System Analogies
System Analogies
4: Analogies
CCBYNCSA
2.3: Purpose of Abstraction
picture_as_pdf
Downloads
Submit Adoption Report
Peer Review
Donate
Sanjoy Mahajan
Olin College of Engineering via MIT OpenCourseWare
Because abstractions are so useful, it is helpful to have methods for making them.
One way is to construct an analogy between two systems. Each common feature leads
to an abstraction; each abstraction connects our knowledge in one system to our
knowledge in the other system. One piece of knowledge does double duty. Like a
mental lever, analogy and, more generally, abstraction are intelligence amplifiers.
clipboard_ef86f9feeee3a5bfdaa9068fded1394e2.png
In the circuit, the voltage source—the Vin on its left side—supplies a current that
flows through the inductor (a wire wrapped around an iron rod) and capacitor (two
metal plates separated by air). As current flows through the capacitor, it alters
the charge on the capacitor. This “charge” is confusingly named, because the net
charge on the capacitor remains zero. Instead, “charge” means that the two plates
of the capacitor hold opposite charges, Q and -Q, with Q ≠ 0. The current changes
Q. The charges on the two plates create an electric field, which produces the
output voltage V out equal to Q/C (where C is the capacitance).
For most of us, the circuit is less familiar than the spring–mass system. However,
by building an analogy between the systems, we transfer our understanding of the
mechanical to the electrical system.
clipboard_edd12f3867ef3a8878c622e667048171f.png
Let’s build more analogy bridges. The derivative of velocity, which is the second
derivative of position, is acceleration (a). Therefore, the derivative of current
(dI/dt) is the analog of acceleration. This analogy will be useful shortly when we
find the circuit’s oscillation frequency.
These variables describe the state of the systems and how that state changes: They
are the kinematics. But without the causes of the motion—the dynamics—the systems
remain lifeless. In the mechanical system, dynamics results from force, which
produces acceleration:
a=Fm(2.4.1)
Acceleration is analogous to change in current dI/dt, which is produced by applying
a voltage to the inductor. For an inductor, the governing relation (analogous to
Ohm’s law for a resistor) is
dIdt=VL(2.4.2)
where L is the inductance, and V is the voltage across the inductor. Based on the
common structure of the two relations, force F and voltage V must be analogous.
Indeed, they both measure effort: Force tries to accelerate the mass, and voltage
tries to change the inductor current. Similarly, mass and inductance are analogous:
Both measure resistance to the corresponding effort. Large masses are hard to
accelerate, and large-L inductors resist changes to their current. (A mass and an
inductor, in another similarity, both represent kinetic energy: a mass through its
motion, and an inductor through the kinetic energy of the electrons making its
magnetic field.)
Turning from the mass–inductor analogy, let’s look at the spring–capacitor analogy.
These components represent the potential energy in the system: in the spring
through the energy in its compression or expansion, and in the capacitor through
the electrostatic potential energy due to its charge.
Force tries to stretch the spring but meets a resistance k: The stiffer the spring
(the larger its k), the harder it is to stretch.
x=Fk(2.4.3)
Analogously, voltage tries to charge the capacitor but meets a resistance 1/C: The
larger the value of 1/C, the smaller the resulting charge.
Q=V1/C(2.4.4)
Based on the common structure of the relations for x and Q, spring constant k must
be analogous to inverse capacitance 1/C. Here are all our analogies.
clipboard_e7e3c8ec6922eebc640675151ebd4de19.png
From this table, we can read off our key result. Start with the natural (angular)
frequency ω
of a spring–mass system: ω=km
Then apply the analogies. Mass m is analogous to inductance L. Spring constant k
is analogous to inverse capacitance 1/C. Therefore, ω
for the LC circuit is 1/LC :
ω=1/CL−−−−√=1LC−−−√(2.4.5)
Because of the analogy bridges, one formula, the natural frequency of a spring–mass
system, does double duty. More generally, whatever we learn about one system helps
us understand the other system. Because of the analogies, each piece of knowledge
does double duty.
F=q1q24πϵ0r2(2.4.6)
Because electrostatic and gravitational forces are both inverse-square forces (the
force is proportional to 1/r2), the energy densities should be analogous. Not
least, there should be a gravitational energy density. But how is it related to the
gravitational field?
