6.

200 Notes: Time Domain R-L-C

Prof. Karl K. Berggren, Dept. of EECS
April 11, 2023

Most resources (including the course text book) treat series and paral-
lel combination of RLC circuits as if they are different circuits. There
is nothing wrong with that approach, but I feel it is more natural
and helpful to treat these as special cases of a single more general
circuit. We will take the single-circuit approach here. I really recom-
mend you also refer to standard texts. This material is intended to
supplement, not supplant, the course textbook.

Some Introductory Material

Let’s start off by re-familiarizing you with some notation you (prob-
ably?) have seen before at some point, but may not be too familiar. I
also introduce some unusual notation.

Notation
A∗ Complex conjugate of A.

ℜ[ A] Real part of A.

ℑ[ A] Imaginary part of A.

c.c. Complex conjugate of previous term in expression. Thus A +

c.c. = A + A∗

∂t Derivative with respect to t (technically this is used as a partial

derivative, while D is used for the standard derivative, but here we
will use it as a convenient notation. You should consider it to be
equivalent to d/dt.

∂tt Second derivative with respect to t (see above).

√
j Complex constant −1 (i.e. equivalent to i in math and physics).
 6.200 notes: time domain r-l-c 2

Some Useful Identities

A + A∗ = 2ℜ[ A]
A − A∗ = 2jℑ[ A]
e jx + e− jx = 2 cos x
 
−1 b
∠( a + bj) = tan
a
π
cos( x − ) = sin( x )
2
a + bj = | a + bj|e∠(a+bj)
p
| a + bj| = a2 + b2

The Problem

In engineering, circuits often exhibit undesired “ringing” due to the

presence of parasitic capacitance and/or inductance. In addition,
even when oscillation is desired (as is often the case) the presence
of a resistor leads to complexity relative to the simpeler L-C case we
discussed previously. For these situations, and understanding of the
response of such systems in the time domain is essential.
The typical LRC circuit consists of a resistor, capacitor, and induc-
tor either in parallel or in a series loop configuration. These two cases
are shown in figure 1 below. Typically the problem will provide an
initial state for the capacitor (an initial voltage vC (0)) or the induc-
tor (an initial current iL (0)) or both. You will then have to find some
current and/or voltage at some later time, or as a function of time.

Series Parallel Figure 1: Series (left) and parallel

(right) LRC circuits.

L
C C L R
R

Of course sources can be added to these problems, and we will

discuss such situations below. For now, we will assume all source
strengths are set to zero and no longer change after t > 0, but where
the capacitor and/or inductor may have a non-zero state at t = 0.
We’ll call the initial inductor state Λ◦ and the initial capacitor state
Q◦ .1 1
Q◦ on a capacitor will result in a
The problem is to now determine the time evolution for t > 0 of voltage V◦ = Q◦ /C. Similarly Λ◦ will
result in a current I◦ = Λ◦ /L
any of the circuit variables. We will focus our solution on vC , i L , but
the exact approach would work for any other variable. Indeed, the
 6.200 notes: time domain r-l-c 3

advantage of taking this more general approach is that we cast the

broadest possible net in solving the problem.
This is known as determining the natural response of an L-R-C
circuit.

L-R-C in Series

We will start by treating the case of an L-R-C circuit in series:

vL
+ −
iL L
+
C vC
iC − R

Step 1: Deriving the Differential Equation

From the constitutive relations for a capacitor and an inductor, we
can write
dv di
iC = C C , and v L = L L . (1)
dt dt
We can then use KVL around the L-R-C loop to derive the equa-
tion:
vC = v L + i L R. (2)
We can also use KCL at either node to state:
i L = −iC (3)
Substituting (1) into (2) and (3) we get:
di L
vC = L + iL R (4a)
dt
dv
i L = −C C . (4b)
dt
Note that these equations reduce to the same coupled first-order
differential equations as arise in an L-C circuit when R → 0.
In this format, the solution is quite computable by numerical
methods, and in practice this is a convenient way to approach the
problem. However, such an approach does not provide the necessary
intuition, so we will take the step of reducing these equations to an
equation of a single variable.
To derive an equation in terms of only i L , we will now substitute
(4a) into (??):2 2
Here we introduce the notation ∂t for
d/dt, and ∂tt for d2 /dt2 simply because
−i L = LC∂tt i L + RC∂t i L (5) it is easier to use when doing algebra.
 
