Skip to main content

Chapter 3 · 2 hours

Frequency Transformation

IOE past exam questions

Past questions and answers

18 questions set from this chapter, 5 of them more than once. Most asked first.

  • Asked 5 times
  • 2074 Chaitra · 2+4 marks
  • 2072 Kartik · 4 marks
  • 2071 Shrawan · 5 marks
  • 2069 Chaitra · 4 marks
  • 2081 Bhadra · 4 marks

What is frequency transformation? Describe (with necessary derivations) the frequency transformation from low pass to band stop filter with a suitable example.

Answer

Frequency transformation

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies.

Low-pass to band-stop transformation

Transformation: replace ss in the low-pass prototype by

s  →  B ss2+ω02,ω0=ω1ω2,B=ω2−ω1s \;\rightarrow\; \frac{B\,s}{s^2 + \omega_0^2}, \qquad \omega_0 = \sqrt{\omega_1\omega_2},\quad B = \omega_2 - \omega_1

It is the band-pass transformation applied to the high-pass (s→1/ss \to 1/s) version. On the jωj\omega axis, ΩLP=Bωω02−ω2\Omega_{LP} = \dfrac{B\omega}{\omega_0^2 - \omega^2}, so:

  • ω=0\omega = 0 and ω=∞⇒ΩLP=0\omega = \infty \Rightarrow \Omega_{LP} = 0 (passbands at low and high frequencies);
  • ω=ω0⇒ΩLP=∞\omega = \omega_0 \Rightarrow \Omega_{LP} = \infty (maximum attenuation, the notch);
  • ω=ω1,ω2⇒∣ΩLP∣=1\omega = \omega_1, \omega_2 \Rightarrow |\Omega_{LP}| = 1 (band edges).

Element transformation:

Series inductor LL:

Z=LBss2+ω02  ⇒  Y=sLB+ω02LBsZ = \frac{LBs}{s^2+\omega_0^2} \;\Rightarrow\; Y = \frac{s}{LB} + \frac{\omega_0^2}{LBs}

becomes a parallel LC tank in the series arm: C′=1LBC' = \dfrac{1}{LB}, L′=LBω02L' = \dfrac{LB}{\omega_0^2}.

Shunt capacitor CC:

Y=CBss2+ω02  ⇒  Z=sCB+ω02CBsY = \frac{CBs}{s^2+\omega_0^2} \;\Rightarrow\; Z = \frac{s}{CB} + \frac{\omega_0^2}{CBs}

becomes a series LC in the shunt arm: L′=1CBL' = \dfrac{1}{CB}, C′=CBω02C' = \dfrac{CB}{\omega_0^2}.

Resistors are unchanged; every LC pair resonates at ω0\omega_0. At ω0\omega_0 the series-arm tanks are open and the shunt-arm series LCs are shorts, so no signal reaches the load.

 LP element       BS element
 series L   -->   series arm: L' || C'
 shunt C    -->   shunt arm:  L'--C' in series
 R          -->   R

Example: third-order Butterworth prototype (Rs=RL=1 ΩR_s = R_L = 1\ \Omega, L1=1L_1 = 1 H, C2=2C_2 = 2 F, L3=1L_3 = 1 H) converted to a band-stop filter with ω0=2000\omega_0 = 2000 rad/s and B=400B = 400 rad/s (ω02=4×106\omega_0^2 = 4\times10^6).

L1=1 H:L1′=L1Bω02=4004×106=0.1 mH,C1′=1L1B=1400=2.5 mFC2=2 F:L2′=1C2B=1800=1.25 mH,C2′=C2Bω02=8004×106=0.2 mFL3=1 H:L3′=0.1 mH,C3′=2.5 mF\begin{aligned} L_1 = 1\ \text{H}:&\quad L_1' = \frac{L_1B}{\omega_0^2} = \frac{400}{4\times10^6} = 0.1\ \text{mH},\quad C_1' = \frac{1}{L_1B} = \frac{1}{400} = 2.5\ \text{mF} \\ C_2 = 2\ \text{F}:&\quad L_2' = \frac{1}{C_2B} = \frac{1}{800} = 1.25\ \text{mH},\quad C_2' = \frac{C_2B}{\omega_0^2} = \frac{800}{4\times10^6} = 0.2\ \text{mF} \\ L_3 = 1\ \text{H}:&\quad L_3' = 0.1\ \text{mH},\quad C_3' = 2.5\ \text{mF} \end{aligned}
  Rs=1   [ Z1 ]          [ Z3 ]
 o-/\/\--[    ]---+------[    ]---+-----o
 |                |               |     +
 Vs            [ Y2 ]           RL=1    Vo
 |                |               |     -
 o----------------+---------------+-----o
ArmPrototypeBand-stop element
Z1 (series)L1=1L_1 = 1 H0.1 mH ∥ 2.5 mF (tank)
Y2 (shunt)C2=2C_2 = 2 F1.25 mH in series with 0.2 mF
Z3 (series)L3=1L_3 = 1 H0.1 mH ∥ 2.5 mF (tank)

Check: 1/0.1 mH×2.5 mF=1/1.25 mH×0.2 mF=20001/\sqrt{0.1\text{ mH}\times2.5\text{ mF}} = 1/\sqrt{1.25\text{ mH}\times0.2\text{ mF}} = 2000 rad/s. Band edges: ω1,2=ω02+(B/2)2∓B/2=1810.0\omega_{1,2} = \sqrt{\omega_0^2 + (B/2)^2} \mp B/2 = 1810.0 and 2210.02210.0 rad/s (3 dB points).

These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • Asked 5 times
  • 2082 Bhadra · 4 marks
  • 2082 Baisakh · 4 marks
  • 2081 Baisakh · 5 marks
  • 2078 Bhadra · 1+4 marks
  • 2079 Bhadra · 5 marks

What is frequency transformation? Describe the frequency transformation from (prototype) low pass to band pass filter with necessary derivations (expressions for R, L and C) and a suitable example.

Answer

Frequency transformation

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies.

Low-pass to band-pass transformation

Transformation: replace ss in the low-pass prototype by

s  →  s2+ω02B s,ω0=ω1ω2,B=ω2−ω1s \;\rightarrow\; \frac{s^2 + \omega_0^2}{B\,s}, \qquad \omega_0 = \sqrt{\omega_1\omega_2},\quad B = \omega_2 - \omega_1

On the jωj\omega axis this gives ΩLP=ω2−ω02Bω\Omega_{LP} = \dfrac{\omega^2 - \omega_0^2}{B\omega}. So:

  • ω=ω0⇒ΩLP=0\omega = \omega_0 \Rightarrow \Omega_{LP} = 0 (centre of passband maps to DC of the LP);
  • ω=ω1,ω2⇒ΩLP=∓1\omega = \omega_1, \omega_2 \Rightarrow \Omega_{LP} = \mp1, since ω1ω2=ω02\omega_1\omega_2 = \omega_0^2 and ω2−ω1=B\omega_2 - \omega_1 = B;
  • ω=0\omega = 0 and ω=∞⇒ΩLP=∓∞\omega = \infty \Rightarrow \Omega_{LP} = \mp\infty (stopbands).

The band-pass response is geometrically symmetric about ω0\omega_0, and its order is twice that of the prototype.

Element transformation:

Series inductor LL:

Z=L s2+ω02Bs=LBs+1(BLω02)sZ = L\,\frac{s^2+\omega_0^2}{Bs} = \frac{L}{B}s + \frac{1}{\left(\dfrac{B}{L\omega_0^2}\right)s}

becomes a series LC: L′=LBL' = \dfrac{L}{B}, C′=BLω02C' = \dfrac{B}{L\omega_0^2}.

Shunt capacitor CC:

Y=C s2+ω02Bs=CBs+1(BCω02)sY = C\,\frac{s^2+\omega_0^2}{Bs} = \frac{C}{B}s + \frac{1}{\left(\dfrac{B}{C\omega_0^2}\right)s}

becomes a parallel LC to ground: C′=CBC' = \dfrac{C}{B}, L′=BCω02L' = \dfrac{B}{C\omega_0^2}.

Resistor RR: unchanged (frequency independent). Every new LC pair resonates at ω0\omega_0 (L′C′=1/ω02L'C' = 1/\omega_0^2).

