Filter Desing Complete

Downloaded from www.jayaram.com.
np
Chapter:01
Filter: Filter can be considered can be considered as frequency selective networks. A filter is
required to separate an unwanted signal from a mixture of wanted and unwanted signals.
The filter specification are generally given in terms of cutoff frequencies, pass
band (P.B) and stop band (s.b) regions. P. B is the frequency band of wanted signal and S.B
is the frequency band of unwanted signal. An ideal filter should pass the wanted signal with
no attenuation and provide infinite attenuation.
Depending upon the components used, filters can be classified as:
1. passive filters: Filters which are the compotnet such as R,L,C are the passive filters. The
Gains of such filters are always less than or equal to unity (i.e GS1). It is to be noted the
L and C are filter components, but R is not.
2. Active filters: The filters which use the components such as transistors, op-amp etc are
the active filters. The Gains of such filters are always greater than or equal to unity. ( G ≥
1)
Gain and Attenuation:
i/p o/p
V1(t) Filter V2(t)
network
Let us consider the filters network with i/p V1(t) having power P1 and o/p V2(t) having
power p2 as shown in fig1. Then the transfer function is given by T(s) = V2(s)/V1(s)
Where , V1(s) and V2(s) are the Laplace Transform of V1(t) .
v 2 ( jw)
Also, T(s) = T(jw) =
v1 ( jw)
Then the voltage gain in db is given by ,
Av = 20log10 T ( jw) dB …………….(1)
Or in term of power , the power gain is given by,
p1
Ap = 10 log10
p2
Now, the voltage attenuation is given by ,
α = 1/Av
α = -20log T ( jw) dB…………….(2)
From equation 1 and 2 ,we can write,
1
Downloaded from www.jayaram.com.np
0.05Av………………..(3)
T ( jw) = 10
-0.05 α
T ( jw) = 10 …………….(4)
Types of filters: ( According to the function)

Filters are classified according to the functions they are to perform. The pattern
of PB and SB that give rise to the most common filters as defined below:
1. Low pass filters: (LPF): A LPF characteristics is one in which the PB extend from ω
= 0 to ω = ωc where ωc is know as cut off frequency.
A
PB SB
w
wc
Fig. 1(a)
2. High pass filter: A high pass filter is a compolement of a low pass filter in that the
frequency range form o to ωc is the SB and from ωc to infinity is the PB.
A
SB PB
wc w
Fig. 1(b)
3. Band pass filter ( BPF): A BPF is one in which the frequency extending form ωL (or
ω1) to ωu (ω2 ) are passed while signals at all other frequencies are stopped.
A
SB PB SB
w
wc
Fig. 1(c)
4. Band stop filter(BSF): A BSF is complement of BPF where signal components at
frequencies form ω1 to ω2 are stopped and all others are passed. These filters are
sometimes known as “Notch filters”.
A
PB SB PB
w
Fig. 1(d) Notch filter
5. All pass filters (APF): It is a filter which passes all range of frequencies , i.e , PB
2
ranges from o to infinity.
A
PB
w
Fig. 1(e)
Non- ideal Characteristics:

Filter Gain curve Attenuation curve
A A
1. LPF 1 TB
0.707
Wp Wc Ws Wp Wc Ws
A A
2.HPF
Ws WcWp
TB
1. From the attenuation curve it to be noted that in the pass band the attenuation is
always less then a maximum value. Designated as αmax
2. In the stop band the attenuation is always larger then a minimum value designated as α
min .
3. Band between PB and SB so defined are known as transition bands. (TB).

Bilinear Transfer function and its poles and zeroes:
We know,
T(s) = P(s)/Q(s) = N(s)/D(s)
a m s m + a m −1 s m −1 + ........... + a1 s + a 0
T(s) =
bn s n + bn −1 s n −1 + ............ + b1 s + b0
When , m = n = 1, then the T(s) of equation (i) will be bilinear , i.e

P( s) a1 s + a 0
T (s) = =
Q( s ) b1 s + b0
a1 ( s + a o / a1 )
=
b1 ( s + b0 / b1 )
G ( s − z1 )  G ( s + z1 ) 
= or T ( s ) = 
( s − p1 )  (s + z 2 ) 
3
If z1< p1 If p1< z1
Here, G = a1/b1 = Gain
Z = -a0/a1 = a zero
P1 = -b0/b1 = a pole
Date:2065/4/22
Realisation of filter with passive elements:

Let us now see how the bilinear transfer function and its various special cases can be
realized with passive elements.
+
v1 + c
-
-
Fig 1.
Plot the magnitude and phase response of the ckt shown in fig (1) and identify the filter.
Solution:
Applying kirchoff’s law for fig 1
1
L∫
V1 = R1 + idt................(i )
1
L∫
V2 = idt......................(ii )
Taking laplace transform of equation (i) and (ii)

1
V1 ( s ) = RI ( s ) + I ( s )..................(iii )
cs
1
I (s)
V2 (s) cs
V2 (s) = =
V1 ( s )  1
I ( s) R + 
 cs 
4
1
1
= cs =
Rcs + 1 RC ( s + 1 / RC )
cs
1
= RC
S + 1 / RC
W0
T (s) =
S + W0
Where, W0 = 1/RC
Now , for magnitude plot,
T(s) = T(jw) = W0/(jw+W0)
w0
∴ T ( jw) =
w 2 + wo
2
Now when
W = 0 T ( jw) = 1
W= wo T ( jw) = 0.707
W = ∞ , T ( jw) = 0
T(jw)
1
0.707
W
Wc
Fig. 2. Magnitude plot
For phase plot:
θ (jw) = tan-1(o/w0) – tan-1(w/wo)

θ (jw) = tan-1(w/w0)
When,
W = 0 , θ (j0) = 0
W = wo , θ (jwo) = -45 ˚
W = ∞ , θ (j ∞ ) = - 90 ˚
5
90
45
Wo
W
-45
-90
2.
R1
1 R1 2
+
v1 + c
- R1 v2
3 -
4 R
Above figure can be modified as:

