Professional Documents
Culture Documents
M. B. Patil
mbpatil@ee.iitb.ac.in
www.ee.iitb.ac.in/~sequel
1
v1
v
v2
0
−1
0 5 10 15 20 0 5 10 15 20
t (msec) t (msec)
Introduction to filters
1
v1
v
v2 v LPF vo = v1
−1
0 5 10 15 20 0 5 10 15 20
t (msec) t (msec)
A low-pass filter with a cut-off frequency ω1 < ωc < ω2 will pass the low-frequency component v1 (t) and
remove the high-frequency component v2 (t).
Introduction to filters
1
v1
v
v2 v LPF vo = v1
v HPF vo = v2
−1
0 5 10 15 20 0 5 10 15 20
t (msec) t (msec)
A low-pass filter with a cut-off frequency ω1 < ωc < ω2 will pass the low-frequency component v1 (t) and
remove the high-frequency component v2 (t).
A high-pass filter with a cut-off frequency ω1 < ωc < ω2 will pass the high-frequency component v2 (t) and
remove the low-frequency component v1 (t).
1
v1
v
v2 v LPF vo = v1
v HPF vo = v2
−1
0 5 10 15 20 0 5 10 15 20
t (msec) t (msec)
A low-pass filter with a cut-off frequency ω1 < ωc < ω2 will pass the low-frequency component v1 (t) and
remove the high-frequency component v2 (t).
A high-pass filter with a cut-off frequency ω1 < ωc < ω2 will pass the high-frequency component v2 (t) and
remove the low-frequency component v1 (t).
There are some other types of filters, as we will see.
H(jω)
vi (t) H(jω) vo (t)
0 ω
0
ωc
H(jω)
vi (t) H(jω) vo (t)
0 ω
0
ωc
Vo (jω)
Vi (jω)
LPF
ω ω
0 0
ωc ωc
H(jω)
vi (t) H(jω) vo (t)
0 ω
0
ωc
Vo (jω)
Vi (jω)
LPF
ω ω
0 0
ωc ωc
Low−pass
H(jω)
0 ω
0
ωc
Ideal filters
Low−pass High−pass
1 1
H(jω)
H(jω)
0 ω 0 ω
0 0
ωc ωc
Ideal filters
Low−pass High−pass
1 1
H(jω)
H(jω)
0 ω 0 ω
0 0
ωc ωc
Band−pass
1
H(jω)
0 ω
0
ωL ωH
Ideal filters
Low−pass High−pass
1 1
H(jω)
H(jω)
0 ω 0 ω
0 0
ωc ωc
Band−pass Band−reject
1 1
H(jω)
H(jω)
0 ω 0 ω
0 0
ωL ωH ωL ωH
1
v1
v2
Filter transfer function Filter output
0
1
v3 1
H(jω)
−1
0
1.5
0 −1
0 0 0.5 1 1.5 2 2.5 0 5 10 15 20
f (kHz) t (msec)
−1.5
0 5 10 15 20
t (msec)
1
v1
v2
Filter transfer function Filter output
0
1
v3 1
H(jω)
−1
0
1.5
0 −1
0 0 0.5 1 1.5 2 2.5 0 5 10 15 20
f (kHz) t (msec)
−1.5
0 5 10 15 20
t (msec)
1
v1
v2
Filter transfer function Filter output
0
1
v3 1
H(jω)
−1
0
1.5
0 −1
0 0 0.5 1 1.5 2 2.5 0 5 10 15 20
f (kHz) t (msec)
−1.5
0 5 10 15 20
t (msec)
1
v1
v2
Filter transfer function Filter output
0
1.5
v3 1
−1 H(jω)
0
1.5
0 −1.5
0 0 0.5 1 1.5 2 2.5 0 5 10 15 20
f (kHz) t (msec)
−1.5
0 5 10 15 20
t (msec)
* In practical filter circuits, the ideal filter response is approximated with a suitable H(jω)
that can be obtained with circuit elements. For example,
1
H(s) =
a 5 s 5 + a4 s 4 + a3 s 3 + a2 s 2 + a1 s + a0
represents a 5th -order low-pass filter.
* In practical filter circuits, the ideal filter response is approximated with a suitable H(jω)
that can be obtained with circuit elements. For example,
1
H(s) =
a 5 s 5 + a4 s 4 + a3 s 3 + a2 s 2 + a1 s + a0
represents a 5th -order low-pass filter.