 LP element       BP element
 series L   -->   series L'--C'
 shunt C    -->   shunt  L' || C'
 R          -->   R

Example: third-order Butterworth prototype (Rs=RL=1 ΩR_s = R_L = 1\ \Omega, series L1=1L_1 = 1 H, shunt C2=2C_2 = 2 F, series L3=1L_3 = 1 H) converted to a band-pass filter with ω0=2000\omega_0 = 2000 rad/s and B=400B = 400 rad/s (ω02=4×106\omega_0^2 = 4\times10^6).

L1=1:L1′=1400=2.5 mH,C1′=4001×4×106=100 μFC2=2:C2′=2400=5 mF,L2′=4002×4×106=50 μHL3=1:L3′=2.5 mH,C3′=100 μF\begin{aligned} L_1 = 1:&\quad L_1' = \frac{1}{400} = 2.5\ \text{mH},\quad C_1' = \frac{400}{1\times4\times10^6} = 100\ \mu\text{F} \\ C_2 = 2:&\quad C_2' = \frac{2}{400} = 5\ \text{mF},\quad L_2' = \frac{400}{2\times4\times10^6} = 50\ \mu\text{H} \\ L_3 = 1:&\quad L_3' = 2.5\ \text{mH},\quad C_3' = 100\ \mu\text{F} \end{aligned}
  Rs=1   [ Z1 ]          [ Z3 ]
 o-/\/\--[    ]---+------[    ]---+-----o
 |                |               |     +
 Vs            [ Y2 ]           RL=1    Vo
 |                |               |     -
 o----------------+---------------+-----o
ArmPrototypeBand-pass element
Z1 (series)L1=1L_1 = 1 H2.5 mH in series with 100 μF
Y2 (shunt)C2=2C_2 = 2 F5 mF ∥ 50 μH
Z3 (series)L3=1L_3 = 1 H2.5 mH in series with 100 μF

Check: 1/2.5 mH×100 μF=20001/\sqrt{2.5\text{ mH}\times100\ \mu\text{F}} = 2000 rad/s. The 3 dB band edges are ω1,2=ω02+(B/2)2∓B/2=1810.0\omega_{1,2} = \sqrt{\omega_0^2+(B/2)^2} \mp B/2 = 1810.0 and 2210.02210.0 rad/s, so ω2−ω1=400\omega_2 - \omega_1 = 400 rad/s. The result is a 6th-order band-pass filter.

These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • Asked 3 times
  • 2079 Bhadra · 1+4 marks
  • 2079 Baisakh · 1+4 marks
  • 2076 Chaitra · 1+4 marks

What is frequency transformation? Design a band stop filter having center frequency 2000 rad/s and bandwidth 400 rad/s from a third order Butterworth low pass filter. [Refer Table of doubly terminated Butterworth element values]

Answer

Frequency transformation

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies. For a band-stop filter, s→Bss2+ω02s \to \dfrac{Bs}{s^2+\omega_0^2}.

Design of the band-stop filter

Prototype (table of doubly terminated Butterworth values, n=3n = 3, Rs=RL=1 ΩR_s = R_L = 1\ \Omega): series L1=1L_1 = 1 H, shunt C2=2C_2 = 2 F, series L3=1L_3 = 1 H.

Element rules for LP → BS:

  • series LL → parallel tank in the series arm: L′=LBω02L' = \dfrac{LB}{\omega_0^2}, C′=1LBC' = \dfrac{1}{LB};
  • shunt CC → series LC in the shunt arm: L′=1CBL' = \dfrac{1}{CB}, C′=CBω02C' = \dfrac{CB}{\omega_0^2}.
L1=1 H:L1′=L1Bω02=4004×106=0.1 mH,C1′=1L1B=1400=2.5 mFC2=2 F:L2′=1C2B=1800=1.25 mH,C2′=C2Bω02=8004×106=0.2 mFL3=1 H:L3′=0.1 mH,C3′=2.5 mF\begin{aligned} L_1 = 1\ \text{H}:&\quad L_1' = \frac{L_1B}{\omega_0^2} = \frac{400}{4\times10^6} = 0.1\ \text{mH},\quad C_1' = \frac{1}{L_1B} = \frac{1}{400} = 2.5\ \text{mF} \\ C_2 = 2\ \text{F}:&\quad L_2' = \frac{1}{C_2B} = \frac{1}{800} = 1.25\ \text{mH},\quad C_2' = \frac{C_2B}{\omega_0^2} = \frac{800}{4\times10^6} = 0.2\ \text{mF} \\ L_3 = 1\ \text{H}:&\quad L_3' = 0.1\ \text{mH},\quad C_3' = 2.5\ \text{mF} \end{aligned}
  Rs=1   [ Z1 ]          [ Z3 ]
 o-/\/\--[    ]---+------[    ]---+-----o
 |                |               |     +
 Vs            [ Y2 ]           RL=1    Vo
 |                |               |     -
 o----------------+---------------+-----o
ArmPrototypeBand-stop element
Z1 (series)L1=1L_1 = 1 H0.1 mH ∥ 2.5 mF (tank)
Y2 (shunt)C2=2C_2 = 2 F1.25 mH in series with 0.2 mF
Z3 (series)L3=1L_3 = 1 H0.1 mH ∥ 2.5 mF (tank)

Check: 1/0.1 mH×2.5 mF=1/1.25 mH×0.2 mF=20001/\sqrt{0.1\text{ mH}\times2.5\text{ mF}} = 1/\sqrt{1.25\text{ mH}\times0.2\text{ mF}} = 2000 rad/s. Band edges: ω1,2=ω02+(B/2)2∓B/2=1810.0\omega_{1,2} = \sqrt{\omega_0^2 + (B/2)^2} \mp B/2 = 1810.0 and 2210.02210.0 rad/s (3 dB points).

These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • Asked 3 times
  • 2070 Chaitra · 1+3 marks
  • 2070 Asar · 5 marks
  • 2081 Baisakh · 1+1+5 marks

What is the significance (importance) of frequency transformation in filter design? How can a band pass filter be obtained from a (normalized) prototype low pass filter? Explain with example.

Answer

Significance of frequency transformation

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies.

  • Only one set of approximation tables (low-pass poles and ladder element values) is needed; HP, BP and BS filters are derived from it.
  • The difficult approximation step (finding T(s)T(s) for given αmax\alpha_{max}, αmin\alpha_{min}) is done once, for the simpler low-pass case; specifications of the other types are converted to an equivalent LP specification.
  • It can be applied either to the transfer function (substitute for ss) or directly to each element of an LC ladder, so the circuit is obtained without re-deriving it.
  • It reduces design time and errors, and keeps the properties of the prototype (equiripple, maximally flat, etc.).

Band-pass filter from the normalized prototype low-pass filter

Transformation: replace ss in the low-pass prototype by

s  →  s2+ω02B s,ω0=ω1ω2,B=ω2−ω1s \;\rightarrow\; \frac{s^2 + \omega_0^2}{B\,s}, \qquad \omega_0 = \sqrt{\omega_1\omega_2},\quad B = \omega_2 - \omega_1

On the jωj\omega axis this gives ΩLP=ω2−ω02Bω\Omega_{LP} = \dfrac{\omega^2 - \omega_0^2}{B\omega}. So:

  • ω=ω0⇒ΩLP=0\omega = \omega_0 \Rightarrow \Omega_{LP} = 0 (centre of passband maps to DC of the LP);
  • ω=ω1,ω2⇒ΩLP=∓1\omega = \omega_1, \omega_2 \Rightarrow \Omega_{LP} = \mp1, since ω1ω2=ω02\omega_1\omega_2 = \omega_0^2 and ω2−ω1=B\omega_2 - \omega_1 = B;
  • ω=0\omega = 0 and ω=∞⇒ΩLP=∓∞\omega = \infty \Rightarrow \Omega_{LP} = \mp\infty (stopbands).

The band-pass response is geometrically symmetric about ω0\omega_0, and its order is twice that of the prototype.

Element transformation:

Series inductor LL:

Z=L s2+ω02Bs=LBs+1(BLω02)sZ = L\,\frac{s^2+\omega_0^2}{Bs} = \frac{L}{B}s + \frac{1}{\left(\dfrac{B}{L\omega_0^2}\right)s}

becomes a series LC: L′=LBL' = \dfrac{L}{B}, C′=BLω02C' = \dfrac{B}{L\omega_0^2}.