To answer that question, our first step is to find the gravitational analog of the
electric field. Rather than thinking of the electric field only as something
electric, focus on the common idea of a field. In that sense, the electric field is
the object that, when multiplied by the charge, gives the force:
force = charge×field(2.4.7)
We use words rather than the normal symbols, such as E for field or q for charge,
because the symbols might bind our thinking to particular cases and prevent us from
climbing the abstraction ladder.
gravitationalfield=forcecharge=forcemass=acceeration(2.4.8)
Indeed, at the surface of the Earth, the field strength is g, also called the
acceleration due to gravity.
The definition of gravitational field is the first half of the puzzle (we are using
divide-and-conquer reasoning again). For the second half, we’ll use the field to
compute the energy density. To do so, let’s revisit the route from electric field
to electrostatic energy density:
E→12ϵ0E2(2.4.9)
With g as the gravitational field, the analogous route is
g→12×somethings×g2(2.4.10)
where the “something” represents our ignorance of what to do about ϵ0
.
To find its equivalent, compare the simplest case in both worlds: the field of a
point charge. A point electric charge q produces a field
E=14πϵ0qr2(2.4.11)
A point gravitational charge m (a point mass) produces a gravitational field (an
acceleration)
g=Gmr2(2.4.12)
where G is Newton's constant.
The gravitational field has a similar structure to the electric field. Both are
inverse-square forces, as expected. Both are proportional to the charge. The
difference is the constant of proportionality. For the electric field, it is 14πϵ0
. For the gravitational field, it is simply G. Therefore, G is analogous to 14πϵ0
; equivalently, ϵ0
is analogous to 14πG
.
12×14πG×g2=g28πG(2.4.13)
We will use this analogy in Section 9.3.3 when we transfer our hard-won knowledge
of electromagnetic radiation to understand the even more subtle physics of
gravitational radiation.
Exercise 2.4.1
: Gravitational energy of the Sun
What is the energy in the gravitational field of the Sun? (Just consider the field
outside the Sun.)
Exercise 2.4.2
: Pendulum period including buoyancy
Exercise 2.4.3
: Comparing field energies
Find the ratio of electrical to gravitational field energies in the fields produced
by a proton.
clipboard_e70e53892dafd4935006cd39847bd6c05.png
To find the resistance R across the ladder (in other words, what the ohmmeter
measures between the nodes A and B), you represent the entire ladder as a single
resistor R. Then the whole ladder is 1 ohm in series with the parallel combination
of 1 ohm and R:
clipboard_e54a1253bbf4244a29e014ad0341840c0.png
The next step in finding R usually invokes the parallel-resistance formula: that
the resistance of R1 and R2 in parallel is
R1R2R1+R2(2.4.14)
For our resistive ladder, the parallel combination of 1 ohm with the ladder is 1
ohm × R/(1 ohm + R). Placing this combination in series with 1 ohm gives a
resistance
1Ω+1Ω×R1Ω+R(2.4.15)
This recursive construction reproduces the ladder, only one unit longer. We
therefore get an equation for R:
R=1Ω+1Ω×R1Ω+R(2.4.16)
The (positive) solution is R=(1+5–√)/2
ohms. The numerical part is the golden ratio ϕ
(approximately 1.618). Thus, the ladder, when built with 1-ohm resistors, offers
a resistance of ϕ
ohms.
Although the solution is correct, it skips over a reusable idea: the parallel
combination. To facilitate its reuse, let’s name the idea with a notation:
R1∥R2(2.4.17)
This notation is self-documenting, as long as you recognize the symbol ∥
to mean “parallel,” a recognition promoted by the parallel bars. A good notation
should help thinking, not hinder it by requiring us to remember how the notation
works. With this notation, the equation for the ladder resistance R is
R=1Ω+1Ω∥R(2.4.18)
(The parallel-combination operator has higher priority than—is computed before—the
addition). This expression more plainly reflects the structure of the system, and
our reasoning about it, than does the version
R=1Ω+1Ω×R1ω+R(2.4.19)
.
The ∥
notation organizes the complexity.
Once you name an idea, you find it everywhere. As a child, after my family bought a
Volvo, I saw Volvos on every street. Similarly, we’ll now look at examples of
parallel combination far beyond the original appearance of the idea in circuits.
For example, it gives the spring constant of two connected springs (Problem 2.16):
clipboard_e3c33813ec6f89a7a8dbb89eedd0521f5.png
Exercise 2.4.4
: Springs as capacitors
Using the analogy between springs and capacitors (discussed in Section 2.4.1),
explain why springs in series combine using the parallel combination of their
spring constants.
