You are on page 1of 6

EE122: Frequency Response of Passive Circuits

Review Notes on the Frequency Response


of Passive Circuits
Prepared by Rizwan Ahmed
Stanford University
January 7, 2003
1 Impedance of Resistors, Capacitors, and Inductors
We write the impedance of a component as Z and the frequency as . Then the impedance of a resistor is
Z
R
= R, The impedance of a capacitor is Z
C
= 1/(jC), and the impedance of an inductor is Z
L
= jL.
Thus a capacitor is a short circuit (Z
C
= 0) at high frequencies and is an open circuit (Z
C
= ) at DC.
An inductor is an open circuit (Z
L
= ) at high frequencies and is a short circuit (Z
L
= 0) at DC. To
understand how we arrive at these equations, recall that inductors store energy with magnetic elds where
as capacitors store energy with electric elds. Using these impedances we can do frequency analysis of our
circuit using normal methods (KVL, KCL, Ohms Law, etc . . .). The input/output relationship is referred
to as the Transfer function, and is often denoted as H(j). At any given frequency, this is just a complex
number. The magnitude and phase of this number give the magnitude and phase of the output in relation
to the input.
2 Identifying the Frequency Response Visually
Good circuit designers can just look at a complex circuit and start identifying components (ampliers, lters,
etc . . . ). Identify the frequency response (Lowpass, Highpass, or Bandpass) of the circuits in the gure below.
R
1
R
2 L
C
R
R
1
C L
R
2
V
out V
out
V
out
hint: Think about how the circuits would behave at DC and very high frequencies.
3 Dening the Cuto Frequency
The cuto frequency is dened where
P
out
(
c
) =
P
max
2
(1)
Plugging in the denition of power across a resistor, we obtain the equation
V
2
(
c
)
R
=
V
2
max
2R
(2)
which leads to
V (
c
)
V
max
=
1

2
(3)
1
EE122: Frequency Response of Passive Circuits
So the ratio of the transfer function is equal to 1/

2 at the cuto frequency.


The decibel is a convenient unit when measuring values over a very wide range and is often used in en-
gineering. If the measurement is voltage, the value in decibels is dened as
V
dB
= 20 log([V [) (4)
plugging in equation (3) we get
V
dB
(
c
) = 20 log

[V
max
[

= 20 log([V
max
[) 20 log(

2)
V
dB
(
c
) = V
dBmax
3dB (5)
So the voltage is 3dB down at the cuto frequency! In the case of power measurements, the value in
decibels is dened as
P
dB
= 10 log(P) (6)
plugging in equation (1) we get
P
dB
(
c
) = 10 log(
P
max
2
) = 10 log(P
max
) 10 log(2)
P
dB
(
c
) = P
dBmax
3dB (7)
So the power is 3dB down at the cuto frequency as well.
4 Bode Plots
Before digital computers were a commonplace, a quick way of sketching the frequency response of a system
was the Bode plot. However, these plots still provide valuable intuition and insight and thus remain an im-
portant concept. The Bode plot is a piecewise linear approximation of the transfer function (both magnitude
and phase) that can be constructed using a set of simple rules. To construct a Bode plot, we must write
the transfer function in the following form (known as the standard form). The z and p terms are called the
zeros and poles respectively.
H(j) = k
(j)

(1
j
z
1
)(1
j
z
2
) (1
j
z
m
)
(1
j
p
1
)(1
j
p
2
) (1
j
p
n
)
(8)
Now if we measure the transfer function in decibels and use the properties of logarithms (multiplication
addition, division subtraction, exponents coecients), we obtain
H(j)
dB
= 20 log([k[) 20 log() + 20 log

1
j
z
1

+ 20 log

1
j
z
m

(9)
20 log

1
j
p
1

20 log

1
j
p
n

The rst term is a constant and the second is linear on a log scale. The other terms are approximated in the
following way. If z than 1

z
1 and 20 log(1) = 0. On the other hand, if z than [1

z
[

z
and
20 log(/z) is a linear term on a log scale. We approximate these terms on the Bode plot by letting them
equal zero until = z and linear afterwards. Of course, this reasoning also holds for the poles. So here are
the rules to making a Bode plot.
1. Construct a log-log plot with on the x-axis and [H(j)[dB on the y-axis. Let the x-axis start at 1.
2. Calculate 20 log([k[), this is the starting point on the y-axis at = 1.
3. If you have a linear term, then start rising by 20dB/decade. If not, then stay constant. Do this until
you encounter a pole or zero.
2
EE122: Frequency Response of Passive Circuits
4. If you encounter a pole, start dropping by 20dB/decade. If you encounter a zero, start rising by
20dB/decade.
5. The rises and drops are cumulative. So for example, if you have a linear term with = 2 and two poles
at = 1000Hz, you would rise at 40dB/decade and atten out after = 1000Hz (This is a second
order high-pass lter).
Now if we look at the phase, we use two very convenient properties of the phase of complex numbers; namely
that multiplication addition and division subtraction. So the phase of the transfer function is given by

H(j) =

k +

(j)

1
j
z
1

1
j
z
m

1
j
p
1

1
j
p
n

(10)
The phase of the rst term is 0 if k 0 and 180

otherwise. The second term has a phase of 90

. The
phase of the pole and zero terms is approximated in the following way. If z than 1
j
z
1 and

1 = 0.
If = z than

1
j
z

= 45

. Finally if z than 1
j
z

j
z
and

j
z

= 90

(all of this
assumes a positive z). Here are the rules for constructing a phase Bode plot.
1. Construct a semilog plot with on the x-axis and

H(j) on the y-axis. Let the x-axis start at 1.
2. Calculate

k +

(j)

. This is the starting point of your phase plot.