1 1
3
+ 2
v1
-
4 4
-
From figure the potential of node 2, is V1/2 and the potential at node 3 is Vs R/(1+1/cs)
∴ V2 = V1/2 - Vs R/(1+1/cs)
V1/V2 = ½ - RCS/RCS+1
T(s) = R(S+1- 2RCS)/2(RCS+1) = -{(RCS+1)/2(RCS+1)}
= RC(S+1/RC)/2RC(s+1/RC)
Where Wo = 1/RC
T(jw) = -1/2 {(jw-wo)/(jw+wo)}
For magnitude plot ,
1 w 2 + ( w0 ) 2
T ( jw) =
2 w 2 + w0
2
6
T(jw)
1 W
T ( jw) =
2
Phase plot:
θ (jw) = tan-1(-w/wo) - tan-1( w/wo)

θ (jw) = -2tan-1 (w/wo)
when,
w = 0, θ (jw) = 0
w = 0, θ (jw) = -90 ˚
w= ∞ , θ (jw) = -180 ˚
90
45
Wo
W
-45
-90
-135
-180
From the magnitude plot, we see that the networking is all pass filter.
Assignment:
3.
+
v1 + v2
-
4.
7
c1
R1
+ R2 v2
v1
-
5.
R1
+
R2
v1 + v2
-
C2
-
Date: 2065/4/28
Example :04
c1
R1
+ R2 v2
v1
-
From fig (i)

R1C1 S + 1
Y1 = c1s+1/R1 =
R1
R1
Z1= 1/Y1 =
R1C1 S + 1
Now applying kirchoff’s voltage law, for fig (i).
V1 = z1i+R2i
V1(s) = (z1s+R2)I1(s)
And ,
V2(s) = R2I(s)
V2 ( s) R2 R2
T(s) = = =
V1 ( s ) Z 1 ( s ) + R2 R1
+ R2
R1C1 S + 1
1
R1 R 2 C1 ( S + )
R2 ( R1C1 S + 1) R1C1
= =
R1 + R2 R1C1 S + R2  R + R2 
R1 R2 C1  S + 1 
 R2 R1C1 
8
1
S+
R1C1
=
1 1
S+ +
R 2 C1 R1C1
S + ω 01 S − (−ω 01 )
Or, T(s) = =
S + ω 02 S − (−ω 02 )
ω > ω 01 -wo1 wo2

And , 02
or , −ω 02 < ω 01
For Magnitude plot:
jw + w01 w 2 + w01
2
T(jw) = =
jw + w02 w 2 + w02
2
w01 R2
Now at w= 0, T ( j 0) = =
w02 R1 + R2
w01
At w = ∞, T ( j∞ ) = =1
w02
T(jw)
1
R2
R1 +R2
W
w=0
For Phase plot,

jw + w01
T(jw) =
jw + w02
Where, w01= 1/R1C1
W02 = 1/R1C1+1/R2C2
 w   w 
Therefore, θ(jw) = tan −1   − tan −1  
 w01   w02 
θ (jw) = θz – θp
Since direct phase plot of above expression is very complicated, we will go it by indirect
method. First we will plot the zero phase and then the pole phase and finally find the net
pole – zero phase.
Zero plot (θz)
9
 w 
θ ( z ) = tan −1   = tan −1 ( wR1C1 )
w
 01 
Now at w = 0
θ (z) = θ (j0) = 0
θ (z) = θ (jw0)= 45 ˚
Now at w = ∞
θ (j ∞ ) = 90 ˚
Pole plot (θp)

θ (p) = tan-1(w/w01)
 
 
−1  w 
= tan  
1 1
 + 
 R1C1 R2 C1 
Now at, w = 0
θp = θ (j0) = 0
at w = w02
θp = θ(w02) = 45 ˚
at w = ∞ , θp = θ (j ∞ ) = 90 ˚
(jw)
zero plot
90
45 pole zero plot
wo2
W
w=0 wo1
-45
Pole plot
-90
Thus the magnitude response of the above network shown that it is a high pass filter with dc
gain R2/(R1+R2) and phase plot signifies it is leading type.
Insertion Gain and insertion loss:
10
T(jw)
1 1
w w
wo wo
w
wo
Insertion gain
T(jw)
1
1
w
wo
Insertion gain
One of the important factor that should be consider in design is that the minimum value of α
should be zero degree. But this is not true in practical case since we are using active element
, this need not be the case because the active element may provided the gain greater than
one (1). If it is necessary to meet the specification exactly then it will be necessary to provide
ck t to reduce the gain. We call this unwanted gain as the insertion gain. On the other hand
there is a loss in the components of passive filter so it provides access attenuation and we
call this loss as insertion loss. To overcome this problem additional compensation circuit is
required.
Chapter- 2
Normalization and Renormalization:

In most of the cases we consider the values of R, L S& C to be the order of unity. It is very
difficult to built the capacitor of 1 f and inductor of 1 H . Besides this the practical values of
capacitors available in the electronic circuit is of the order of microfarad or Pico farad. The
circuit considered so for have normalized elemental values but practically these values are
not realizable. So we perform scaling to get the realizable components.
There are mainly two reasons for resorting the normalized design.
1. Numerical computation become simple and it is easier to manipulate the numbers of
the order of unity.
2. If we have the normalized design of the filter then it is easy to generate the
11
filter of similar characteristics of varying center frequency and impedance level
without redesigning the whole circuit.
The actual or the required elemental values of the
Filter ckt which is obtained after scaling is called demoralized values of the circuit.
Scaling: While designing the ckt sometimes the value of components may not be available
so we change them with the available one, which is called scaling. To obtained the elemental
values of the required filter we amplitude and frequency scale the normalized design.
Types of scaling:
1. Impedance (Magnitude or amplitude) scaling: In this scaling, the magnitude of the

impedance is increased or decreased. To scale in magnitude , z(s) (the impedance) is
multiplied by a constant factor Km .
Z(s)
If Km . 1, then it is called scale up.