* Some commonly used approximations (polynomials) are the Butterworth, Chebyshev, Bessel,
and elliptic functions.
* In practical filter circuits, the ideal filter response is approximated with a suitable H(jω)
that can be obtained with circuit elements. For example,
1
H(s) =
a 5 s 5 + a4 s 4 + a3 s 3 + a2 s 2 + a1 s + a0
represents a 5th -order low-pass filter.
* Some commonly used approximations (polynomials) are the Butterworth, Chebyshev, Bessel,
and elliptic functions.
* Coefficients for these filters are listed in filter handbooks. Also, programs for filter design
are available on the internet.
* A practical filter may exhibit a ripple. Amax is called the maximum passband ripple, e.g., Amax = 1 dB.
* A practical filter may exhibit a ripple. Amax is called the maximum passband ripple, e.g., Amax = 1 dB.
* Amin is the minimum attenuation to be provided by the filter, e.g., Amin = 60 dB.
* A practical filter may exhibit a ripple. Amax is called the maximum passband ripple, e.g., Amax = 1 dB.
* Amin is the minimum attenuation to be provided by the filter, e.g., Amin = 60 dB.
* ωs : edge of the stop band.
* A practical filter may exhibit a ripple. Amax is called the maximum passband ripple, e.g., Amax = 1 dB.
* Amin is the minimum attenuation to be provided by the filter, e.g., Amin = 60 dB.
* ωs : edge of the stop band.
* ωs /ωc (for a low-pass filter): selectivity factor, a measure of the sharpness of the filter.
* A practical filter may exhibit a ripple. Amax is called the maximum passband ripple, e.g., Amax = 1 dB.
* Amin is the minimum attenuation to be provided by the filter, e.g., Amin = 60 dB.
* ωs : edge of the stop band.
* ωs /ωc (for a low-pass filter): selectivity factor, a measure of the sharpness of the filter.
* ωc < ω < ωs : transition band.
1
For a low-pass filter, H(s) = n .
X
ai (s/ωc )i
i=0
Coefficients (ai ) for various types of filters are tabulated in handbooks. We now look at |H(jω)| for two
commonly used filters.
1
For a low-pass filter, H(s) = n .
X
ai (s/ωc )i
i=0
Coefficients (ai ) for various types of filters are tabulated in handbooks. We now look at |H(jω)| for two
commonly used filters.
Butterworth filters:
1
|H(jω)| = p .
1 + (ω/ωc )2n
2
1
For a low-pass filter, H(s) = n .
X
ai (s/ωc )i
i=0
Coefficients (ai ) for various types of filters are tabulated in handbooks. We now look at |H(jω)| for two
commonly used filters.
Butterworth filters:
1
|H(jω)| = p .
1 + (ω/ωc )2n
2
Chebyshev filters:
1
|H(jω)| = p where
1 + 2 Cn2 (ω/ωc )
1
For a low-pass filter, H(s) = n .
X
ai (s/ωc )i
i=0
Coefficients (ai ) for various types of filters are tabulated in handbooks. We now look at |H(jω)| for two
commonly used filters.
Butterworth filters:
1
|H(jω)| = p .
1 + (ω/ωc )2n
2
Chebyshev filters:
1
|H(jω)| = p where
1 + 2 Cn2 (ω/ωc )
H(s) for a high-pass filter can be obtained from H(s) of the corresponding low-pass filter by (s/ωc ) → (ωc /s) .