Shunt capacitor CC:

Y=C s2+ω02Bs=CBs+1(BCω02)sY = C\,\frac{s^2+\omega_0^2}{Bs} = \frac{C}{B}s + \frac{1}{\left(\dfrac{B}{C\omega_0^2}\right)s}

becomes a parallel LC to ground: C′=CBC' = \dfrac{C}{B}, L′=BCω02L' = \dfrac{B}{C\omega_0^2}.

Resistor RR: unchanged (frequency independent). Every new LC pair resonates at ω0\omega_0 (L′C′=1/ω02L'C' = 1/\omega_0^2).

 LP element       BP element
 series L   -->   series L'--C'
 shunt C    -->   shunt  L' || C'
 R          -->   R

Example (transfer function method): first-order prototype TLP(s)=1s+1T_{LP}(s) = \dfrac{1}{s+1}. Substituting s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs}:

TBP(s)=1s2+ω02Bs+1=Bss2+Bs+ω02T_{BP}(s) = \frac{1}{\dfrac{s^2+\omega_0^2}{Bs} + 1} = \frac{Bs}{s^2 + Bs + \omega_0^2}

This is the standard second-order band-pass function with centre frequency ω0\omega_0, bandwidth BB and Q=ω0/BQ = \omega_0/B. For ω0=1000\omega_0 = 1000 rad/s and B=100B = 100 rad/s: TBP(s)=100ss2+100s+106T_{BP}(s) = \dfrac{100s}{s^2 + 100s + 10^6}, Q=10Q = 10.

Example (element method): the third-order Butterworth ladder (1 H, 2 F, 1 H) with ω0=2000\omega_0 = 2000 rad/s, B=400B = 400 rad/s gives series arms of 2.5 mH in series with 100 μF and a shunt arm of 5 mF in parallel with 50 μH.

  • Asked 2 times
  • 2080 Baisakh · 1+4 marks
  • 2072 Chaitra · 6 marks
  • 2080 Baisakh · 5 marks

What is frequency transformation and what are its applications in filter design? How can you obtain a high pass filter from a given (normalized) low pass filter? Explain with a suitable example.

Answer

Frequency transformation and its applications

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies.

Applications in filter design:

  • Only one set of approximation tables (low-pass poles and ladder element values) is needed; HP, BP and BS filters are derived from it.
  • The difficult approximation step (finding T(s)T(s) for given αmax\alpha_{max}, αmin\alpha_{min}) is done once, for the simpler low-pass case; specifications of the other types are converted to an equivalent LP specification.
  • It can be applied either to the transfer function (substitute for ss) or directly to each element of an LC ladder, so the circuit is obtained without re-deriving it.
  • It reduces design time and errors, and keeps the properties of the prototype (equiripple, maximally flat, etc.).

High-pass filter from a normalized low-pass filter

Transformation: replace ss in the normalized low-pass prototype by

s  →  ωcs(normalized: s→1/s)s \;\rightarrow\; \frac{\omega_c}{s} \quad (\text{normalized: } s \to 1/s)

On the jωj\omega axis ΩLP=−ωc/ω\Omega_{LP} = -\omega_c/\omega: low frequencies map to the LP stopband and high frequencies to the LP passband, and ω=ωc\omega = \omega_c maps to the LP band edge Ω=1\Omega = 1. So the attenuation the LP has at Ω\Omega appears in the HP at ωc/Ω\omega_c/\Omega.

Element transformation (normalized, ωc=1\omega_c = 1):

  • inductor LL: Z=sL→L/s=1s(1/L)Z = sL \to L/s = \dfrac{1}{s(1/L)}, i.e. a capacitor C′=1/LC' = 1/L;
  • capacitor CC: Y=sC→C/s=1s(1/C)Y = sC \to C/s = \dfrac{1}{s(1/C)}, i.e. an inductor L′=1/CL' = 1/C;
  • resistors unchanged.

Then frequency-scale to ωc\omega_c and magnitude-scale to the required impedance.

Example: third-order Butterworth low-pass prototype (Rs=RL=1 ΩR_s = R_L = 1\ \Omega; series L1=1L_1 = 1 H, shunt C2=2C_2 = 2 F, series L3=1L_3 = 1 H). Required: high-pass, ωc=1000\omega_c = 1000 rad/s, terminations 1 kΩ.

  1. LP → HP (normalized): series C1=1/1=1C_1 = 1/1 = 1 F, shunt L2=1/2=0.5L_2 = 1/2 = 0.5 H, series C3=1C_3 = 1 F.
  2. Scale with kf=1000k_f = 1000, km=1000k_m = 1000 (C′=C/(kmkf)C' = C/(k_mk_f), L′=Lkm/kfL' = Lk_m/k_f):
C1′=C3′=1103×103=1 μF,L2′=0.5×103103=0.5 HC_1' = C_3' = \frac{1}{10^3\times10^3} = 1\ \mu\text{F}, \qquad L_2' = \frac{0.5\times10^3}{10^3} = 0.5\ \text{H}
  Rs=1k   1uF            1uF
 o-/\/\---||-----+--------||-----+-----o
 |               |               |     +
 Vs            0.5 H           RL=1k   Vo
 |               |               |     -
 o---------------+---------------+-----o

Transfer function check: TLP(s)=1s3+2s2+2s+1T_{LP}(s) = \dfrac{1}{s^3+2s^2+2s+1} (for equal terminations the passband gain is 1/2, omitted here); s→1000/ss \to 1000/s gives THP(s)=s3s3+2000s2+2×106s+109T_{HP}(s) = \dfrac{s^3}{s^3 + 2000s^2 + 2\times10^6 s + 10^9}, a third-order Butterworth high-pass with 3 dB frequency 1000 rad/s.

  • 2080 Bhadra · 1+4 marks

How frequency transformation reduces the design steps required to design a filter? Design a band stop filter having center frequency 2000 rad/s and bandwidth 400 rad/s from a 3rd order Butterworth low pass filter. (Refer Table)

Answer

How frequency transformation reduces design steps

Without frequency transformation, a band-stop filter would need its own approximation (finding a suitable T(s)T(s) with the required notch band) and its own synthesis, which is long and error-prone. With frequency transformation:

  1. The band-stop specification is converted to an equivalent normalized low-pass specification.
  2. The order and the ladder element values are read directly from the standard low-pass table (no new approximation).
  3. Each element of the low-pass ladder is replaced by a simple LC pair using fixed formulas (no new synthesis).

So one set of low-pass tables, plus a few substitution rules, covers HP, BP and BS designs.

Design of the band-stop filter

The LP → BS transformation is s→Bss2+ω02s \to \dfrac{Bs}{s^2+\omega_0^2}. Element rules:

  • series LL → parallel tank in the series arm: L′=LBω02L' = \dfrac{LB}{\omega_0^2}, C′=1LBC' = \dfrac{1}{LB};
  • shunt CC → series LC in the shunt arm: L′=1CBL' = \dfrac{1}{CB}, C′=CBω02C' = \dfrac{CB}{\omega_0^2}.

Prototype from the table (n=3n = 3, Rs=RL=1 ΩR_s = R_L = 1\ \Omega): series L1=1L_1 = 1 H, shunt C2=2C_2 = 2 F, series L3=1L_3 = 1 H.