R 1
⇒ 0 = ∂tt i L + ∂t i L + iL . (6)
L LC
 6.200 notes: time domain r-l-c 4

Remarkably, if we do the opposite and substitute (??) into (4a) the

form of the equation doesn’t change, just the variable, so

−vC = LC∂tt vC + RC∂t vC (7)

 
R 1
⇒ 0 = ∂tt vC + ∂t vC + v . (8)
L LC C
Even more remarkably, it turns out that any circuit variable we
choose (even i R or v R ) will have exactly the same form. To write this
most generally, we will use x to represent vC or i L or any variable we
may be interested in, and so we can write:
 
R 1
∂tt x + ∂t x + x = 0. (9)
L LC
Isn’t linearity a miraculous thing?

Step 2: Identifying Some Special Constants

Equation (9) is an important form for us, but the coefficients are a
bit cumbersome. Anticipating some of the features of the solution
(sorry... it will be clear why we do this soon), we rewrite it in an even
more general form as:

∂tt x + 2α∂t x + ω◦2 vC = 0. (10)

where
R
α≡ (11)
2L
2 1
ω◦ = . (12)
LC
We have deliberately expressed this in terms of α and ω◦ instead
of R, L, and C, because different topologies will always have the same
overall form, thus by simply changing the values of α and ω◦ , one
doesn’t have to resolve the entire differential equation every time one
has a new circuit topology. For example, for the case of R, L, C in
parallel, ω◦ is unchanged, but α = 1/(2RC ).
Both α and ω◦ have units of inverse time. ω◦ of course represents
a frequency but, as we’ll see below, α represents a rate of decay (an
inverse of a time constant, similar to 1/τ).

Step 3: Finding a Solution

Let’s try a solution of the form Aest .3 Substituting this into (10), we 3
I know this seems unsatisfying... but
find this is literally the only differential
equation—and the only solution—you’ll
ever see in circuit theory. So can we just
s2 Aest + 2αsAest + ω◦2 Aest = 0 (13) accept it and move on?
2
⇒s + 2αs + ω◦2 = 0. (14)
 6.200 notes: time domain r-l-c 5

which is known as the “characteristic equation” of this system.

The solutions to this equation by the quadratic formula will be
√
b b2 − 4ac
− ± (15)
2a 2a
where a = 1, b = 2α, and c = ω◦2 , which is then
p
2α 4α2 − 4ω◦2
s =− ± (16)
2 q 2
s =−α± α2 − ω◦2 . (17)

This trial solution naturally gives us three cases, based on the sign
of α2 − ω◦2 . If ω◦ < α, the situation is termed “underdamped” for
reasons we shall see in a moment. If ω◦ = α, the situation is termed
“critically damped,” which is unusual and thus is of little interest
to us. And finally, if ω◦ > α, the situation is termed underdamped,
again, for reasons we shall see in a moment. We will primarily con-
cern ourselves with the underdamped case, because it has many
applications, but overdamped circuits are also useful in some situa-
tions.

The Underdamped Case

In the underdamped case, where α < ω◦ , the expression α2 − ω◦2 from
(17) will be negative and so the routes of the characteristic equation
will be complex. In that case, we can rewrite (17) as:

s± = −α ± jωd (18)

where we use j as the imaginary constant instead of i to avoid confu-

sion with current,4 and we define a new frequency parameter 4
Using j instead of i is standard prac-
q tice in electrical engineering, to avoid
confusion with the current symbol. But
ωd ≡ ω◦2 − α2 . (19)
why is current denoted i instead of c (or
j, even)? Because it is derived from the
When the damping term α is small relative to ω◦ , ωd ≈ ωo . french term “intensité,” meaning “in-
The fact that there are two roots for the characteristic equation in tensity” which originated before people
this case suggests that two valid solutions will exist. Because it is a understood that current represented a
flow of particles.
linear equation, superposition of these two solutions should also be
a valid solution (as was the case with the L-C circuit), thus generally,
the solution should be of the form:5 5
The case we are working here is for a
homogeneous equation, thus this will
x = A+ e s+ t + A− e s− t (20) apply for the homogeneous solution.
−αt jωd t −αt − jωd t Circuits with voltage sources will be
= A+ e e + A− e e (21) slightly different, as we will discuss
  below.
= e−αt A+ e jωd t + A− e− jωd t . (22)

