V ) =
0 V < 0.

• Sometimes one instead takes the I(V ) characteristic

(
∞ V > V0 ,
I(V ) =
0 V < V0

which means that it “costs” voltage V0 to go through the diode in the forward direction.
More realistically, I(V ) smoothly increases when V passes V0 , but you don’t often see
this in Olympiad problems because it makes the math very messy.

• Zener diodes are bidirectional diodes. An ideal Zener diode has


∞
 V > V0 ,
I(V ) = 0 −V0 < V < V0 ,

−∞ V < −V0 .


• Many familiar objects such as fuses (wires which break when I passes a threshold)
and spark gaps (breaks that conduct when V passes a threshold) can be thought of as
nonlinear circuit elements in the same way.

Analytically, these three cases are easily handled by casework. For instance, a diode acts just
like a wire for positive V , and just like a break for negative V . In each case, the circuit is no
more complicated than an ordinary one with linear circuit elements. Then you put the cases
together to get the full behavior.

Example 8

A capacitor of capacitance C is charged so that its voltage is VC . The capacitor is placed in

series with a resistor R and a diode with I(V ) characteristic
(
∞ V > V0 ,
I(V ) =
0 V < V0 .

The diode is oriented so that the initial voltage across it is positive. What happens next?

Solution
We use casework. If Vc < V0 , the voltage on the capacitor is not enough to get current to
flow through the diode, so nothing happens. If Vc > V0 , current flows, and the diode acts like
a battery of emf V0 oriented in the opposite direction. This is just a discharging RC circuit,

17
 Kevin Zhou Physics Olympiad Handouts

so the capacitor’s voltage is

V (t) = (VC − V0 )e−t/RC + V0 .

After a long time, the voltage on the capacitor falls to V0 and current stops flowing.

Idea 6
It is difficult to solve a nonlinear circuit analytically if I(V ) is not very simple. In these cases:

• One can find the answer graphically as the intersection of I(V ) and another curve.

• One can solve for the answer iteratively on a calculator.

• If V stays within a narrow range, one can take a linear approximation to I(V ). This
effectively replaces the element with a battery in series with a resistor, so the problem
can be solved just like those in E3.

[2] Problem 22 (Kalda). Find the current in the circuit given below.

The nonlinear element is a diode with the I(V ) characteristic shown.

Idea 7
The power delivered to any circuit element is still P = IV . However, some nonlinear circuit
elements can be active, providing net power to the circuit, like batteries.

Example 9: Kalda 64

The circuit below containing an ideal diode makes it possible to charge a rechargeable battery
of voltage E = 12 V with a direct voltage source of a voltage V0 = 5 V < E.

18
 Kevin Zhou Physics Olympiad Handouts

To do this, the switch K is periodically opened and closed, with the opened and closed periods
having equal length τ = 10 ms. Find the average charging current assuming L = 1 H.

Solution
The intuition here is that, using an inductor and a switch, one can generate emfs larger
than what we put in, because the current wants to keep flowing through the inductor
when the switch is opened; this allows us to get enough emf to charge the battery.
This idea is also used in the ignition coils of old-fashioned cars, where a voltage large
enough to ionize air is produced, making a spark and starting the engine. There’s
also a fluid analogue, called the hydraulic ram, used to raise water. The point of the
diode here is just to keep current from flowing the other way during the other half of the cycle.

When the switch is closed, no current can flow through the battery, and the current through
the inductor builds up linearly, since there is an emf V0 across the inductor. When the switch
is opened, the emf across the inductor is V0 − E = −7 V, causing its current to decrease
while simultaneously charging the battery. After a time (5/7)τ with the switch open, the
current through the inductor falls to zero, and the diode causes current to stop flowing.

Quantitatively, while the switch is closed, the current through the inductor builds up to Eτ /L.
When the switch is open, current flows for a time (5/7)τ , linearly falling to zero, so the total
charge is
1 Eτ 5
Q= τ.
2 L 7
A cycle takes time 2τ , so
Q 5 Eτ
I= = = 21 mA.
2τ 28 L
By the way, your phone and laptop chargers probably have rectangular bricks containing a
switched-mode power supply. This consists of one part that converts the AC wall power to
DC, and a second part similar to the circuit above, but set up to output a lower DC voltage.
You could also use a transformer to lower the AC voltage, but a switch-mode power supply
is more space-efficient, and it easily copes with a range of input AC voltages and frequencies.

[3] Problem 23 (Kalda). An alternating voltage V = V0 cos(2πνt) is applied to the leads of the circuit
shown below. Treat the diode as ideal.

Assuming the current in the inductor begins at zero, what is the average current through the
inductor at late times?

[3] Problem 24. EFPhO 2010, problem 9.

19
 Kevin Zhou Physics Olympiad Handouts

[3] Problem 25. EFPhO 2008, problem 6.

[3] Problem 26. EFPhO 2013, problem 8. This one has a nice mechanical analogy.

[3] Problem 27. 01Y‚ IPhO 2001, problem 1c.

[3] Problem 28. 01^‚ USAPhO 2018, problem A2.

[4] Problem 29. 01@ EuPhO 2022, problem 2. A nice application of casework.

20