L1=1 H:L1′=L1Bω02=4004×106=0.1 mH,C1′=1L1B=1400=2.5 mFC2=2 F:L2′=1C2B=1800=1.25 mH,C2′=C2Bω02=8004×106=0.2 mFL3=1 H:L3′=0.1 mH,C3′=2.5 mF\begin{aligned} L_1 = 1\ \text{H}:&\quad L_1' = \frac{L_1B}{\omega_0^2} = \frac{400}{4\times10^6} = 0.1\ \text{mH},\quad C_1' = \frac{1}{L_1B} = \frac{1}{400} = 2.5\ \text{mF} \\ C_2 = 2\ \text{F}:&\quad L_2' = \frac{1}{C_2B} = \frac{1}{800} = 1.25\ \text{mH},\quad C_2' = \frac{C_2B}{\omega_0^2} = \frac{800}{4\times10^6} = 0.2\ \text{mF} \\ L_3 = 1\ \text{H}:&\quad L_3' = 0.1\ \text{mH},\quad C_3' = 2.5\ \text{mF} \end{aligned}
  Rs=1   [ Z1 ]          [ Z3 ]
 o-/\/\--[    ]---+------[    ]---+-----o
 |                |               |     +
 Vs            [ Y2 ]           RL=1    Vo
 |                |               |     -
 o----------------+---------------+-----o
ArmPrototypeBand-stop element
Z1 (series)L1=1L_1 = 1 H0.1 mH ∥ 2.5 mF (tank)
Y2 (shunt)C2=2C_2 = 2 F1.25 mH in series with 0.2 mF
Z3 (series)L3=1L_3 = 1 H0.1 mH ∥ 2.5 mF (tank)

Check: 1/0.1 mH×2.5 mF=1/1.25 mH×0.2 mF=20001/\sqrt{0.1\text{ mH}\times2.5\text{ mF}} = 1/\sqrt{1.25\text{ mH}\times0.2\text{ mF}} = 2000 rad/s. Band edges: ω1,2=ω02+(B/2)2∓B/2=1810.0\omega_{1,2} = \sqrt{\omega_0^2 + (B/2)^2} \mp B/2 = 1810.0 and 2210.02210.0 rad/s (3 dB points).

These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2082 Chaitra (new course) · 3 marks

Define frequency transformation and mention its importance in filter design.

Answer

Frequency transformation is a substitution for the complex frequency variable ss that converts a normalized low-pass prototype filter (passband edge 1 rad/s) into a low-pass filter of another cutoff, or into a high-pass, band-pass or band-stop filter, while keeping the same type of response (Butterworth, Chebyshev, etc.).

Required filterSubstitution in TLP(s)T_{LP}(s)
Low-pass, edge ωp\omega_ps→s/ωps \to s/\omega_p
High-pass, edge ωc\omega_cs→ωc/ss \to \omega_c/s
Band-pass, ω0\omega_0, BBs→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs}
Band-stop, ω0\omega_0, BBs→Bss2+ω02s \to \dfrac{Bs}{s^2+\omega_0^2}

Importance in filter design:

  • Only the low-pass approximation problem has to be solved; tables of normalized LP poles and ladder element values serve all filter types.
  • It can be applied to the transfer function or element by element to an LC ladder (e.g. a series L becomes a series LC in a band-pass filter), so circuits are obtained directly.
  • It saves design effort and reduces errors.

Example: TLP(s)=1s+1T_{LP}(s) = \dfrac{1}{s+1} with s→ωc/ss \to \omega_c/s gives THP(s)=ss+ωcT_{HP}(s) = \dfrac{s}{s+\omega_c}.

  • 2083 Baisakh · 4 marks

Design a band-pass filter with center frequency 10 kHz and bandwidth 4 kHz using a normalized low-pass filter with transfer function T(s) = 1/(s² + 1.414s + 1).

Answer

Use the low-pass to band-pass transformation s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs} on the normalized prototype T(s)=1s2+1.414s+1T(s) = \dfrac{1}{s^2+1.414s+1} (second-order Butterworth).

Specifications in rad/s

ω0=2π×104=62 831.85 rad/s,ω02=3.9478×109B=2π×4000=25 132.74 rad/s,B2=6.3165×108\begin{aligned} \omega_0 &= 2\pi\times10^4 = 62\,831.85\ \text{rad/s}, \quad \omega_0^2 = 3.9478\times10^9 \\ B &= 2\pi\times4000 = 25\,132.74\ \text{rad/s}, \quad B^2 = 6.3165\times10^8 \end{aligned}

Substitution

TBP(s)=1(s2+ω02Bs)2+1.414s2+ω02Bs+1=B2s2(s2+ω02)2+1.414Bs(s2+ω02)+B2s2T_{BP}(s) = \frac{1}{\left(\dfrac{s^2+\omega_0^2}{Bs}\right)^2 + 1.414\dfrac{s^2+\omega_0^2}{Bs} + 1} = \frac{B^2s^2}{(s^2+\omega_0^2)^2 + 1.414Bs(s^2+\omega_0^2) + B^2s^2}

Expanding the denominator:

s4+1.414B s3+(2ω02+B2)s2+1.414Bω02 s+ω04s^4 + 1.414B\,s^3 + (2\omega_0^2 + B^2)s^2 + 1.414B\omega_0^2\,s + \omega_0^4
CoefficientExpressionValue
s3s^31.414B1.414B3.5538×1043.5538\times10^4
s2s^22ω02+B22\omega_0^2 + B^28.5273×1098.5273\times10^9
s1s^11.414Bω021.414B\omega_0^21.4030×10141.4030\times10^{14}
s0s^0ω04\omega_0^41.5585×10191.5585\times10^{19}
TBP(s)=6.3165×108 s2s4+3.5538×104s3+8.5273×109s2+1.4030×1014s+1.5585×1019T_{BP}(s) = \frac{6.3165\times10^8\, s^2}{s^4 + 3.5538\times10^4 s^3 + 8.5273\times10^9 s^2 + 1.4030\times10^{14}s + 1.5585\times10^{19}}

Check: at s=jω0s = j\omega_0, ∣TBP∣=1|T_{BP}| = 1 (0 dB at 10 kHz), and the 3 dB points are ΩLP=±1\Omega_{LP} = \pm1, i.e. 44 kHz apart.

Realization as two cascaded biquads

The LP poles −0.707±j0.707-0.707 \pm j0.707 each map to a pair of BP poles. Factoring the denominator:

TBP(s)=25 132.7 ss2+15 256.8s+2.9699×109⋅25 132.7 ss2+20 280.9s+5.2479×109T_{BP}(s) = \frac{25\,132.7\,s}{s^2 + 15\,256.8s + 2.9699\times10^9}\cdot\frac{25\,132.7\,s}{s^2 + 20\,280.9s + 5.2479\times10^9}
Sectionω0i\omega_{0i} (rad/s)f0if_{0i}QQ
154 496.58.673 kHz3.572
272 442.111.530 kHz3.572

Note f01f02=(10 kHz)2f_{01}f_{02} = (10\text{ kHz})^2. Each section can be built as a band-pass biquad (e.g. MFB or Tow-Thomas) and the two are cascaded with buffering.

Answer: a fourth-order band-pass filter with the transfer function above, centre frequency 10 kHz, 3 dB bandwidth 4 kHz.

  • 2081 Bhadra · 4 marks

Design a Band pass filter having center frequency at 1500 rad/sec and bandwidth 300 rad/sec from a 4th order Butterworth low pass resistively terminated lossless filter. [Refer Table of doubly terminated Butterworth element values]

Answer

Use the low-pass to band-pass transformation s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs} with ω0=1500\omega_0 = 1500 rad/s (ω02=2.25×106\omega_0^2 = 2.25\times10^6) and B=300B = 300 rad/s.

Prototype (table of doubly terminated Butterworth values, n=4n = 4, Rs=RL=1 ΩR_s = R_L = 1\ \Omega, taken as starting with a series inductor): L1=0.7654L_1 = 0.7654 H, C2=1.848C_2 = 1.848 F, L3=1.848L_3 = 1.848 H, C4=0.7654C_4 = 0.7654 F.

Element rules:

  • series LL → series L′=L/BL' = L/B and C′=B/(Lω02)C' = B/(L\omega_0^2);
  • shunt CC → parallel C′=C/BC' = C/B and L′=B/(Cω02)L' = B/(C\omega_0^2).
L1:L1′=0.7654300=2.5513 mH,C1′=3000.7654×2.25×106=174.20 μFC2:C2′=1.848300=6.16 mF,L2′=3001.848×2.25×106=72.150 μHL3:L3′=1.848300=6.16 mH,C3′=3001.848×2.25×106=72.150 μFC4:C4′=0.7654300=2.5513 mF,L4′=3000.7654×2.25×106=174.20 μH\begin{aligned} L_1:&\quad L_1' = \frac{0.7654}{300} = 2.5513\ \text{mH},\quad C_1' = \frac{300}{0.7654\times2.25\times10^6} = 174.20\ \mu\text{F} \\ C_2:&\quad C_2' = \frac{1.848}{300} = 6.16\ \text{mF},\quad L_2' = \frac{300}{1.848\times2.25\times10^6} = 72.150\ \mu\text{H} \\ L_3:&\quad L_3' = \frac{1.848}{300} = 6.16\ \text{mH},\quad C_3' = \frac{300}{1.848\times2.25\times10^6} = 72.150\ \mu\text{F} \\ C_4:&\quad C_4' = \frac{0.7654}{300} = 2.5513\ \text{mF},\quad L_4' = \frac{300}{0.7654\times2.25\times10^6} = 174.20\ \mu\text{H} \end{aligned}
  Rs=1  [ Z1 ]        [ Z3 ]
 o-/\/\-[    ]--+-----[    ]--+------+---o
 |              |             |      |   +
 Vs          [ Y2 ]        [ Y4 ]  RL=1  Vo
 |              |             |      |   -
 o--------------+-------------+------+---o
ArmBand-pass element
Z12.5513 mH in series with 174.20 μF
Y26.16 mF ∥ 72.150 μH
Z36.16 mH in series with 72.150 μF
Y42.5513 mF ∥ 174.20 μH