Next, we have to determine the values of A+ and A− . Before we

go there, let’s talk about the (literal) complexity of our proposed solu-
tion. You have every right at this point to be very concerned that our
 6.200 notes: time domain r-l-c 6

proposed solution appears to be complex! After all, e jωd t is certainly

and x is a current or voltage and thus certainly is not! So what gives?
Well, the key is that A+ and A− must be complex conjugates of each
other... so we can write A+ ≡ A and then A− = A∗ . As a result, the
term Ae jωd t is the complex conjugate of A∗ e− jωd t and their sum will
be real, so x will be real. We can finally then write:
 
x (t) = e−αt Ae jωd t + A∗ e− jωd t (23)
 
= e−αt Ae jωd t + c.c. (24)

where we introduce the notation c.c. to represent the complex conju-

gate of the first term in the expression.6 6
Complex conjugation is just the
process of replacing all the j symbols
with − j symbols, i.e multiplying the
Using Initial Conditions imaginary part of the number by -1.

Now, back to figuring out A+ and A− (or just A now that we know
they’re just complex conjugates). There are two unknowns (the real
and imaginary parts of A), so we are going to need two pieces of
information about the circuit to solve for them. In principle we could
use the values of x at any two points in time, but the the t = ∞ case
does us no good because e−αt factor is zero there, so we lose any
useful information. Luckily, we have been given two facts that we can
use as initial conditions.
Although we have two initial conditions, there is only one variable
x so there is only one x (0). Luckily, we can use ∂t x (0) as one of our
conditions. But “we don’t know ∂t x (0),”do I hear you cry? Not so!
We may not have been told it explicitly, but the current can be used
to calculate the derivative of the voltage and vice versa. This can be
figured out from (4a) and (??). Rearranging these expressions, we can
write:
vC R
∂t i L = − iL (25)
L L
i
∂t vC = − L (26)
C
Because we know i L (0) = Λ◦ /L and vC (0) = Q◦ /C (remember
this was part of the initial setup of the problem?), we can substitute
for these values above and find:
di L (0) Q◦ RΛ◦
= − 2 (27)
dt LC L
dvC (0) Λ◦ 2
=− C . (28)
dt LC
This is a bit more complicated than typical... often, either Λ◦ or Q◦
will be zero, and these conditions will simplify greatly as a result.
 6.200 notes: time domain r-l-c 7

So we now know x (0) and ∂t x (0). We can find the two equations
for these expressions be evaluating (24) at t = 0, and by taking its
derivative, and then evaluating that at t = 0.

x (0) = A + c.c (29)

    
∂t x (t) = −αe−αt Ae jωd t + c.c. + e−αt jωd Ae jωd t + c.c.
t =0 t =0
(30)
= (−α + jωd ) A + c.c. (31)

Notice that (29), and (31) are two equations and have two un-
knowns (remember, x (0) and ∂t x (0) are now both known, set by the
initial conditions on current and voltage given in the problem), and
so can be solved by using standard linear algebra methods. Mak-
ing the observation that the coefficient of A in (31) is just s+ we can
write: ! ! !
x (0) 1 1 A
= (32)
∂ t x (0) s+ s∗+ A∗
which can be inverted to give
! ! !
A 1 s∗+ −1 x (0)
= ∗ (33)
A∗ s+ − s+ −s+ 1 ∂ t x (0)
! !
1 −α − jωd −1 x (0)
=− . (34)
2jωd α − jωd 1 ∂ t x (0)