If Km< 1, then it is called scale down.
Let, Rold = old value of Resistor.

Lold = old value of inductor
Cold = old value of capacitor.
The new values of R, L and C are given by
Rnew = Km Rold ………….(i)

Also,
XLKm = Lold S Km = (KmLold)S = LnewS
Lnew = Km Lold ……………….(ii)
Again,
1 1
Xc Km = 1/cold s . Km = =
 C old  C new .S
  s
 Km 
C 0ld
C new = ……………………….(iii)
Km
Example 01:
Perform Impedance scaling to the following network.
12
R=1 ohm
+ +
V2 C =1F V1
- -
Solution:
Rold = 1 Ω
Cold = 1 F
Now , let us assume that,
Cnew = 10 µF
Note: Generally we assume new value of capacitor 1µF or 10 µF.

We know that
Cnew = Cold/Km
Km = Cold/Cnew = 1F/10 µF = 105
Therefore, Rnew = Km .Rold
= 105 * 1 Ω
Rnew = 100K
R=100k
+ +
V2 C =10 uf V1
- -
Fig(ii) scaled ckt.
The transfer function for fig. (i) ,

Told(s) = 1/(s+1)
1
R new C new
And, Tnew = = 1/s+1
1
S+
R new C new
Thus we see that there is no change in the following transfer function while doing magnitude
scaling.
Date: 2065/5/3
2. Frequency scaling:
In frequency scaling our objective is to scale the frequency without affecting the
magnitude of the impedance , i.e
ZL = ( = XL) = LS = jWL
Z L = WL is a constant.
Similarly,
Zc ( = Xc) = 1/cs = 1/ jwc
13
1
Zc = is constant.
wc
To do so any change in w must be compensated by corresponding change in L and c
If, w = old corner frequency
Ω = new corner frequency.
Ω = Kf w
Where,
Kf = frequency scaling factor.
If Kf > 1, then it is called expansion scaling
If, Kf < 1 , then it is called compression scaling.
ωo
Expansion
ωo = 10 Ωo = 10
3
Compression
3
ωo = 10 Ωo = 1
Also, if T(jw) is old Transfer function, then the new transfer fucnti is T(j Ω)
= T (jKfw)
The resistance is unaffected by frequency scaling , i.e

Rnew= Rold …………….(v)
For inductor,
Xl = Ls = jwL = jwkf . L/kf
Or, XL = j(wkf) ( Lold/kf) since, L = Lold
= j Ω ( Lold/kf)
Lold = Lold / Kf ………….(vi)

For capacitor,
Cnew = Cold/ kf…………………………(vii)
14
3. Both magnitude and Frequency scaling:
It is not necessary that we scale magnitude and scale in frequency separately. We can do
both at once. Cobining all the above equations.
Rnew = KmRold …………….(Viii)

Lnew = Km/kf. Lold ……………(ix)
Cnew = Cold/Km.kf ………….(x)
These three equations are know as element scaling equations.
Example 01:
1Ω
1F
Solution:
W0 = 1 , Ω = 1000
Therefore, kf = Ωo/wo = 1000
Now we know that
Cnew = Cold /kf = 1F/ 1000 = 1 mF
And , Rnew = Rold = 1 Ω
1k
1mF
Fig (ii): after frequency scaling.
Now,
1
R0ld C old 1
Told(S) = =
1 s +1
s+
Rold C old
1
R new C new 10
And, Tnew(s) = =
1 s + 10
s+
R new C new
Example 02:
R=1/10Ω
1Ω
1F
15
Perform frequency scaling with Ωo = 1 Ω
Example 03:
R1
C1
v1 R2 v2
T(s) = (s+0.5)/(s+3)
Perform magnitude and frequency scaling separately with wo = 3 and Ω0 = 300.
Solution:
The transfer function of the above figure is
1
s+
R1C1
T(s) = …………….(i)
1 1
s+ +
R1C1 R 2C1
But given ,
T(s) = (s+0.5)/(s+3) ……………….(ii)
Comparing equation (i) and (ii)
1/R1C1 = 0.5
R1C1 = 2 ……………..(iii)
Again, ( 1/R1+ 1/R2)1/C1 = 3…………..(iv)
Let , C1= 1 F
For equation (iii) R1× 1 = 2
R1 = 2 Ω
Therefore from equation (iv)
(1/2 + 1/R2) 1/2 = 3
Therefore, R2 = 2/5 Ω
In order to perform magnitude scaling
R1old = 2 Ω
R2old = 2/5 Ω = 0.4 Ω
Cold = 1 F
Say, C1new = 10 µF
Then, Km = Cold/Cnew = 1F/ 10 µF
Km = 105
Therefore, Rnew = km R2old = 105 × 0.4 Ω = 40 k
The selected ckt will be :
16
200k
40.4F
v1 40k v2
Again for frequency scaling,

Wo = 3 , Ω0 = 3000
Therefore , kf = Ωo/ wo = 3000/ 3 = 1000
Therefore, R1new = R1old = 2 Ω
R2new = R2old = 0.4 Ω
C1old = C1old/kf = 1F/ 1000 = 1 mF.
Example 04:
C2= 1/10 F R2= 1/100Ω
_
R1= 1Ω C1= 1 F
+
Perform magnitude scaling to the ckt given.