Each LC pair resonates at 1500 rad/s. The result is an 8th-order band-pass filter with 3 dB edges at ω1,2=15002+1502∓150=1357.5\omega_{1,2} = \sqrt{1500^2+150^2} \mp 150 = 1357.5 and 1657.51657.5 rad/s. These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2075 Chaitra · 1+1+4 marks

What is frequency transformation? What is its importance in filter design? Design a bandpass filter having ω0 = 2000 rad/s and B = 400 rad/s from a third order Butterworth lowpass filter. [Refer Table 1]

Answer

Frequency transformation

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies.

Importance in filter design

  • Only one set of approximation tables (low-pass poles and ladder element values) is needed; HP, BP and BS filters are derived from it.
  • The difficult approximation step (finding T(s)T(s) for given αmax\alpha_{max}, αmin\alpha_{min}) is done once, for the simpler low-pass case; specifications of the other types are converted to an equivalent LP specification.
  • It can be applied either to the transfer function (substitute for ss) or directly to each element of an LC ladder, so the circuit is obtained without re-deriving it.
  • It reduces design time and errors, and keeps the properties of the prototype (equiripple, maximally flat, etc.).

Design of the band-pass filter

LP → BP: s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs}, with ω0=2000\omega_0 = 2000 rad/s and B=400B = 400 rad/s.

  • series LL → series L′=L/BL' = L/B, C′=B/(Lω02)C' = B/(L\omega_0^2);
  • shunt CC → parallel C′=C/BC' = C/B, L′=B/(Cω02)L' = B/(C\omega_0^2).

Prototype from Table 1 (n=3n = 3, Rs=RL=1 ΩR_s = R_L = 1\ \Omega): series L1=1L_1 = 1 H, shunt C2=2C_2 = 2 F, series L3=1L_3 = 1 H.

L1=1:L1′=1400=2.5 mH,C1′=4001×4×106=100 μFC2=2:C2′=2400=5 mF,L2′=4002×4×106=50 μHL3=1:L3′=2.5 mH,C3′=100 μF\begin{aligned} L_1 = 1:&\quad L_1' = \frac{1}{400} = 2.5\ \text{mH},\quad C_1' = \frac{400}{1\times4\times10^6} = 100\ \mu\text{F} \\ C_2 = 2:&\quad C_2' = \frac{2}{400} = 5\ \text{mF},\quad L_2' = \frac{400}{2\times4\times10^6} = 50\ \mu\text{H} \\ L_3 = 1:&\quad L_3' = 2.5\ \text{mH},\quad C_3' = 100\ \mu\text{F} \end{aligned}
  Rs=1   [ Z1 ]          [ Z3 ]
 o-/\/\--[    ]---+------[    ]---+-----o
 |                |               |     +
 Vs            [ Y2 ]           RL=1    Vo
 |                |               |     -
 o----------------+---------------+-----o
ArmPrototypeBand-pass element
Z1 (series)L1=1L_1 = 1 H2.5 mH in series with 100 μF
Y2 (shunt)C2=2C_2 = 2 F5 mF ∥ 50 μH
Z3 (series)L3=1L_3 = 1 H2.5 mH in series with 100 μF

Check: 1/2.5 mH×100 μF=20001/\sqrt{2.5\text{ mH}\times100\ \mu\text{F}} = 2000 rad/s. The 3 dB band edges are ω1,2=ω02+(B/2)2∓B/2=1810.0\omega_{1,2} = \sqrt{\omega_0^2+(B/2)^2} \mp B/2 = 1810.0 and 2210.02210.0 rad/s, so ω2−ω1=400\omega_2 - \omega_1 = 400 rad/s. The result is a 6th-order band-pass filter.

These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2075 Asoj · 1+4 marks

What is the importance of frequency transformation? Obtain a bandpass filter having ω0 = 2000 rad/s and B = 400 rad/s from fourth order Butterworth lowpass filter. [Refer table 2]

Answer

Importance of frequency transformation

Frequency transformation converts a normalized low-pass prototype into HP, BP or BS filters by a change of variable, so:

  • Only one set of approximation tables (low-pass poles and ladder element values) is needed; HP, BP and BS filters are derived from it.
  • The difficult approximation step (finding T(s)T(s) for given αmax\alpha_{max}, αmin\alpha_{min}) is done once, for the simpler low-pass case; specifications of the other types are converted to an equivalent LP specification.
  • It can be applied either to the transfer function (substitute for ss) or directly to each element of an LC ladder, so the circuit is obtained without re-deriving it.
  • It reduces design time and errors, and keeps the properties of the prototype (equiripple, maximally flat, etc.).

Band-pass filter from the fourth-order Butterworth low-pass filter

LP → BP: s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs}, ω0=2000\omega_0 = 2000 rad/s (ω02=4×106\omega_0^2 = 4\times10^6), B=400B = 400 rad/s.

Prototype (Table 2, n=4n = 4, Rs=RL=1 ΩR_s = R_L = 1\ \Omega, starting with a series inductor): L1=0.7654L_1 = 0.7654 H, C2=1.848C_2 = 1.848 F, L3=1.848L_3 = 1.848 H, C4=0.7654C_4 = 0.7654 F.

L1:L1′=0.7654400=1.9135 mH,C1′=4000.7654×4×106=130.65 μFC2:C2′=1.848400=4.62 mF,L2′=4001.848×4×106=54.113 μHL3:L3′=4.62 mH,C3′=54.113 μFC4:C4′=1.9135 mF,L4′=130.65 μH\begin{aligned} L_1:&\quad L_1' = \frac{0.7654}{400} = 1.9135\ \text{mH},\quad C_1' = \frac{400}{0.7654\times4\times10^6} = 130.65\ \mu\text{F} \\ C_2:&\quad C_2' = \frac{1.848}{400} = 4.62\ \text{mF},\quad L_2' = \frac{400}{1.848\times4\times10^6} = 54.113\ \mu\text{H} \\ L_3:&\quad L_3' = 4.62\ \text{mH},\quad C_3' = 54.113\ \mu\text{F} \\ C_4:&\quad C_4' = 1.9135\ \text{mF},\quad L_4' = 130.65\ \mu\text{H} \end{aligned}
  Rs=1  [ Z1 ]        [ Z3 ]
 o-/\/\-[    ]--+-----[    ]--+------+---o
 |              |             |      |   +
 Vs          [ Y2 ]        [ Y4 ]  RL=1  Vo
 |              |             |      |   -
 o--------------+-------------+------+---o
ArmBand-pass element
Z1 (series)1.9135 mH in series with 130.65 μF
Y2 (shunt)4.62 mF ∥ 54.113 μH
Z3 (series)4.62 mH in series with 54.113 μF
Y4 (shunt)1.9135 mF ∥ 130.65 μH

All LC pairs resonate at 2000 rad/s; the 3 dB band edges are 1810.0 and 2210.0 rad/s. These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2074 Asoj · 1+3+4 marks

What is frequency transformation in filter design? How can you obtain a bandpass filter from given lowpass filter at normalized frequency? Obtain a bandpass filter having ω1 = 100 rad/s and ω2 = 10000 rad/s from following lowpass filter at normalized frequency. [Figure: source V1, 1 Ω source resistor, series 0.7654 H, shunt 1.8485 F, series 1.8485 H, shunt 0.7654 F, 1 Ω load (output V2)]

Answer

Frequency transformation

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies.