Butterworth filters:
1 ǫ = 0.5 0
n=1
|H| (dB)
n=1 2
|H|
2 3
3
4 5 4
−100
5
0
0 1 2 3 4 5 0.01 0.1 1 10 100
ω/ωc ω/ωc
Chebyshev filters:
1 ǫ = 0.5 0
n=1
|H| (dB)
n=1
2
|H|
2
3
3 4
4 −100
5
5
0
0 1 2 3 4 5 0.01 0.1 1 10 100
ω/ωc ω/ωc M. B. Patil, IIT Bombay
Practical filters (high-pass)
Butterworth filters:
1 0
n=1
n=1
|H| (dB)
2
|H|
2
3
3
4
4 5
5 ǫ = 0.5 −100
0
0 1 2 3 4 0.01 0.1 1 10 100
ω/ωc ω/ωc
Chebyshev filters:
1 0
n=1
n=1
|H| (dB)
2
|H|
3
2
4
5
3
4 ǫ = 0.5 −100
5
0
0 1 2 3 4 0.01 0.1 1 10 100
ω/ωc ω/ωc M. B. Patil, IIT Bombay
Passive filter example
R
Vs Vo
100 Ω
C
5 µF
Passive filter example
R
Vs Vo (1/sC) 1
100 Ω H(s) = = ,
R + (1/sC) 1 + (s/ω0 )
C
5 µF with ω0 = 1/RC → f0 = ω0 /2π = 318 Hz
(Low−pass filter)
Passive filter example
R
Vs Vo (1/sC) 1
100 Ω H(s) = = ,
R + (1/sC) 1 + (s/ω0 )
C
5 µF with ω0 = 1/RC → f0 = ω0 /2π = 318 Hz
(Low−pass filter)
20
0
|H| (dB)
−20
−40
−60
101 102 103 104 105
f (Hz)
(SEQUEL file: ee101 rc ac 2.sqproj)
−20
|H| (dB)
−40
−60
−80
102 103 104 105
f (Hz)
(SEQUEL file: ee101 rlc 3.sqproj)
* Op-amp filters can be designed without using inductors. This is a significant advantage since inductors are
bulky and expensive. Inductors also exhibit nonlinear behaviour (arising from the core properties) which is
undesirable in a filter circuit.
* Op-amp filters can be designed without using inductors. This is a significant advantage since inductors are
bulky and expensive. Inductors also exhibit nonlinear behaviour (arising from the core properties) which is
undesirable in a filter circuit.
* With op-amps, a filter circuit can be designed with a pass-band gain.
* Op-amp filters can be designed without using inductors. This is a significant advantage since inductors are
bulky and expensive. Inductors also exhibit nonlinear behaviour (arising from the core properties) which is
undesirable in a filter circuit.
* With op-amps, a filter circuit can be designed with a pass-band gain.
* Op-amp filters can be easily incorporated in an integrated circuit.
* Op-amp filters can be designed without using inductors. This is a significant advantage since inductors are
bulky and expensive. Inductors also exhibit nonlinear behaviour (arising from the core properties) which is
undesirable in a filter circuit.
* With op-amps, a filter circuit can be designed with a pass-band gain.
* Op-amp filters can be easily incorporated in an integrated circuit.
* However, there are situations in which passive filters are still used.
- high frequencies at which op-amps do not have sufficient gain
- high power which op-amps cannot handle
RL
Op-amp filters: example
R2
10 k
C
R1 10 nF
Vs
1k
Vo
RL
Op-amp filters are designed for op-amp operation in the linear region
→ Our analysis of the inverting amplifier applies, and we get,
R2 k (1/sC )
Vo = − Vs (Vs and Vo are phasors)
R1
R2 1
H(s) = −
R1 1 + sR2 C
Op-amp filters: example
R2
10 k
C
R1 10 nF
Vs
1k
Vo
RL
Op-amp filters are designed for op-amp operation in the linear region
→ Our analysis of the inverting amplifier applies, and we get,
R2 k (1/sC )
Vo = − Vs (Vs and Vo are phasors)
R1
R2 1
H(s) = −
R1 1 + sR2 C
This is a low-pass filter, with ω0 = 1/R2 C (i.e., f0 = ω0 /2π = 1.59 kHz).
Op-amp filters: example
R2
20
10 k
C
R1 10 nF
|H| (dB)
Vs
1k 0
Vo
RL
−20
101 102 103 104 105
f (Hz)
Op-amp filters are designed for op-amp operation in the linear region
→ Our analysis of the inverting amplifier applies, and we get,
R2 k (1/sC )
Vo = − Vs (Vs and Vo are phasors)
R1
R2 1
H(s) = −
R1 1 + sR2 C
This is a low-pass filter, with ω0 = 1/R2 C (i.e., f0 = ω0 /2π = 1.59 kHz).
Op-amp filters: example
R2
20
10 k
C
R1 10 nF
|H| (dB)
Vs
1k 0
Vo
RL
−20
101 102 103 104 105
f (Hz)
Op-amp filters are designed for op-amp operation in the linear region
→ Our analysis of the inverting amplifier applies, and we get,
R2 k (1/sC )
Vo = − Vs (Vs and Vo are phasors)
R1
R2 1
H(s) = −
R1 1 + sR2 C
This is a low-pass filter, with ω0 = 1/R2 C (i.e., f0 = ω0 /2π = 1.59 kHz).