The Final Answer

Now that we know A, we can back-substitute into (24) to write out
the full expression for x (t):
  
α + jωd 1
x (t) = e−αt x (0) − ∂t x (0) e jωd t + c.c. (35)
2jωd 2jωd

While this is indeed a daunting equation... for simpler cases, it is

immediately reduceable. For example, in many cases either x (0) or
∂t x (0) will equal zero.
Additionally, this form can always be reduced to a more intuitive
form as follows:
x (t) = Xe−αt cos (ωd t + ϕ◦ ). (36)
where7 7
We have used the relation Ae jB +
−α + jωd 1 c.c. = 2| A| cos( B + ∠ A), where the ∠
X= x (0) − ∂ t x (0), (37) symbols represents the polar angle of
jωd jωd
its operand.
which is real, and
−α + jωd
 
1
ϕ◦ = ∠ x (0) − ∂ t x (0) , (38)
jωd jωd
 6.200 notes: time domain r-l-c 8

which is of course also real. This form means that these solutions
will always have the form of a decaying oscillation. The cosine term
enforces oscillation with radial frequency ωd in units of radians per
time (ωd /2π in cycles per unit time). The exponential term multiplies
the cosine and causes it to decay with a time constant of τ = 1/α.8 8
If you’ve forgotten from physics,
The figure below shows a characteristic case, where a variable x (t) radial frequency is the rate of change
of an angle, and is measured in units of
(a current or voltage) is decaying in time. radians per second. To convert to cycles
x (t) per second (conventional frequency, in
units of hertz), one must divide by the
number of radians in a cycle (namely
2π). Thus ω = 2π f .

e−αt

T = 2π/ωd

Examples

Let’s apply our solution to two examples.

+
+ vL L
− iL
vC C
+
− iC vR R
−

Suppose the inductor starts out un-fluxed (i.e. i L (0) = 0), and the
capacitor starts out with some initial voltage across is V◦ . Find the
current through the inductor and voltage across the capacitor as a
function of time in this circuit.
We’ll deal with finding the current in the inductor first.
First, let’s set up the differential equation for the problem. We will
use our definitions of α, ω◦ , and ωd , substituting in GC = 0 to find:
r
R 1 1 R2
α= ω◦ = √ ωd = − 2. (39)
2L LC LC 4L
 6.200 notes: time domain r-l-c 9

We could now write out our differential equation... but we don’t

need to! We just recognize we are trying to solve for i L , so we should
set x ≡ i L in our solution above and jump straight to the solution as
shown in (10).
To determine the initial conditions, we observe directly that we
were told that i L (0) = 0, which is our x (0). We only need to deter-
mine ∂t x (0) which we can from our equation (27) above, recognizing
that if i L (0) = 0, then Λ◦ = 0 and if vC (0) = V◦ , Q◦ = CV◦ so
∂t i L (0) = V◦ /L, which is our ∂t x (0) condition.
With these two initial conditions, we can use our solution above to
write:
 
−αt V◦ jωd t
i L (t) = e − e + c.c. (40)
2jωd L
V◦ −αt
= e cos (ωd t + π/2) (41)
ωd L
V◦ −αt
=− e sin (ωd t) (42)
ωd L
where we have used our expressions (37) and (38). Additionally, in
the second step we used the fact that ∠(−1/j) = π/2, and in the
third step we used the fact that cos(θ + π/2) = − sin(θ ).
We were also interested in vC (t) for this problem. We know that α,
ω◦ and ωd are unchanged, so we can skip straight to the initial condi-
tions. In this case we observe that vC (0) = V◦ and because i L (0) = 0,
the ∂t x (t) term will be zero. With these two initial conditions, we can
use our solution above in a single step as:
−α + jωd
 
vC (t) = e−αt V◦ e jωd t + c.c. (43)
2jωd
−α + jωd jωd t
 
= V◦ e−αt e + c.c. (44)
2jωd

Let’s pause for a second to point out that the coefficient of the e jωd t
term can be written in polar notation as
−α + jωd 1 ∠j
q
= α2 + ωd2 e j∠(−α+ jωd ) · e (45)
2jωd 2ωd
q
α2 + ωd2
= e j(∠(−α+ jωd )−∠ j) (46)
2ωd
s
α2 + ωd2 jϕ◦
= e (47)
4ωd2
s
1 jϕ◦ α2
= e 1+ 2 (48)
2 ωd