Note: Take ‘Cnew’ as the new value of capacitor for ‘Cold’ where ‘Cold’ represents the largest
value in the circuit.
Solution:
Here, R1old = 1 Ω
R2old = 2 Ω
C1old = 1 F
C2old = 1/10 F.
Take, Cnew = 10 µF.
Then for, magnitude scaling,
Cnew = Cold/km
Km = C1old/ C1new = 1F/ 10 µF = 105
Therefore, C2new = C2old/km = 0.1 F/ 105
C2new = 1 µF
Similarly,
R1old = km. R1old = 105× 1 Ω = 100 k
R2new = km. R2old = (1/100). 105 Ω = 1 k.
1k
_
100k 10 uF
+
17
Fig: Magnitude Scaling Ckt.
18
Chapter: 3
One port and two port passive network:
Positive real function: The filter circuit is complex transfer function that may be realizable
depending upon weather the transfer function exhibits PRF properties. I the transfer function
is PRF only ckt is realizable. There are two types of passive network : [i] one port network
[ii] Two port network.
I(s) I1(s) I2(s)
V(s) 1- port V1(s) 2- port

n/w n/w
Fig. 1(a) one port n/w Fig. 2(b) two port n/w
One port network: Let us suppose of fig of 1(a),

Then, z(s) = V(s) / I(s)
If V(s) = 3s+2
I(s) = 1
Then, z(s) = 3s+2
= Ls +R
3H
V(s) 2Ω
Thus , the function is realization but if, z(s) = 3s-2 , then it is not realizable.
Date: 2065/ 5/10

Why? (छटे
ु को छ)
(i) If F(s) denote the function in S-domain, the F(s) indicates either driving point
impedance or driving admittance. Which ever is concern to us.
(ii) F(s) should be for real value of S.
(iii) The value of F(s) must be greater than or equal to zero. i.e Re[f(s)] ≥ 0.
Thus in brief a PRF must be real and +ve .
If F(s) = LS = jWL L must be +ve.
F(s) = 1/CS = 1/jwc C must be +ve
F(s) = R R must be +ve.
Properties of Passive n/w.

A passive network is one
(i) The element of which one are +ve and real.
19
(ii) The average Power dissipated (APD) by the n/w. for a sinusoidal i/p must be +ve.
For one port n/w APD = 1/2 Re[ z(s)][I(s)]2 ≥ 0
Properties of PRF:
1. If F(s) is +ve and real , then 1/F(s) is also +ve and real.
2. The sum of DRFS is always PRF but the difference may not be PRF.
Example: Z1(s) = 5s+ 3 (PRF)
Z2(s) = 2s+ 5 ( PRF)
Then, z1(s)+z2(s) = 7s+8 (PRF)
But, Z1(s) – Z2(s) = 3s-2 (not PRF)
3. The Poles and zero’s of PRF cannot be in the right half of the S-Plain.
4. Only poles with real residues can exists on the jw axis.
Example: F(s) = 6s/(s2+ σ 2 )
In this case, S = ± σ j
Residue = σ real and +ve.
5. The poles and zeroes of PRF Occurs in pairs.
6. The highest power of numerator and denominator polynomial may differ atmost by
unity.
S 5 + 4 S 4 + 3S 3 + 3S 2 + 3S 1 + 2
Example:
S 6 + 4 S 4 + 2 S 3 + 3S 2 + 3 K
7. The lowest power of numerator and denominator polynomial may differ atmost by
unity.
S 5 + 4 S 4 + 3S 3 + 3S 2 + 3S
Example:
S 6 + 4 S 4 + 2 S 3 + 3S 2 + 3 K
8. The real part of F(s) must be greater than or equal to zero. i.e Re[F(s)] ≥ 0
But , if Re[F(s)] = 0 , then the ckt do not consist resistive components. Hence only
capacitive and inductive components are presents. Hence only capacitive and inductive
components are present . Such a n/w whose transfer function satisfies this condition is
known as lossless n/w.
Example: Determine weather the function is PRF.
(i) z(s) = 2s2+5/s(s2+1)
Hence , z(s) = 2s2+5/s(s+1)
A/s + Bs/(s2+1) = A/s + B/ (s2+1)/s
2s 2 + 5
A= .s
s ( s + 1) s = 0
2 s 2 + 5 ( s 2 + 1)
B= .
s ( s 2 + 1) s s2
2(−1) + 5
= = −3
(−1)
Z(s) = 5/3 + -3s/(s2+1)
Here, (-3) , the residues ( s2 = -1) is –ve , therefore z(s) is not PRF.
( s + 1)( s + 4) s ( s + 4) + 2( s + 4)
(ii) z(s) = =
( s + 1)( s + 3) s ( s + 3) + 1( s + 3)
20
s 2 + 6s + 8
=
s 2 + 4s + 3
2s + 5
=1+
( s + 1)( s + 3)
= z1 ( s ) + z 2 ( s )
2s + 5 A B 3 / 2 1/ 2
Where z2(s) = = + = +
( s + 1)( s + 3) s + 1 S + 3 s + 1 s + 3
3 / 2 1/ 2
Therefore, z(s) = 1 + +
s +1 s + 3
It is not PRF.
8 s 3 + 4 s 2 + 3s + 1
(iii) z(s) =
8 s 3 + 3s
s 2 + 2s + 8
(iv) Y(s) =
s ( s + 4)
Basic ckt Synthesis Techniques:
Any one port n/w each can be represented by either admittance function Y(s) or
impedance function z(s) . i.e
a n s n + a n −1 s n −1 + a n − 2 s n − 2 + ............. + a1 s + a 0
F ( s) =
bm s m + bm −1 s m −1 + bm − 2 s m − 2 + ............ + b1 s + b0
P(s)
=
Q( s)
N ( s)
=
D( s)
Z ( s)
=
P( s)
Design of LC Ckt . (Loss less ckt):

Consider a impedance function as
E n ( s) + On ( s)
Z(s) =
E m ( s) + Om ( s)
Where En(s) and Om(s) denote the even parts of numerator and denominator respectively
and On(s) and On(s) denote odd part.
s5 + s4 + s3 + s + 1 N (s)
Z(s) = =
s + s + s + s + s + s + 1 Q( s )
6 5 4 3 2
( s 4 + s 2 + 1) ( s 5 + s 3 + 5)
+
E n ( s) On ( s)
=
( s 6 + s 4 + s 2 + 1) ( s 5 + s 3 + 1)
+
E m ( s) O m ( s)
For the loss less function , it is to be noted that,
Re[z(s)] = 0 ………….(i)
E n ( s) + On ( s) E m ( s) − Om ( s)
Now, z(s) = ×
E m ( s) + Om ( s) E m ( s) − Om ( s)
E n ( s ) E m ( s ) + On ( s ) E m ( s ) − E n ( s )Om ( s ) − On ( s )Om ( s )
=
E m ( s) − Om ( s)
2 2
21
E n ( s ) E m ( s ) + On ( s ).Om ( s ) O n ( s ) E m ( s ) − E n ( s )O m ( s )
= +
E m ( s ) + Om ( s ) E m ( s) − Om ( s)
2 2 2 2
E n ( s ) E m ( s ) − On ( s ).Om ( s )
= Re[ z(s)] = …………(ii)
E m ( s) − Om ( s)
2 2
Therefore from equation (i) and (ii).