Obtaining a band-pass filter from the normalized low-pass filter

Replace ss by s2+ω02Bs\dfrac{s^2+\omega_0^2}{Bs}, where ω0=ω1ω2\omega_0 = \sqrt{\omega_1\omega_2} and B=ω2−ω1B = \omega_2 - \omega_1. Applied to each element:

  • series LL → series L′=L/BL' = L/B with C′=B/(Lω02)C' = B/(L\omega_0^2);
  • shunt CC → parallel C′=C/BC' = C/B with L′=B/(Cω02)L' = B/(C\omega_0^2);
  • resistors unchanged.

This maps ω0→0\omega_0 \to 0 and ω1,ω2→∓1\omega_1, \omega_2 \to \mp1 of the low-pass prototype, so the LP passband becomes the band ω1\omega_1 to ω2\omega_2.

Design for ω1=100\omega_1 = 100 rad/s, ω2=10000\omega_2 = 10000 rad/s

ω0=100×10000=1000 rad/s,ω02=106,B=10000−100=9900 rad/s\omega_0 = \sqrt{100\times10000} = 1000\ \text{rad/s}, \quad \omega_0^2 = 10^6, \quad B = 10000 - 100 = 9900\ \text{rad/s}

Given ladder (4th-order Butterworth, 1 Ω terminations): series 0.7654 H, shunt 1.8485 F, series 1.8485 H, shunt 0.7654 F.

0.7654 H:L′=0.76549900=77.313 μH,C′=99000.7654×106=12.934 mF1.8485 F:C′=1.84859900=186.72 μF,L′=99001.8485×106=5.3557 mH1.8485 H:L′=186.72 μH,C′=5.3557 mF0.7654 F:C′=77.313 μF,L′=12.934 mH\begin{aligned} 0.7654\ \text{H}:&\quad L' = \frac{0.7654}{9900} = 77.313\ \mu\text{H},\quad C' = \frac{9900}{0.7654\times10^6} = 12.934\ \text{mF} \\ 1.8485\ \text{F}:&\quad C' = \frac{1.8485}{9900} = 186.72\ \mu\text{F},\quad L' = \frac{9900}{1.8485\times10^6} = 5.3557\ \text{mH} \\ 1.8485\ \text{H}:&\quad L' = 186.72\ \mu\text{H},\quad C' = 5.3557\ \text{mF} \\ 0.7654\ \text{F}:&\quad C' = 77.313\ \mu\text{F},\quad L' = 12.934\ \text{mH} \end{aligned}
  Rs=1  [ Z1 ]        [ Z3 ]
 o-/\/\-[    ]--+-----[    ]--+------+---o
 |              |             |      |   +
 Vs          [ Y2 ]        [ Y4 ]  RL=1  Vo
 |              |             |      |   -
 o--------------+-------------+------+---o
ArmPrototypeBand-pass element
Z10.7654 H77.313 μH in series with 12.934 mF
Y21.8485 F186.72 μF ∥ 5.3557 mH
Z31.8485 H186.72 μH in series with 5.3557 mF
Y40.7654 F77.313 μF ∥ 12.934 mH

Check: 1/77.313 μH×12.934 mF=10001/\sqrt{77.313\ \mu\text{H}\times12.934\text{ mF}} = 1000 rad/s. These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2073 Chaitra · 2+4 marks

What is frequency transformation? Obtain the bandpass filter from lowpass filter given in figure 1 having center frequency 10⁴ rad/s and bandwidth of 9.9 × 10⁴ rad/s. [Figure 1: source V1, R1 = 1 Ω, series L1 = 2.024 H, shunt C1 = 0.994 F, series L2 = 2.024 H, load R2 = 1 Ω]

Answer

Frequency transformation

Frequency transformation is a change of the complex frequency variable that converts a normalized low-pass prototype filter (cutoff 1 rad/s) into a high-pass, band-pass or band-stop filter (or a low-pass with a different cutoff). The magnitude response of the prototype is mapped onto the new frequency axis, so the same attenuation values appear at the corresponding new frequencies. For band-pass: s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs}.

Band-pass filter from the given low-pass filter

Given: ω0=104\omega_0 = 10^4 rad/s (ω02=108\omega_0^2 = 10^8), B=9.9×104B = 9.9\times10^4 rad/s (values used as given; they correspond to band edges of about ω1≈1000\omega_1 \approx 1000 and ω2≈105\omega_2 \approx 10^5 rad/s, since ω1ω2=108\omega_1\omega_2 = 10^8 and ω2−ω1=9.9×104\omega_2 - \omega_1 = 9.9\times10^4).

Prototype (Figure 1): R1=1 ΩR_1 = 1\ \Omega, series L1=2.024L_1 = 2.024 H, shunt C1=0.994C_1 = 0.994 F, series L2=2.024L_2 = 2.024 H, R2=1 ΩR_2 = 1\ \Omega.

Element rules: series LL → series L/BL/B and B/(Lω02)B/(L\omega_0^2); shunt CC → parallel C/BC/B and B/(Cω02)B/(C\omega_0^2).

L1=2.024:L1′=2.0249.9×104=20.444 μH,C1′=9.9×1042.024×108=489.13 μFC1=0.994:C′=0.9949.9×104=10.040 μF,L′=9.9×1040.994×108=995.98 μHL2=2.024:L2′=20.444 μH,C2′=489.13 μF\begin{aligned} L_1 = 2.024:&\quad L_1' = \frac{2.024}{9.9\times10^4} = 20.444\ \mu\text{H},\quad C_1' = \frac{9.9\times10^4}{2.024\times10^8} = 489.13\ \mu\text{F} \\ C_1 = 0.994:&\quad C' = \frac{0.994}{9.9\times10^4} = 10.040\ \mu\text{F},\quad L' = \frac{9.9\times10^4}{0.994\times10^8} = 995.98\ \mu\text{H} \\ L_2 = 2.024:&\quad L_2' = 20.444\ \mu\text{H},\quad C_2' = 489.13\ \mu\text{F} \end{aligned}
  Rs=1   [ Z1 ]          [ Z3 ]
 o-/\/\--[    ]---+------[    ]---+-----o
 |                |               |     +
 Vs            [ Y2 ]           RL=1    Vo
 |                |               |     -
 o----------------+---------------+-----o
ArmBand-pass element
Z1 (series)20.444 μH in series with 489.13 μF
Y2 (shunt)10.040 μF ∥ 995.98 μH
Z3 (series)20.444 μH in series with 489.13 μF

Check: 1/20.444 μH×489.13 μF=1041/\sqrt{20.444\ \mu\text{H}\times489.13\ \mu\text{F}} = 10^4 rad/s. These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2071 Chaitra · 2+3 marks

What is the importance of frequency transformation in filter design? The circuit given in figure below is a lowpass filter having passband frequency of 1 rad/s. Obtain a band pass filter having ωo = 2000 rad/s and B = 400 rad/s. [Figure: source V1, 1 Ω source resistor, shunt 2.0237 F, series 0.9941 H, shunt 2.0237 F, 1 Ω load (output V2)]

Answer

Importance of frequency transformation

  • Only one set of approximation tables (low-pass poles and ladder element values) is needed; HP, BP and BS filters are derived from it.
  • The difficult approximation step (finding T(s)T(s) for given αmax\alpha_{max}, αmin\alpha_{min}) is done once, for the simpler low-pass case; specifications of the other types are converted to an equivalent LP specification.
  • It can be applied either to the transfer function (substitute for ss) or directly to each element of an LC ladder, so the circuit is obtained without re-deriving it.
  • It reduces design time and errors, and keeps the properties of the prototype (equiripple, maximally flat, etc.).

Band-pass filter from the given low-pass filter

LP → BP: s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs} with ω0=2000\omega_0 = 2000 rad/s (ω02=4×106\omega_0^2 = 4\times10^6), B=400B = 400 rad/s.

  • shunt CC → parallel C′=C/BC' = C/B, L′=B/(Cω02)L' = B/(C\omega_0^2);
  • series LL → series L′=L/BL' = L/B, C′=B/(Lω02)C' = B/(L\omega_0^2).

Given ladder: shunt C1=2.0237C_1 = 2.0237 F, series L2=0.9941L_2 = 0.9941 H, shunt C3=2.0237C_3 = 2.0237 F, 1 Ω terminations.