where we have defined ϕ◦ ≡ ∠(−α + jωd ) − ∠ j. We can use the fact

that ∠ j = π/2 and in general ∠( a + jb) = tan−1 (b/a), where one
 6.200 notes: time domain r-l-c 10

has to take care to get the correct quadrant of the arctan function, to
write:
 
vC (t) = V◦ e−αt e j(ωd t+ϕ◦ ) + c.c. (49)
s
α2
= V◦ e−αt 1 + 2 cos (ωd t + ϕ◦ ), (50)
ωd

where ϕ◦ = ∠(−α + jωd ) − ∠( j) = tan−1 (−ωd /α) − π/2 and we

know that we are in the correct quadrant because α and ωd are both
real and positive.9 9
Remember whenever switching from
Finally, it is most instructive to study the two solutions together an e jϕ + c.c. notation to cos notation, a
factor of two creeps in... be careful.
and graphically.

vC (t),i L (t) Figure 2: Characteristic decaying

oscillation observed in RLC circuits.
Time-domain comparison of i L (red)
and vC (blue), normalized to have the
same amplitude at t = 0. Current is
needed to charge and discharge the
capacitor, thus it leads the capacitor
voltage. Period of oscillation (time
between zero crossings) is T = 2π/ωd .
t

i L ωd L Figure 3: Parametric plot of current in

inductor vs. voltage across capacitor,
showing oscillation. The current access
is scaled by a factor of
√ “characteristic
impedance” ωd L = L/C to give it
vC units of volts, and to permit plotting
on the same scale as voltage, otherwise
plots would appear elliptical. System
spirals in with time.
 6.200 notes: time domain r-l-c 11

Energy?

Dipping down into the physics of this system briefly, we notice that
there must be some slow decay of energy associated with the decay-
ing voltage and current. This energy must appear through heat in the
inductor.
How fast does this decay occur, exactly? We start out by observing
E ∝ v2 and asking how many radians of phase (i.e. ω◦ t) must evolve
before system energy drops by a factor of 1e ?

1
v2 ∝ e−2αt = ⇒ 2αt = 1
e
1
⇒t=
2α
ω0 1 L2 Z
ω0 t = = √ = 0 =Q
2α 2 LC R R
You thus see that Q has meaning in time domain: it represents the
number of radians of phase that evolve for the energy to drop by a
factor of 1/e.
A more convenient way to think of it is to realize that in Q cycles
(i.e. in time t = QT = 2πQ/ω◦ = 2πω◦ /(2αω◦ ) = π/α the voltage is
reduced by a factor of e−αt = e−απ/α = e−π ≈ 0.04 = 4%.

What’s Next?

Derived parameters like ω◦ , α, and ωd are tremendously helpful

when trying to determine quickly how a circuit behaves. But there
are even more parameters that we haven’t discussed here that can
further help with interpretation. Discussion of how energy moves
back and forth between circuits, how it is dissipated, and how over-
damped cases should be treated are all interesting and worthwhile
areas to look into. Furthermore, circuits that at first glance don’t
appear to fall into the simple structure shown can arise, but with
some effort can be mapped onto this structure. Ultimately, using the
frequency domain rather than the time domain to analyze circuit
behavior provides even more powerful methods.

Glossary and Definitions

Characteristic Equation: Polynomial equation used to determine the

eigenvalues of a matrix. In this case, refers to the polynomial form
equation that results from substitution of a trial solution.

Complex Conjugate: Takes the imaginary part of a complex number

and multiplies it by -1. The real part of the number is unchanged.
 6.200 notes: time domain r-l-c 12

Thus ( a + bj)∗ = ( a − bj) where a and b are real numbers, and j is

the imaginary constant j2 = −1.