Another surprising example is the following spring–mass system with two masses:
clipboard_e04b97ed24a0a455c83279194550ccaee.png
ω=k(m+M)mM(2.4.20)
This form looks complicated until we use the ∥
abstraction:
ω=km∥M(2.4.21)
Now the frequency makes more sense. The two masses act like their parallel
combination m∥M
clipboard_ef612e58b4d06803294eb544a3d2ca31d.png
In the spirit of notation that promotes insight, use lowercase (“small”) m for the
mass that is probably smaller, and uppercase (“big”) M for the mass that is
probably larger. Then write m∥M
rather than M∥m
. These two forms produce the same result, but the m∥M
order minimizes surprise: The parallel combination of m and M is smaller than
either mass (Problem 2.17), so it is closer to m, the smaller mass, than to M.
Writing m∥M
, rather than M∥m
, places the most salient information first.
Exercise 2.4.5
: Using the resistance analogy
The answer lies in the analogy between mass and resistance. Resistance appears in
Ohm’s law:
Voltage is an effort. Current, which results from the effort, is a flow. Therefore,
the more general form—one step higher on the abstraction ladder—is
identifies force as the effort, mass as the resistance, and acceleration as the
flow.
Because the spring can wiggle either mass, just as current can flow through either
of two parallel resistors, the spring feels a resistance equal to the parallel
combination of the resistances—namely, m∥M
.
Exercise 2.4.6
: Three springs connected
What is the effective spring constant of three springs connected in a line, with
spring constants 2, 3, and 6 newtons per meter, respectively?
Resistors are easy to handle. When a circuit contains only resistors, we can
immediately and completely describe how it behaves. In particular, we can write the
voltage at any point in the circuit as a linear combination of the voltages at the
source nodes. If only we could do the same when the circuit contains capacitors and
inductors.
flow=1resistance×effort(2.4.23)
For a capacitor, flow will still be current. But we’ll need to find the capacitive
analog of effort. This analogy will turn out slightly different from the
electrical–mechanical analogy between capacitance and spring constant (Section
2.4.1), because now we are making an analogy between capacitors and resistors (and,
eventually, inductors). For a capacitor,
charge=capacitance×voltage(2.4.24)
To turn charge into current, we differentiate both sides to get
current=capacitance×d(voltage)dt(2.4.25)
clipboard_e6290618f78e82aabba2f160e135222e2.png
To make the analogy quantitative, let’s apply to the capacitor the simplest voltage
whose form is not altered by differentiation:
V=V0ejωt(2.4.26)
where V is the input voltage, V0 is the amplitude, ω
is the angular frequency, and j is the imaginary unit −−−√1
. The voltage V is a complex number; but the implicit understanding is that the
actual voltage is the real part of this complex number. By finding how the current
I (the flow) depends on V (the effort), we will extend the idea of resistance to a
capacitor.
With this exponential form, how can we represent the more familiar oscillating
voltages V1 cos ω
t or V1 sin ω
t, where V1 is a real voltage?
ejωt=cosωt+jsinωt(2.4.27)
To make V1 cos ω
t, set V0=V1
in V=V0ejωt
. Then
V=V1(cosωt+jsinωt)(2.4.28)
and the real part of V is just V1 cos ω
t.
V=jV1(cosωt+jsinωt)=V1(jcosωt−sinωt)(2.4.29)
.
dVdt=jω×V0ejωt=jωV(2.4.30)
With this changing voltage, the capacitor equation,
current=capacitance×d(voltage)dt(2.4.31)
becomes
current=capacitance×jω×voltage(2.4.32)
Let's compare this form to its analog for a resistor (Ohm's law):
current=1resistance×voltage(2.4.33)
Matching up the pieces, we find that a capacitor offers a resistance
ZC=1jωC(2.4.34)
This more general resistance, which depends on the frequency, is called impedance
and denoted Z. (In the analogy of Section 2.4.1 between capacitors and springs, we
found that capacitor offered a resistance to being charged of 1/C. Impedance, the
result of an analogy between capacitors and resistors, contains 1/C as well, but
also contains the frequency in the 1/j ω
factor.)