C1:C1′=2.0237400=5.0593 mF,L1′=4002.0237×4×106=49.414 μHL2:L2′=0.9941400=2.4853 mH,C2′=4000.9941×4×106=100.59 μFC3:C3′=5.0593 mF,L3′=49.414 μH\begin{aligned} C_1:&\quad C_1' = \frac{2.0237}{400} = 5.0593\ \text{mF},\quad L_1' = \frac{400}{2.0237\times4\times10^6} = 49.414\ \mu\text{H} \\ L_2:&\quad L_2' = \frac{0.9941}{400} = 2.4853\ \text{mH},\quad C_2' = \frac{400}{0.9941\times4\times10^6} = 100.59\ \mu\text{F} \\ C_3:&\quad C_3' = 5.0593\ \text{mF},\quad L_3' = 49.414\ \mu\text{H} \end{aligned}
  Rs=1         [ Z2 ]
 o-/\/\--+-----[    ]-----+------+---o
 |       |                |      |   +
 Vs   [ Y1 ]           [ Y3 ]  RL=1  Vo
 |       |                |      |   -
 o-------+----------------+------+---o
ArmBand-pass element
Y1 (shunt)5.0593 mF ∥ 49.414 μH
Z2 (series)2.4853 mH in series with 100.59 μF
Y3 (shunt)5.0593 mF ∥ 49.414 μH

Each LC pair resonates at 2000 rad/s. These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2080 Bhadra · 4 marks

Following circuit is a low pass filter having αp = 1dB and ωp = 1 rad/s. Obtain a bandpass filter ω0 = 400 rad/sec and bandwidth of 150 rad/sec. [Figure: source V1, 1 Ω source resistor, series 1.2817 H, shunt 1.9093 F, series 1.4126 H, shunt 1.0495 F, 1 Ω load]

Answer

Use the low-pass to band-pass transformation s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs} with ω0=400\omega_0 = 400 rad/s (ω02=1.6×105\omega_0^2 = 1.6\times10^5) and B=150B = 150 rad/s. The prototype has αp=1\alpha_p = 1 dB at ωp=1\omega_p = 1 rad/s, so the band-pass filter will have 1 dB ripple between band edges ω1,2=4002+752∓75=331.97\omega_{1,2} = \sqrt{400^2+75^2} \mp 75 = 331.97 and 481.97481.97 rad/s.

Element rules:

  • series LL → series L′=L/BL' = L/B and C′=B/(Lω02)C' = B/(L\omega_0^2);
  • shunt CC → parallel C′=C/BC' = C/B and L′=B/(Cω02)L' = B/(C\omega_0^2).

Given ladder: series 1.2817 H, shunt 1.9093 F, series 1.4126 H, shunt 1.0495 F, 1 Ω terminations.

1.2817 H:L′=1.2817150=8.5447 mH,C′=1501.2817×1.6×105=731.45 μF1.9093 F:C′=1.9093150=12.729 mF,L′=1501.9093×1.6×105=491.02 μH1.4126 H:L′=1.4126150=9.4173 mH,C′=1501.4126×1.6×105=663.67 μF1.0495 F:C′=1.0495150=6.9967 mF,L′=1501.0495×1.6×105=893.28 μH\begin{aligned} 1.2817\ \text{H}:&\quad L' = \frac{1.2817}{150} = 8.5447\ \text{mH},\quad C' = \frac{150}{1.2817\times1.6\times10^5} = 731.45\ \mu\text{F} \\ 1.9093\ \text{F}:&\quad C' = \frac{1.9093}{150} = 12.729\ \text{mF},\quad L' = \frac{150}{1.9093\times1.6\times10^5} = 491.02\ \mu\text{H} \\ 1.4126\ \text{H}:&\quad L' = \frac{1.4126}{150} = 9.4173\ \text{mH},\quad C' = \frac{150}{1.4126\times1.6\times10^5} = 663.67\ \mu\text{F} \\ 1.0495\ \text{F}:&\quad C' = \frac{1.0495}{150} = 6.9967\ \text{mF},\quad L' = \frac{150}{1.0495\times1.6\times10^5} = 893.28\ \mu\text{H} \end{aligned}
  Rs=1  [ Z1 ]        [ Z3 ]
 o-/\/\-[    ]--+-----[    ]--+------+---o
 |              |             |      |   +
 Vs          [ Y2 ]        [ Y4 ]  RL=1  Vo
 |              |             |      |   -
 o--------------+-------------+------+---o
ArmBand-pass element
Z1 (series)8.5447 mH in series with 731.45 μF
Y2 (shunt)12.729 mF ∥ 491.02 μH
Z3 (series)9.4173 mH in series with 663.67 μF
Y4 (shunt)6.9967 mF ∥ 893.28 μH

Each pair resonates at 400 rad/s (e.g. 1/8.5447 mH×731.45 μF=4001/\sqrt{8.5447\text{ mH}\times731.45\ \mu\text{F}} = 400). These values are for 1 Ω1\ \Omega terminations. For practical values, magnitude-scale by kmk_m (e.g. km=1000k_m = 1000 for 1 kΩ terminations): multiply every L by kmk_m and divide every C by kmk_m.

  • 2078 Bhadra · 1+1+5 marks

What is frequency transformation? What are the importance? Obtain a band pass filter having band center (ω0) at 1K rad/sec and bandwidth of 100 rad/sec from fourth order Butterworth lowpass ladder circuit. [Refer Table 2]

Answer

Frequency transformation is the change of the complex-frequency variable ss in a normalized low-pass prototype (cut-off ω=1\omega = 1 rad/s) so that the same circuit or transfer function gives a high-pass, band-pass or band-stop response, or a low-pass response at another frequency. Typical substitutions are s→s/ωps \to s/\omega_p (LP→LP), s→ωp/ss \to \omega_p/s (LP→HP) and s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs} (LP→BP).

Importance

  • Only one set of design tables (Butterworth, Chebyshev, Bessel low-pass prototypes) is needed; every other filter type is derived from it.
  • The approximation is done once, on the simple low-pass problem; the transformed filter keeps the same ripple, selectivity and order.
  • Element values can be converted directly in the circuit (each LL and CC is replaced by a simple combination), so no new synthesis is needed.
  • Normalized values (ω=1\omega = 1, 1 Ω1\ \Omega) are easy to handle and are later scaled to practical values.

Band-pass design

Prototype (Table 2, 4th-order Butterworth, Rs=RL=1 ΩR_s = R_L = 1\ \Omega, ωc=1\omega_c = 1 rad/s): g1=0.7654g_1 = 0.7654, g2=1.848g_2 = 1.848, g3=1.848g_3 = 1.848, g4=0.7654g_4 = 0.7654. Taking the ladder that starts with a series inductor: L1=0.7654L_1 = 0.7654 H (series), C2=1.848C_2 = 1.848 F (shunt), L3=1.848L_3 = 1.848 H (series), C4=0.7654C_4 = 0.7654 F (shunt).

Given ω0=1000\omega_0 = 1000 rad/s and B=100B = 100 rad/s, so ω02=106\omega_0^2 = 10^6.

LP → BP rules (substitute s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs}):

Prototype elementBecomesValues
Series LLseries L′L'–C′C' (series resonant)L′=L/BL' = L/B, C′=B/(Lω02)C' = B/(L\omega_0^2)
Shunt CCshunt L′′∥C′′L'' \parallel C'' (parallel resonant)C′′=C/BC'' = C/B, L′′=B/(Cω02)L'' = B/(C\omega_0^2)
ResistorunchangedRR

Each new pair resonates at ω0\omega_0 since L′C′=L′′C′′=1/ω02L'C' = L''C'' = 1/\omega_0^2.