(SEQUEL file: ee101 op filter 1.sqproj)
M. B. Patil, IIT Bombay
Op-amp filters: example
R2
C 10 k
R1
Vs
1k 100 nF Vo
RL
Op-amp filters: example
R2
C 10 k
R1
Vs
1k 100 nF Vo
RL
R2 sR2 C
H(s) = − =− .
R1 + (1/sC ) 1 + sR1 C
Op-amp filters: example
R2
C 10 k
R1
Vs
1k 100 nF Vo
RL
R2 sR2 C
H(s) = − =− .
R1 + (1/sC ) 1 + sR1 C
This is a high-pass filter, with ω0 = 1/R1 C (i.e., f0 = ω0 /2π = 1.59 kHz).
Op-amp filters: example
20
R2
C 10 k
R1
0
Vs
|H| (dB)
1k 100 nF Vo
RL −20
−40
101 102 103 104 105
f (Hz)
R2 sR2 C
H(s) = − =− .
R1 + (1/sC ) 1 + sR1 C
This is a high-pass filter, with ω0 = 1/R1 C (i.e., f0 = ω0 /2π = 1.59 kHz).
Op-amp filters: example
20
R2
C 10 k
R1
0
Vs
|H| (dB)
1k 100 nF Vo
RL −20
−40
101 102 103 104 105
f (Hz)
R2 sR2 C
H(s) = − =− .
R1 + (1/sC ) 1 + sR1 C
This is a high-pass filter, with ω0 = 1/R1 C (i.e., f0 = ω0 /2π = 1.59 kHz).
R2
100 k
C2
R1 C1 80 pF
Vs
10 k 0.8 µF Vo
RL
Op-amp filters: example
R2
100 k
C2
R1 C1 80 pF
Vs
10 k 0.8 µF Vo
RL
R2 k (1/sC2 ) R2 sR1 C1
H(s) = − =− .
R1 + (1/sC1 ) R1 (1 + sR1 C1 )(1 + sR2 C2 )
Op-amp filters: example
R2
100 k
C2
R1 C1 80 pF
Vs
10 k 0.8 µF Vo
RL
R2 k (1/sC2 ) R2 sR1 C1
H(s) = − =− .
R1 + (1/sC1 ) R1 (1 + sR1 C1 )(1 + sR2 C2 )
R2
20
100 k
C2
|H| (dB)
R1 C1 80 pF
Vs
10 k 0.8 µF Vo
RL
0
100 102 104 106
f (Hz)
R2 k (1/sC2 ) R2 sR1 C1
H(s) = − =− .
R1 + (1/sC1 ) R1 (1 + sR1 C1 )(1 + sR2 C2 )
R2
20
100 k
C2
|H| (dB)
R1 C1 80 pF
Vs
10 k 0.8 µF Vo
RL
0
100 102 104 106
f (Hz)
R2 k (1/sC2 ) R2 sR1 C1
H(s) = − =− .
R1 + (1/sC1 ) R1 (1 + sR1 C1 )(1 + sR2 C2 )
C1
20
a 1−a
R1A R2 R1B
Vs C2 0.7 a=0.9
|H| (dB)
0.5
R3A R3B 0
R1A = R1B = 470 Ω 0.3
C1
20
a 1−a
R1A R2 R1B
Vs C2 0.7 a=0.9
|H| (dB)
0.5
R3A R3B 0
R1A = R1B = 470 Ω 0.3
* Equalizers are implemented as arrays of narrow-band filters, each with an adjustable gain (attenuation)
around a centre frequency.
C1
20
a 1−a
R1A R2 R1B
Vs C2 0.7 a=0.9
|H| (dB)
0.5
R3A R3B 0
R1A = R1B = 470 Ω 0.3
* Equalizers are implemented as arrays of narrow-band filters, each with an adjustable gain (attenuation)
around a centre frequency.
* The circuit shown above represents one of the equalizer sections.