E n ( s ) E m ( s ) − On ( s ).Om ( s )
=0
E m ( s) − Om ( s)
2 2
E n ( s ) E m ( s ) − On ( s).Om ( s ) = 0
E n ( s ) E m ( s ) = On ( s).Om ( s )
E m ( s) On ( s)
= ………….(iii)
Om ( s ) E m ( s )
The above equation (iii) indicates that LC ckt is even to odd ( or odd ) to even function.
Properties of LC Ckt:
a n s n + a n − 2 s n − 2 + a n − 2 s n − 4 + ............. + a 0
1. F ( s) =
bm s m + bm − 2 s m − 2 + bm − 4 s m − 4 + ............ + b0
The coefficients an and bm must be real and +ve and F(s) must be even to odd or odd to even
function.
2. The highest power of numerator and denominator can differ atmost by unity ( in this
case it is 2). So does the lowest power.
3. The succeeding power of ‘s ‘ in numerator and denominator must differ by the order of
s 4 + 17 s 2 + 165s 0
2 all the way through . Example:
s 3 + 4s
4. The poles and zeros must be alternatively placed on the jw axis and lie only on the
imaginary axis.
5. There must be either a pole or a zero at the origin.
Example: Test whether the following function is LC.
(i) z(s) = K (s2+1)(s2+5)/(s2+2)(s2+10) k>0
It is not LC ckt function because,
1. There is neither pole or zero at the origin though the pole zero are alternatively placed on
the imaginary axis.
2. It is not even to odd or odd to even function.
(ii) Z(s) = z(s2+1)(s2+9)/s(s2+4)

(iii) Z(s) = k s(s2+4)/(s2+1)(s2+3) , k> 0
(iv) Z(s) = s5+4s3+5/(4s4+s2)
Date: 2065/5/12
Design of LC ckt by Foster’s Method:
In this case ,
k0 2k s
F(s) = + 2 i 2 + ............ + k ∞ s ……………….(i)
S s + wi
This equation may represent z(s) or Y(s)
Case I : ( i.e when F(s) = z(s))
22
Then,
k0 2k s
Z(s) = + 2 i 2 + ............ + k ∞ s
S s + wi
Here,
- ko/s will represent a capacitive reactance of 1/ko F.
- 2ki(s)/(s2+w2) will represent LC parallel combination.
Having capacitor of value 1/2ki F and inductor of value 2ki/wi2. Thus the final circuit will
be:
2
2ki/wi
1/k k∞
0
.......
z(s) 1/2ki
This method of circuit synthesis is known as foster impedance or series or 1st method for LC
ckt.
Case – II
In this case , F(s) = Y(s) , then equation (i) becomes

k0 2k s
Y(s) = + 2 i 2 + ............ + k ∞ s
S s + wi
Here,
- KO/s represents admittance of inductor having value of 1/ko H.
- K ∞s represent admittance of capacitor having value K ∞ F.
- 2ki(s)/s2+w2 represents admittance of series LC combination having inductor of value
1/2ki H and capacitor value wi2/2ki
The ckt can be realize as :
1/2ki
k
1/k ι ∞
wi 2 /2ki
This method of circuit synthesis is known as foster admittance or parallel or 2nd method for
LC ckt.
Example 01: Design a Foster series n/w for the following n/w.
s ( s 2 + 4)
F ( s) =
2( s 2 + 1)( s 2 + 9)
Solution:
It is Foster’s series n/w
23
s ( s 2 + 4)
F ( s) = z ( s) =
2( s 2 + 1)( s 2 + 9)
s ( s 2 + 4) As Bs
Now, z ( s) = 2 = 2 + 2
2( s + 1)( s + 9) s + 1 s + 9
2
s ( s 2 + 4) ( s 2 + 9)
Where, A = .
2( s 2 + 1)( s 2 + 9) s s 2 = −1
−1+ 4
= = 3 / 16
2( − 1 + 9)
Therefore, A = 3/16
s ( s 2 + 4) ( s 2 + 9)
And B = .
2( s 2 + 1)( s 2 + 9) s s 2 = −9
−9+4 −5 5
= = =
2(−9 + 1) 2 × −8 16
Therefore, B = 5/16
(3 / 16) s (5 / 6) s
z (s) = + 2 = z1 ( s ) + z 2 ( s )
s2 +1 s +9
The ckt will be as follows.
L 1= 3/16 H L 2 = 5/144 H
z(s) C1 =16/3 F C2 = 16/5 F
• The first part of z(s) ( i.e z1(s) ) represents parallel LC combination having inductor L1
of value 3/16 H and capacitor of value 16/3 F.
• The 2nd part of z(s) (i.e z2(s) ) represents parallel LC combination having inductor L2 of
value 5/144 H and capacitor C2 of value 16/5 F.
Example 02: Design Foster parallel n/w for the function