Calculations

L1′=0.7654100=7.654 mH,C1′=1000.7654×106=130.65 μFC2′′=1.848100=18.48 mF,L2′′=1001.848×106=54.11 μHL3′=1.848100=18.48 mH,C3′=1001.848×106=54.11 μFC4′′=0.7654100=7.654 mF,L4′′=1000.7654×106=130.65 μH\begin{aligned} L_1' &= \frac{0.7654}{100} = 7.654\ \text{mH}, & C_1' &= \frac{100}{0.7654\times10^6} = 130.65\ \mu\text{F} \\ C_2'' &= \frac{1.848}{100} = 18.48\ \text{mF}, & L_2'' &= \frac{100}{1.848\times10^6} = 54.11\ \mu\text{H} \\ L_3' &= \frac{1.848}{100} = 18.48\ \text{mH}, & C_3' &= \frac{100}{1.848\times10^6} = 54.11\ \mu\text{F} \\ C_4'' &= \frac{0.7654}{100} = 7.654\ \text{mF}, & L_4'' &= \frac{100}{0.7654\times10^6} = 130.65\ \mu\text{H} \end{aligned}
ArmTypeLLCC
1series L–C7.654 mH130.65 µF
2shunt L ∥ C54.11 µH18.48 mF
3series L–C18.48 mH54.11 µF
4shunt L ∥ C130.65 µH7.654 mF

Terminations stay Rs=RL=1 ΩR_s = R_L = 1\ \Omega.

o-[Rs]-[L1 C1]-+-[L3 C3]-+-----+
               |         |     |
          [L2||C2]  [L4||C4]  [RL]
               |         |     |
o--------------+---------+-----+

Check: L1′C1′=7.654×10−3×130.65×10−6=10−6=1/ω02L_1'C_1' = 7.654\times10^{-3}\times130.65\times10^{-6} = 10^{-6} = 1/\omega_0^2. The resulting filter is 8th order, centred at 1000 rad/s with a 3 dB bandwidth of 100 rad/s. For practical values, the circuit may finally be impedance scaled (e.g. to 1 kΩ: multiply each LL by 1000, divide each CC by 1000).

Answer: series arms 7.654 mH + 130.65 µF and 18.48 mH + 54.11 µF; shunt arms 54.11 µH ∥ 18.48 mF and 130.65 µH ∥ 7.654 mF, with 1 Ω terminations.

  • 2076 Asoj · 1+1+3 marks

What is frequency transformation? What are its importance. The low pass filter shown below has a cutoff frequency at 1 rad/sec. Transform it into a band pass filter having center frequency at 10000 rad/sec and bandwidth of 1000 rad/sec. [Figure: 1 Ω source resistor, shunt 1 F capacitor, series 2 H inductor, shunt 1 F capacitor, 1 Ω load]

Answer

Frequency transformation is the change of the complex-frequency variable ss in a normalized low-pass prototype (cut-off ω=1\omega = 1 rad/s) so that the same circuit or transfer function gives a high-pass, band-pass or band-stop response, or a low-pass response at another frequency. Typical substitutions are s→s/ωps \to s/\omega_p (LP→LP), s→ωp/ss \to \omega_p/s (LP→HP) and s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs} (LP→BP).

Importance

  • Only one set of design tables (Butterworth, Chebyshev, Bessel low-pass prototypes) is needed; every other filter type is derived from it.
  • The approximation is done once, on the simple low-pass problem; the transformed filter keeps the same ripple, selectivity and order.
  • Element values can be converted directly in the circuit (each LL and CC is replaced by a simple combination), so no new synthesis is needed.
  • Normalized values (ω=1\omega = 1, 1 Ω1\ \Omega) are easy to handle and are later scaled to practical values.

Band-pass transformation of the given ladder

Prototype (ωc=1\omega_c = 1 rad/s): Rs=1 ΩR_s = 1\ \Omega, shunt C1=1C_1 = 1 F, series L2=2L_2 = 2 H, shunt C3=1C_3 = 1 F, RL=1 ΩR_L = 1\ \Omega (3rd-order Butterworth). Given ω0=104\omega_0 = 10^4 rad/s, B=103B = 10^3 rad/s, ω02=108\omega_0^2 = 10^8.

LP → BP rules (substitute s→s2+ω02Bss \to \dfrac{s^2+\omega_0^2}{Bs}):

Prototype elementBecomesValues
Series LLseries L′L'–C′C' (series resonant)L′=L/BL' = L/B, C′=B/(Lω02)C' = B/(L\omega_0^2)
Shunt CCshunt L′′∥C′′L'' \parallel C'' (parallel resonant)C′′=C/BC'' = C/B, L′′=B/(Cω02)L'' = B/(C\omega_0^2)
ResistorunchangedRR

Each new pair resonates at ω0\omega_0 since L′C′=L′′C′′=1/ω02L'C' = L''C'' = 1/\omega_0^2.

C1′′=11000=1 mF,L1′′=10001×108=10 μHL2′=21000=2 mH,C2′=10002×108=5 μFC3′′=1 mF,L3′′=10 μH\begin{aligned} C_1'' &= \frac{1}{1000} = 1\ \text{mF}, & L_1'' &= \frac{1000}{1\times10^8} = 10\ \mu\text{H} \\ L_2' &= \frac{2}{1000} = 2\ \text{mH}, & C_2' &= \frac{1000}{2\times10^8} = 5\ \mu\text{F} \\ C_3'' &= 1\ \text{mF}, & L_3'' &= 10\ \mu\text{H} \end{aligned}
o-[1 Ω]-+--[2 mH]-[5 uF]--+------+
        |                 |      |
  [10uH || 1mF]     [10uH || 1mF] [1 Ω]
        |                 |      |
o-------+-----------------+------+

Check: 2×10−3×5×10−6=10−8=1/ω022\times10^{-3}\times5\times10^{-6} = 10^{-8} = 1/\omega_0^2 and 10×10−6×10−3=10−810\times10^{-6}\times10^{-3} = 10^{-8}, so every arm resonates at 10410^4 rad/s.

Answer: shunt arms: 10 µH ∥ 1 mF (both ends); series arm: 2 mH in series with 5 µF; source and load 1 Ω.

  • 2073 Shrawan · 4 marks

The following low pass filter has passband frequency ωp of 1 rad/s. Transform it into a highpass filter having passband frequency of 2KHz. [Figure: source Vs, series 3 H inductor, shunt 2/3 F capacitor, series 1 H inductor, 2 Ω load resistor]

Answer

For LP → HP we substitute s→ωp/ss \to \omega_p/s. Then a series inductor LL (impedance sLsL) becomes ωpL/s\omega_p L/s, i.e. a series capacitor, and a shunt capacitor becomes a shunt inductor. Resistors are unchanged.

Cnew=1L ωp,Lnew=1C ωpC_{new} = \frac{1}{L\,\omega_p}, \qquad L_{new} = \frac{1}{C\,\omega_p}

Given: prototype ωp=1\omega_p = 1 rad/s with series L1=3L_1 = 3 H, shunt C2=2/3C_2 = 2/3 F, series L3=1L_3 = 1 H, RL=2 ΩR_L = 2\ \Omega. New passband edge fp=2f_p = 2 kHz:

ωp=2π×2000=12566.37 rad/s\omega_p = 2\pi\times2000 = 12566.37\ \text{rad/s}

Element values

C1=13×12566.37=26.53 μFL2=1(2/3)×12566.37=119.37 μHC3=11×12566.37=79.58 μFRL=2 Ω (unchanged)\begin{aligned} C_1 &= \frac{1}{3\times12566.37} = 26.53\ \mu\text{F} \\ L_2 &= \frac{1}{(2/3)\times12566.37} = 119.37\ \mu\text{H} \\ C_3 &= \frac{1}{1\times12566.37} = 79.58\ \mu\text{F} \\ R_L &= 2\ \Omega\ \text{(unchanged)} \end{aligned}
LP elementHP elementValue
series L1=3L_1 = 3 Hseries C1C_126.53 µF
shunt C2=2/3C_2 = 2/3 Fshunt L2L_2119.37 µH
series L3=1L_3 = 1 Hseries C3C_379.58 µF
RL=2 ΩR_L = 2\ \OmegaRLR_L2 Ω
o-(Vs)-[C1]-+-[C3]-+
            |      |
          [L2]   [RL]
            |      |
o-----------+------+

Answer: C1=26.53 μC_1 = 26.53\ \muF, L2=119.37 μL_2 = 119.37\ \muH, C3=79.58 μC_3 = 79.58\ \muF, RL=2 ΩR_L = 2\ \Omega; the circuit passes frequencies above 2 kHz. (These values can be impedance scaled later if a larger load resistance is needed.)

Questions from Old Question Collection (BEI EX 606 and BEX EX 704) (Scanned IOE papers: BEI EX 606 2078–2083 and BEX EX 704 2069–2076), Old Question Collection (EX 704) (IOE BEX EX 704 papers from 2069 to 2081) and 2080 course paper (ENEX 301) (IOE ENEX 301 new-course paper, 2082 Chaitra). Answers are written for this site; check them against your class notes.

Chapter titles and hours from the IOE syllabus ↗