(SEQUEL file: ee101 op filter 4.sqproj)
M. B. Patil, IIT Bombay
M. B. Patil, IIT Bombay
Sallen-Key filter example (2nd order, low-pass)
40
C1
R1 20
R2
Vs
V1
Vo 0
|H| (dB)
C2
RL −20
R1 = R2 = 15.8 kΩ RB
C1 = C2 = 10 nF RA −40
RA = 10 kΩ, RB = 17.8 kΩ
−60
(Ref.: S. Franco, "Design with Op Amps and analog ICs") 101 102 103 104 105
f (Hz)
C1
R1 20
R2
Vs
V1
Vo 0
|H| (dB)
C2
RL −20
R1 = R2 = 15.8 kΩ RB
C1 = C2 = 10 nF RA −40
RA = 10 kΩ, RB = 17.8 kΩ
−60
(Ref.: S. Franco, "Design with Op Amps and analog ICs") 101 102 103 104 105
f (Hz)
RA
V+ = V− = Vo ≡ Vo /K .
RA + RB
C1
R1 20
R2
Vs
V1
Vo 0
|H| (dB)
C2
RL −20
R1 = R2 = 15.8 kΩ RB
C1 = C2 = 10 nF RA −40
RA = 10 kΩ, RB = 17.8 kΩ
−60
(Ref.: S. Franco, "Design with Op Amps and analog ICs") 101 102 103 104 105
f (Hz)
RA
V+ = V− = Vo ≡ Vo /K .
RA + RB
(1/sC2 ) 1
Also, V+ = V1 = V1 .
R2 + (1/sC2 ) 1 + sR2 C2
C1
R1 20
R2
Vs
V1
Vo 0
|H| (dB)
C2
RL −20
R1 = R2 = 15.8 kΩ RB
C1 = C2 = 10 nF RA −40
RA = 10 kΩ, RB = 17.8 kΩ
−60
(Ref.: S. Franco, "Design with Op Amps and analog ICs") 101 102 103 104 105
f (Hz)
RA
V+ = V− = Vo ≡ Vo /K .
RA + RB
(1/sC2 ) 1
Also, V+ = V1 = V1 .
R2 + (1/sC2 ) 1 + sR2 C2
1 1
KCL at V1 → (Vs − V1 ) + sC1 (Vo − V1 ) + (V+ − V1 ) = 0 .
R1 R2
C1
R1 20
R2
Vs
V1
Vo 0
|H| (dB)
C2
RL −20
R1 = R2 = 15.8 kΩ RB
C1 = C2 = 10 nF RA −40
RA = 10 kΩ, RB = 17.8 kΩ
−60
(Ref.: S. Franco, "Design with Op Amps and analog ICs") 101 102 103 104 105
f (Hz)
RA
V+ = V− = Vo ≡ Vo /K .
RA + RB
(1/sC2 ) 1
Also, V+ = V1 = V1 .
R2 + (1/sC2 ) 1 + sR2 C2
1 1
KCL at V1 → (Vs − V1 ) + sC1 (Vo − V1 ) + (V+ − V1 ) = 0 .
R1 R2
K
Combining the above equations, H(s) = .
1 + s [(R1 + R2 )C2 + (1 − K )R1 C1 ] + s 2 R1 C1 R2 C2
(SEQUEL file: ee101 op filter 5.sqproj)
M. B. Patil, IIT Bombay
Sixth-order Chebyshev low-pass filter (cascade design)
5.1 n 10 n 62 n
Vo
Vs
10.7 k 10.2 k 8.25 k 6.49 k 4.64 k 2.49 k
2.2 n 510 p 220 p RL
20
(Ref.: S. Franco, "Design with Op Amps and analog ICs")
−20
|H| (dB)
−40
−60
−80
102 103 104 105
f (Hz)
M. B. Patil, IIT Bombay
Third-order Chebyshev high-pass filter
20
100 n 15.4 k 154 k
7.68 k 0
Vo
Vs −20
|H| (dB)
100 n 100 n RL
54.9 k
−40
−60
(Ref.: S. Franco, "Design with Op Amps and analog ICs")
40
5k
5k
7.4 n
5k 20
7.4 n
Vs 5k
|H| (dB)
5k
0
−20
370 k Vo
5k −40
102 103 104 105
(Ref.: J. M. Fiore, "Op Amps and linear ICs") f (Hz)
SEQUEL file: ee101_op_filter_8.sqproj
Vo
89 k 10 k
1k
|H| (dB)
−20
−40
101 102
f (Hz)
M. B. Patil, IIT Bombay