s ( s 2 + 4)
F ( s) =
2( s 2 + 1)( s 2 + 9)
Solution:
It is Foster’s parallel n/w
s ( s 2 + 4)
F ( s) = Y ( s) =
2( s 2 + 1)( s 2 + 9)
s ( s 2 + 4) As Bs
Now, z ( s) = 2 = 2 + 2
2( s + 1)( s + 9) s + 1 s + 9
2
s ( s 2 + 4) ( s 2 + 9)
Where, A = .
2( s 2 + 1)( s 2 + 9) s s 2 = −1
24
−1+ 4
= = 3 / 16
2( − 1 + 9)
Therefore, A = 3/16
s ( s 2 + 4) ( s 2 + 9)
And B = .
2( s 2 + 1)( s 2 + 9) s s 2 = −9
−9+4 −5 5
= = =
2(−9 + 1) 2 × −8 16
Therefore, B = 5/16
(3 / 16) s (5 / 6) s
Y (s) = + 2 = Y1 ( s ) + Y2 ( s )
s2 +1 s +9
The ckt will be as follows:
Figure:
• The first part of Y(s) ( i.e Y1(s) ) represents series LC combination having inductor L1 of
value 16/3 H and capacitor of value 16/3 F.
• The 2nd part of Y(s) (i.e Y2(s) ) represents series LC combination having inductor L2 of
value 16/5 H and capacitor C2 of value 144/5 F.
2( s 2 + 1)( s 2 + 9)
Example 03: Design Foster parallel n/w for the function F ( s) =
s ( s 2 + 4)
Solution:
It is Foster Parallel ,
2( s 2 + 1)( s 2 + 9)
F ( s) = Y ( s) =
s ( s 2 + 4)
2 s 4 + 20 s 2 + 18
=
s 3 + 4s
S3+4s )2s4+20s2+18( 2s
2s4+8s2
12s2+18
12 s 2 + 18
Therefore, Y(s) = 2s + 3
s + 4s
12 s 2 + 18
= 2s +
s ( s 2 + 4)
Y(s) = Y1(s) + Y2(s)
12 s 2 + 18 A Bs 9 / 2 (15 / 2) s
Now Y2(s) = 2s + = + 2 = + 2
s ( s + 4) s s + 4
2
s s +4
9 / 2 (15 / 2) s
Y(s) = 2s + + 2 = Y1 ( s ) + Y2 ( s ) + Y3 ( s )
s s +4
Here Y1(s) = 2s , so C1 = 2 F
25
9/2
Y2 (s) = , So, L1 = 2/9 H
s
(15 / 2).s
And Y3(s) = 2
s +4
L2 = 2/15 H
C2 = 8/15 F
Therefore, The final ckt will be
L 2 =2/15 H
L 1= 2/9H
C1 = 2 F
C2 = 8/15 F
Fig. Foster’s parallel n/w of LC ckt.
Assignment:
2( s 2 + 1)( s 2 + 9)
1. z(s) =
s ( s 2 + 1)
2( s 2 + 2)( s 2 + 4)
2. Y(s) =
( s 2 + 3)( s 2 + 1)
Date: 2065/5/17
Continued Fraction method or cauer method for LC Ckt

1. case- I
It is removed by successive removal of pole at ∞. The ckt will be as follows:
L1 L2
....
C1 C2 Cn
Fig. For F(s) = z(s)
L1 L2
....
C1 C2 C3 Cn
V(s)
Fig. For F(s) = Y(s)
Example 01: Synthesis the following function in cauer form.
26
2 s 5 + 12 s 3 + 16 s
Z(s) =
s 4 + 4s 2 + 3
Solution:
In cauer n/w we proceed as follows:
S4 +4s2+3) 2s5 + 12s3 + 16s (2s z1(s)
2s5+8s3+6s
4s3+10s) s4 + 4s2 +3 (s/4 Y2(s)
S4 + 10s2/4
3S2/2+3) 4S3 +10S (8s/3 Z3(s)
3
4S +8S
2S) 3S2/2 +3 (3s/4 Y4(s)
2
3s /2
3) 2s (2s/3 Z5(s)
2s
2H 8/3 2/3
1/4 3/4
Z(s)
Fig. Cauer n/w for LC series ckt
2 s 5 + 12 s 3 + 16 s
Example: 02: Y(s) =
s 4 + 4s 2 + 3
2F 8/3 F 2/3 F
Y(s)
Fig: Cauer n/w for LC parallel ckt.
Example:03: Synthesis the following ckt in cauer form.

s ( s 2 + 2)( s 2 + 4) s ( s 2 + 2)( s 2 + 4)
(i) Y(s) = (ii) Z(s) =
( s 2 + 1)( s 2 + 3) ( s 4 + 1)( s 2 + 3)
Cauer –II:
This is the case of removal of pole at origin.
C1 C2
....
L1 L2 Ln
Fig. Caure II n/w for LC series ckt.

27
Example:01: Synthesize the following function in cauer form.
s 4 + 4s 2 + 3
Z(s) =
2 s 5 + 12 s3 + 16 s
Solution:
Since Z(s) is the case of pole at origin (i.e s = 0 ) z(s) can be rewrite as:
3 + 4s 2 + s 4
Z(s) =
16 s + 12 s 3 + 2 s 5
16s+12s3+2s5 ) 3+4s2+ s4 (3/16s z1(s)
3+9s2/4+ 3s4/8
7s2/4+5s4/8)16s+12s3+2s5 (64/7s Y2(s)
16s+40s3/7
44s3/7+2s5) 7s2/4+5s4/8 (49/176s Z3(s)
2 4
7s /4+44s /88
3s4/44) 44s3/7 +2s5 ((44)2/21s Y4(s)
3
44s /7
2s5) 3s2/44(3/88s Z5(s)
3s2/44
16/3 176/40 88/3
Z(s)
7/64 21/44.44
Fig. Cauer II n/w for LC ckt
s 4 + 4s 2 + 3
Example:02: Y(s) = 5
2 s + 12 s 3 + 16 s
7/64 21/1936
Y(s) 176/49 88/3

16/3
Fig. Caure II n/w for parallel LC ckt.
R-C one port n/w: (R-C impedance /R-L admittance)