Using impedance, we can describe what happens to any sinusoidal signal in a circuit
containing capacitors. Our thinking is aided by the compact notation—the capacitive
impedance ZC (or even RC). The notation hides the details of the capacitor
differential equation and allows us to transfer our intuition about resistance and
flow to a broader class of circuits.
clipboard_efae8b40ace435aac32286922163d14c9.png
The simplest circuit with resistors and capacitors is the so-called low-pass RC
circuit. Not only is it the simplest interesting circuit, it will also be, thanks
to further analogies, a model for heat flow. Let's apply the impedance analogy to
this circuit.
clipboard_eb9730cf6047423d6623cd76080285de8.png
To help us make and use abstractions, let’s imagine defocusing our eyes. Under
blurry vision, the capacitor looks like a resistor that just happens to have a
funny resistance RC=1jωC
. Now the entire circuit looks just like a pure-resistance circuit. Indeed, it is
the simplest such circuit, a voltage divider. Its behavior is described by one
number: the gain, which is the ratio of output to input voltage Vout/Vin
.
gain=1jωCR+1jωC(2.4.36)
After clearing the fractions by multiplying by jωC
in the numerator and denominator, the gain simplifies to
gain=11+jωRC(2.4.37)
Why is the circuit called a low-pass circuit?
At high frequencies ( ω
→ ∞), the jωRC
term in the denominator makes the gain zero. At low frequencies ( ω
→ 0), the jωRC
term disappears and the gain is 1. High-frequency signals are attenuated by the
circuit; low-frequency signals pass through mostly unchanged. This abstract, high-
level description of the circuit helps us understand the circuit without our
getting buried in equations. Soon we will transfer our understanding of this
circuit to thermal systems.
The gain contains the circuit parameters as the product RC. In the denominator of
the gain, jωRC
is added to 1; therefore, jωRC
, like 1, must have no dimensions. Because j is dimensionless (is a pure number),
ωRC
must be itself dimensionless. Therefore, the product RC has dimensions of time.
This product is the circuit’s time constant—usually denoted τ
.
The time constant has two physical interpretations. To construct them, we imagine
charging the capacitor using a constant input voltage V0; eventually (after an
infinite time), the capacitor charges up to the input voltage (Vout = V0) and holds
a charge Q=CV0
. Then, at t= 0, we make the input voltage zero by connecting the input to ground.
clipboard_ee0d67a4c2bec76f4336727c009ab5b08.png
The capacitor discharges through the resistor, and its voltage decays
exponentially:
clipboard_ec304937418f6b288ce1961e74c472ef4.png
Exercise 2.4.7
: Impedance of an inductor
V=LdIdt(2.4.38)
where L is the inductance. Find an inductor’s frequency-dependent impedance ZL.
After finding this impedance, you can analyze any linear circuit as if it were
composed only of resistors.
2.4.5 Thermal systems
The RC circuit is a model for thermal systems—which are not obviously connected to
circuits. In a thermal system, temperature difference, the analog of voltage
difference, produces a current of energy. Energy current, in less fancy words, is
heat flow. Furthermore, the current is proportional to the temperature difference—
just as electric current is proportional to voltage difference. In both systems,
flow is proportional to effort. Therefore, heat flow can be understood by using
circuit analogies.
clipboard_e5ce9975f3522e9020c62bab2a9060a5d.png
As an example, I often prepare a cup of tea but forget to drink it while it is hot.
Like a discharging capacitor, the tea slowly cools toward room temperature and
becomes undrinkable. Heat flows out through the mug. Its walls provide a thermal
resistance; by analogy to an RC circuit, let's denote the thermal resistance Rt.
The heat is stored in the water and mug, which form a heat reservoir. This
reservoir, of heat rather than of charge, provides the thermal capacitance, which
we denote Ct. (Thus the mug participates in the thermal resistance and
capacitance.) Resistance and capacitance are transferable ideas.
The product RtCt is, by analogy to the RC circuit, the thermal time constant τ
. To estimate τ
with a home experiment (the method we used in Section 1.7), heat up a mug of tea;
as it cools, sketch the temperature gap between the tea and room temperature. In my
extensive experience of tea neglect, an enjoyably hot cup of tea becomes lukewarm
in half an hour. To quantify these temperatures, enjoyably warm may be 130 °F (≈ 55
°C),room temperature is 70 °F (≈ 20 °C), and lukewarm may be 85 °F (≈ 30 °C).
clipboard_e3c06051b0dd288be9fa3b2de2fb3f42a.png
Based on the preceding data, what is the approximate thermal time constant of the
mug of tea?