3. If you encounter a pole, construct a linear drop of 90

from 0.1z to 10z. If you encounter a negative


zero ('z < 0), construct a linear rise of 90

from 0.1z to 10z. If you encounter a positive zero


('z > 0), construct a linear drop of 90

from 0.1z to 10z.


If the steps sound a bit confusing, it is only because it is dicult to put them into words. The actual steps
are very straightforward. Constructing Bode plots becomes much clearer after doing a few examples.
5 Some Common Filters
You will all use lters in your projects. The forms for some common transfer functions is given below.
A lowpass lter with a given
c
and a rollo of n 20dB/decade. H(0) is the DC gain and n is the
order of the lter.
H(j) =
1
(1 +
j

c
)
n
H(0) (11)
A highpass lter with a given
c
and a rollo of n 20dB/decade. H() is the gain at innite frequency
and n is the order of the lter.
H(j) =

1 +
j

n
H() (12)
A bandpass lter with 20dB/decade rollo in both directions. is the bandwidth (the amount of
spectrum between the 3dB points) and H(
o
) is the transfer function evaluated at the center frequency
of the passband. While we did not put this in standard form it is easy to see that it has a linear term
and two poles. Thus it will rise and then eventually fall at 20dB/decade. With some algebraic eort,
you can show that the and
o
are indeed the bandwidth and center frequency respectively.
H(j) =
j
(j)
2
+(j) +
2
o
H(
o
) (13)
It is easy to realize these transfer functions with circuits. The importance of writing the transfer function
in these forms is that they give an easy way of identifying the important parameters of the lter (cuto
frequency in the case of LP and HP lters, and center frequency and bandwidth in the case of the BP lter).
3
EE122: Frequency Response of Passive Circuits
6 Examples
Example 1
R
L
Type of Filter
V
out
V
in
First, we see that this circuit is a voltage divider. So we write
H(j) =
R
R +jL
(14)
or in standard form
H(j) =
1
1 +j/(R/L)
(15)
So we can immediately see that the cuto frequency is
c
= R/L. The magnitude and phase of the transfer
function is given by
[H(j)[ =
1

1 + (
L
R
)
2
(16)

H(j) =

1

1 +
j
R/L

(17)
We can simplify the equation for the phase to get

H(j) = tan
1

L
R

(18)
The gure below plots the true magnitude and phase response (blue) and the Bode approximations (red).
Example 2
R
2
Type of Filter
V
in
R
1
V
out
L
We see that the impedance of R
2
| L is
jLR
2
R
2
+jL
. So we write
H(j) =
jLR
2
R
2
+jL
R
1
+
jLR
2
R
2
+jL
(19)
4
EE122: Frequency Response of Passive Circuits
Doing some algebra, we get
H(j) =
R
2
R
1
+R
2

(j)/(R
1
R
2
/L(R
1
+R
2
))
1 + (j)/(R
1
R
2
/L(R
1
+R
2
))
(20)
Comparing with the standard form of the highpass lter we see that, H() =
R
2
R
1
+R
2
and
c
=
R
1
R
2
L(R
1
+R
2
)
.
Now suppose R
2
R
1
and R
1
= 100, and you need a
c
= 1kHz. What value would you choose for
the inductor?
Since R
2
R
1
, we can ignore R
1
in the denominator of the expression for the cuto frequency. Then
R
2
s cancel and we are left with
c
= R
1
/L. Thus L = R
1
/
c
= R
1
/(2f
c
). Then,
L =
100
2 1000Hz
=
1
20
H 15.9 mH (21)
The gure below plots the true magnitude and phase response (blue) and the Bode approximations (red).
Example 3
R
Type of Filter
V
in
V
out
C L
We write the transfer function
H() =
R
R +jL + 1/(jC)
(22)
or equivalently as
H() =
j(R/L)
j
2

2
+j(R/L) + 1/(LC)
(23)
so
o
= 1/

LC and = R/L. We also calculate the magnitude and phase of the transfer function.
[H(j)[ =
(R/L)

(
R
L
)
2
+ (
1
LC

2
)
2
(24)

H(j) =

(j(R/L))

1
LC

2
+j(R/L)

(25)
This simplies down to

H(j) = 90

tan
1

(R/L)
1/(LC)
2

(26)
The next gure plots the true magnitude and phase response.
5
EE122: Frequency Response of Passive Circuits
6

You might also like