Critically Damped: An approximate definition is the situation in

which damping rate in a system matches the rate of oscillation
or equivalently the decay time constant matches the oscillation
period. This regime is of little interest to us here, but of value in
advanced topics, particularly in mechanics.

Duality: Inductors and capacitors are often referred to as “dual” cir-

cuit elements. Duality means that the role of current and voltage
are reversed. Thus we could also say that resistance and conduc-
tance are dual variables. It turns out that, geometrically, a node
and a loop in a circuit are also dual topological elements. Duality
is a deep concept in circuits, and covered in detail in advanced
classes.

Initial Conditions: Values of circuit variables at the initiation of a

region in which a differential equation is applied.

Natural Frequency: The natural frequency ω◦ is the frequency at

which a perfect L-C oscillator would resonate in the absence of a
driving source.

Overdamped: Situation in which damping rate in a system is large

relative to the rate of oscillation or equivalently the decay time
constant is short relative to the oscillation period. Rapid non-
oscillatory decay is characteristic of this regime.

Parasitics: Undesired circuit elements that arise due to imperfections

in the construction and implementation of circuit components in
the real world. For example, series resistance is unavoidable in
inductors made of normal metals (not superconductors), and a
parallel resistor is unavoidable in capacitors.

Radial Frequency: Frequency expressed in units per unit time instead

of the more conventional cycles per unit time. There are 2π radi-
ans in a cycle, thus radial frequency ω is related to conventional
frequency f through the relation ω = 2π f .

Underdamped: Situation in which damping rate in a system is small

relative to the rate of oscillation or equivalently the decay time
constant is long relative to the oscillation period. Decaying oscilla-
tions are characteristic of this regime.
 6.200 notes: time domain r-l-c 13

Appendix: Just the Maths, Ma’am

In this appendix, we work through a sample problem focussing

on just the algebra. It provides a more concise example relative to
the verbose description in the text, and may help illuminate the
problem structure.

Problem Statement

v ( t = 0) = v0
i ( t = 0) = 0
The 0 here simplifies algebra a bit, but not otherwise necessary
Find v(t) for t ≥ 0.
We can assume we are in the underdamped case: Z0 > R.

Starting Equations

dv
i=C
dt
di
vL = L
dt
di
KVL ⇒ L +v + |{z}iR = 0
dt
|{z} vR
vL

d2 v dv
⇒ LC + v + CR =0 ÷ LC
vt2 dt
1 R
ω0 ≡ √ α=
LC 2L
d2 v R dv 1
⇒ + + =0
dt2 L dt LC
d2 v dv
⇒ + 2α + ω02
dt2 dt

Characteristic equation:

d st
Noting dt Ae = sAest

⇒ s2 + 2αs + ω02 = 0
 6.200 notes: time domain r-l-c 14

Roots of the Characteristic Equation

a=1 b = 2α c = ω02
q
± (2α)2 − 4ω02 q
s± = −α = −α ± α2 − ω02
2
q
L
In underdamped case Z0 > R ⇒ C = ω0 L > R ⇒ ω0 > α

∴ α2 − ω02 < 0
q
⇒ spm = −α ± j ω02 − α2 − α ± jωd
q
where ωd = ω02 − α2

Form of Solution

v = A+ e s+ t + A − es− t
= A+ e−αt e jωd t + A− e−αt e− jωd t
= A+ e−αt e jωd t + c.c.
|{z}
complex conjugate of previous term
| {z }
x +c.c.( x )= Re( x )

Using Boundary Conditions

v(0) = V0 ⇒ A + A∗ = 2Re( A) = V0
⇒ Re( a) = V0 /2
dv
i (0) ∝ | t =0 = 0
dt

⇒
= −α · 2 · Re( A) + jωd (2jlm( A))
= −αV0 − 2ωdlm( A) = 0

αV0
⇒ lm( A) = −
2ωd
V0 jαV0
⇒A= −
2 2ωd
⇒ v(t) = 2 · Re( Ae jωd t e−αt )
αV0 −αt
⇒ v(t) = V0 e−αt cosωd t + e sinωd t
ωd