In one thermal time constant, the temperature gap falls by a factor of e (just as
the voltage gap falls by a factor of e in one electrical time constant). For my mug
of tea, the temperature gap between the tea and the room started at 60 °F:
enjoyably warm130oF−room
temperature70oF=60oF.(2.4.39)
In the half hour while the tea cooled in the microwave, the temperature gap fell to
15°F:
lukewarm85oF−room
temperature70oF=15oF.(2.4.40)
Therefore, the temperature gap decreased by a factor of 4 in half an hour. Falling
by the canonical factor of e (roughly 2.72) would require less time: perhaps 0.3
hours (roughly 20 minutes) instead of 0.5 hours. A more precise calculation would
be to divide 0.5 hours by ln 4, which gives 0.36 hours. However, there is little
point doing this part of the calculation so precisely when the input data are far
less precise. Therefore, let’s estimate the thermal time constant τ
as roughly 0.3 hours.
Using this estimate, we can understand what happens to the tea mug when, as it
often does, it spends a lonely few days in the microwave, subject to the daily
variations in room temperature. This analysis will become our model for the daily
temperature variations in a house.
clipboard_e2878c0502287318c6d7568e62a103c6d.png
First, set up the circuit analogy. The output signal is still the tea's
temperature. The input signal is the (sinusoidally) varying room temperature.
However, the ground signal, which is our reference temperature, cannot also be the
room temperature. Instead, we need a constant reference temperature. The simplest
choice is the average room temperature Tavg. (After we have transferred this
analysis to the temperature variation in houses, we'll see that the conclusion is
the same even with a different reference temperature.)
The gain connects the amplitudes of the output and input signals:
2π×1cycledayfω×0.3
hrτ×1day24hr1≈0.1.(2.4.42)
The system is driven by a low-frequency signal: ω
is not large enough to make ωτ
comparable to 1. As the gain expression reminds us, the mug of tea is a low-pass
filter for temperature variations. It transmits this low-frequency input
temperature signal almost unchanged to the output—to the tea temperature.
Therefore, the inside (tea) temperature almost exactly follows the outside (room)
temperature.
clipboard_e43d43d929a974fd8dc45a01ea45944a9.png
The opposite extreme is a house. Compared to the mug, a house has a much higher
mass and therefore thermal capacitance. The resulting time constant τ=RtCt
is probably much longer for a house than for the mug. As an example, when I
taught in sunny Cape Town, where houses are often unheated even in winter, the
mildly insulated house where I stayed had a thermal time constant of approximately
0.5 days.
2π×1cycledayfω×0.5daysτ≈3.
(2.4.43)
What consequence does ωτ≈3
have for the indoor temperature? In the Cape Town winter, the outside temperature
varied daily between 45 °F and 75 °F; let’s also assume that it varied
approximately sinusoidally. This 30 °F peak-to-peak variation, after passing
through the house low-pass filter, shrinks by a factor of approximately 3. Here is
how to find that factor by estimating the magnitude of the gain.
|gain|≈|11+3j|=112+32−−−−−−√≈13(2.4.45)
In general, when ωτ
≫ 1, the magnitude of the gain is approximately 1/ ωτ
.
clipboard_e3ee897d9acfa41530f61e1c6783648de.png
Our comfort depends not only on the temperature variation (I like a fairly steady
temperature), but also on the average temperature.
It turns out that the average temperature indoors is equal to the average
temperature outdoors! To see why, let’s think carefully about the reference
temperature (our thermal analog of ground). Before, in the analysis of the
forgotten tea mug, our reference temperature was the average indoor temperature.
Because we are now trying to determine this value, let’s instead use a known
convenient reference temperature—for example, the cool 10 °C, which makes for round
numbers in Celsius or Fahrenheit (50 °F).
The input signal (the outside temperature) varied in winter between 45° F and 75°F.
Therefore, it has two pieces: (1) our usual varying signal with the 30°F peak-to-
peak variation, and (2) a steady signal of 10°F.
clipboard_e3614101a1bfb7f0947e691d1fca1bff8.png
The steady signal is the difference between the average outside temperature of 60°F
and the reference signal of 50°F.
Let’s handle each piece in turn—we are using divide-and-conquer reasoning again. We
just analyzed the varying piece: It passes through the house low-pass filter and,
with ωτ≈3
, it shrinks significantly in amplitude. In contrast, the nonvarying part, which
is the average outside temperature, has zero frequency by definition. Therefore,
its dimensionless parameter ωτ
is exactly 0. This signal passes through the house low-pass filter with a gain of
1. As a result, the average output signal (the inside temperature) is also 60°F:
the same steady 10°F signal measured relative to the reference temperature of 50°F.
Exercise 2.4.8
: When is the house coldest?
This page titled 2.4: Analogies is shared under a CC BY-NC-SA license and was
authored, remixed, and/or curated by Sanjoy Mahajan (MIT OpenCourseWare) .