1. Foster 1st method:
In this case,
F(s) = z(s) , gives R-C impedance n/w.
28
1/k k∞
0
.......
z(s) 1/k1
Foster method defines F(s) as

F(s) = z(s) = ko/s + k1/(s+σ1)+ k2/(s+σ2) + ………+k∞
Here,
- ko/s represent capacitive reactance having capacitor of value 1/ko F.
- k∞ represent resistor of value k∞ Ω .
- ki/(s+σi) represents RC parallel in which the resister has a value of ki/σi Ω and a
capacitor has value of 1/ki F.
Properties of RC impedance N/w:

1. the poles of RC –impedance n/w are on the –ve real axis.
2. As in LC ckt, residues of poles (kis) are real and +ve i. z(s ) must be PRF.
3. At two critical frequencies i.e when s = o , i.e σ = 0 when s = ∞ i.e σ = ∞
4. z(0) = ∞ if C0 is present
= ∑Ri , if C0 is missing
5. z(∞ ) = k∞ , R∞ is present
= 0, R∞ is missing
6. z(0) ≥ z(∞) is always true.
7. The critical frequency nearest to the origin must be a pole.
8. The poles and zeroes must be alternatively placed.
Example:01 State giving reasons which of the following if not RC impedance.

( s + 1)( s + 4)( s + 9)
(a) Z(s) =
s ( s + 2)( s + 5)
( s + 1)( s + 8)
(b) Z(s) =
( s + 2)( s + 4)
( s + 2)( s + 4)
(c) Z(s) =
( s + 1)
( s + 1)( s + 2)
(d) Z(s) =
s ( s + 3)
6( s + 2)( s + 4)
Example:02: Synthesis the following function in Foster series form: F(s) =
s ( s + 3)
Solution:
6( s + 2)( s + 4)
Since it is foster series function z(s) =
s ( s + 3)
This is the RC impedance n/w.
Now,
(i) z(0) = ∞ , C0 is present .
(ii) z(∞ ) = σ , R∞ is also present.
29
Z(s) = ko /s + k∞ + k1/(s+3) = ko /s + k1/(s+3)+6
6( s + 2)( s + 4)
Ko = .s = (6.2.4)/3 = 16
s ( s + 3) s=0
K2 = 2
∴ Z(s) = 16/s + 2/(s+3) + 6
The component values are as follows:
16/s ⇒ 1/cos ⇒ c0 = 1/16 F
σ ⇒ R∞ ⇒ R ∞ = 6 Ω
2/(s+3) ⇒ R1 = 2/3 Ω and C1 = ½ F
The ckt will be:

2/3
1/16 6
.......
z(s) 1/2
Date: 2065/5/19
6( s + 2)( s + 4)
F(s) =
s ( s + 3)
6( s + 2)( s + 4)
F(s) = z(s) =
s ( s + 3)
= 6+ 16/s + 2/(s+3)
Forster parallel method for R-C one port n/w:
In this case,
F(s) = Y(s)
Y(s) = ko/s + k1/(s+σ1)+ k2/(s+σ2) + ………+k∞
... ...
R1 R2
Lo
L1 L2
Fig. (i) R-L admittance n/w for foster 2nd method in this case
- ko/s represents inductor of value 1/ko

- k∞ represents inductor of value 1/ko
- ki/(s+σi) represents RL series ckt having inductor of value 1/ki H and resister of
value σi/k Ω .
30
Properties:
Same as RC- impedance.
Example: 01: Synthesis the following function in foster parallel.

6( s + 2)( s + 4)
F(s) =
s ( s + 3)
Solution:
Since it is Foster parallel,
6( s + 2)( s + 4)
F(s) = Y(s) =
s ( s + 3)
= 6 + 16/s + 2/(s+3)
∴ The ckt will be:
2/3
1/16
1/6
1/2
Fig. R-L admittance ckt from foster parallel
Continued Fraction method or cauer method for R-C impedance or R-L Admittance:
1. If F(s) = z(s) , then it yields cauer 1 n/w.
2. If F(s) = Y(s) , then it yields cauer 2 n/w.
For cauer 1 n/w:

In this case F(s) = z(s)
Example:01: Synthesize the following function cauer 1 form.

6( s + 2)( s + 4)
F(s) =
s ( s + 3)
Solution:
6( s + 2)( s + 4) 6 s 2 + 36 s + 48
F(s) = z(s) = =
s ( s + 3) s 2 + 3s
Now,
S2 +3s)6S2+36s+48(6 Z1(s)
S2+18s
18s+48) s2+ 3s (s/18 Y2(s)
2
S +8s/3
s/3) 18s+ 4s (54 Z3(s)
18s
48) s/3 (s/3.48 Y4(s)
The ckt will be: s/3
6 54
1/18 1/144
Fig. Caure 1 n/w

31
Cauer 2 n/w:
6( s + 2)( s + 4)
Example: 02: Realise the given function in cauer 2 n/w F(s) =
s ( s + 3)
Solution:
In this case,
6( s + 2)( s + 4)
F(s) = Y(s) =
s ( s + 3)
In this case circuit will be :
1/18 1/144
1/54
1/6
Fig. Caure 2 method
R-L one-Port n/w: (R-L impedance or R-C admittance n/w)
1. Foster Series method: It yields R-L impedance ckt for which

F(s) = (s) = ko + kis/(s+ σ1) + k2s/(s+σ2) + ……..+ k∞s
k1 k2
ko k∞
....
k1/ σ1 k2/ σ2
z(s)
In this case,
- k0 represent resistor of value ko Ω .
- k∞ s represent inductor of value k∞ H.
- kis/(s+σi) represent RL parallel ckt with resistor of value ki and inductor of value
ki/σi .
This method of synthesis is know as foster series (1st) method for R-L one port n/w.
Properties of R-L impedance n/w:

1. Poles are on the –ve real axis.
2. The residue of pole must be real and +ve i.e F(s) must be PRF.
3. z(0) = k0 if R0 is present.
= 0 if R0 is missing.
4. z(∞ ) = ∞ if L is present.
= ∑Ri if L is missing.
5. z(∞ ) ≥ z(0)
6. Zero is nearest to the origin.
7. The pole and zero must be alternatively placed.
32
2. Foster parallel method:

In this case,
F(s) = Y(s) = ko +k1s/(s+σi) + k2s/(s+σ2) + ……….+ k∞
The ckt will be as follows:
1/k 1/k
1 2 k
∞
Y(s) 1/k 0
k2/σ2
k1/σ1
This method of synthesis is known as Foster parallel method which yields R-C admittance
n/w.
Properties:
Some as that of R-L impedance except F(s) = Y(s)
4( s + 1)( s + 3)
Example:01: Given F(s) = . Realise the above function in (a) Foster series
( s + 2)( s + 6)
(b) Foster parallel.
Solution:
Since zero is nearest to the origin , (i.e s = -1f) the function yields R-L one port n/w.
4( s + 1)( s + 3)
(a) Foster series: In this case F(s) = z(s) =
( s + 2)( s + 6)
Thus, it yields R-L impedance n/w. To check the availability of components, we use.
Z(0) = (4×1×3)/(2×6 ) = 1 = ko . i.e Ro is present .
Z(∞ ) = 4 = ∑Ri , L∞ is missing.
4( s + 1)( s + 3) 1 k k
∴ z (s) /s = = + 1 + 2
( s + 2)( s + 6) s s+2 s+6
4( s + 1)( s + 3)
K1 = .( s + 2)
s ( s + 2)( s + 6) s = −2
4(−2 + 1)(−2 + 3)
=
− 2( −2 + 6)
=½
4( s + 1)( s + 3)
K2 = .( s + 6)
s ( s + 2)( s + 6) s = −6
4(−6 + 1)(−6 + 3)
=
− 6( −6 + 2)
K2 = 5/2
1 (1 / 2).s (5 / 2).s
∴ z(s)/s = + +
s s+2 s+6
33
1/2 5/2
1
1/4 5/12
z(s)
Fig. Foster series n/w
(b) Foster parallel:

In this case,
4( s + 1)( s + 3)
F(s) = Y(s) =
( s + 2)( s + 6)
Which yields R-C admittance n/w.
(1 / 2).s (5 / 2).s
∴ Y(s) = 1 + +
s+2 s+6
2 2/ 5
Y(s) 1
1 12 /5
Fig. Foster Parallel ckt.
Cauer Method for R-L one port n/w:

(1) If F(s) = z(s) , it is called cauer 1 method which yields R-L impedance ckt.
(2) If F(s) = Y(s) , it is called caure 2 method which yields R-C admittance ckt.
Example: 01: Synthesize the following function in
4 s 2 + 16 s + 12
(a) caure 1 n/w. (b) cauer 2 n/w.
s 2 + 8s + 12
Solution:
(a) cauer 1 n/w:
In this case
4( s + 1)( s + 3) 4 s 2 + 16 s + 12
F(s) = z(s) = = 2
( s + 2)( s + 6) s + 8s + 12
S2+8s+12 ) 4s2+ 16s+12 ( 4

4s2+ 32s+4s
-ve
This way the ckt cannot be realize. Therefore z(s) is rewritten in form as:
12 + 16 s + 4 s 2
Z(s) =
12 + 8s + s 2
34
12+8s +s2 ) 12+16s+4s2 ( 1 Z1(s)

2
12+ 8s+s
8s +3s2 ) 12+8s+s2 ( 3/2s Y2(s)
12+9s/2
7s/2 +s2 ) 8s+3s2 ( 16/7 Z3(s)
8s+16s2/7
5s2/7 ) 7s/2+s2 ( 49/10s Y4(s)
7s/2
s2 ) 5s2/7 ( 5/7 Z5(s)
5s2/7
1 16/7
2/3 10/49 5/7
Fig. cauer 1 n/w
(b) Cauer 2 n/w:

In this case,
4 s 2 + 16 s + 12 12 + 16 s + 4 s 2
F(s) = Y(s) = =
s 2 + 8s + 12 12 + 8s + s 2
1/2ki
k
1/k ι ∞
wi 2 /2ki
Fig. Cauer 2 n/w
Assignment: 03
( s + 2)( s + 4)
1. F(s) = Find the n/w of the form (a) Foster series (b) Foster parallel.
( s + 1)( s + 3)
( s + 1)( s + 3)
2. Realize the n/w function F(s) = (a) 1st Foster method. (b) 2nd foster
( s + 2)( s + 4)
method.
( s + 2)( s + 4)
3. Realise the n/w function Y(s) = as a cauer n/w.
( s + 1)( s + 3)
( s + 1)( s + 3)
4. z(s) = Realise the function in foster and cauer n/w.
( s + 2)( s + 2)
35
( s + 2)( s + 4)
5. Realise the n/w Y(s) =
( s + 1)( s + 6)
Two port n/w:
1. Z-Parameter
2. Y – Parameter
3. ABCD Parameter
4. Transformation of one parameter to other
5. T and π n/w
6. Interconnection of two port n/w
a. Cascade b. series c. parallel.
Date: 2065/5/24
Chapter: 4
Low pass Filter Approximations:
T(jw) T(jw)
1
1
PB SB
w W
wo=1 Wp Ws
Fig. (a) Ideal case (b) Non ideal case
The desirable feature of low pass approximation are

1. Minimum pass band attenuation, αp
2. Maximum stop band attenuation, αs
3. Low transition band ratio, ws/wp
4. Simple network.
The approximation Method are:
1. Butterworth
2. Chebyshev
3. Inverse chebyshev
4. Ellipse or Cauer
5. Bessel –Thomson
1. Butterworth low pass approximation: Generally signal become contaminated with
high frequency signal. It is evident that low pass filter are required to remove such
unwanted signals from the useful one. The desirable LPF response is shown in fig . 1(a)
Below the normalize frequency i.e w0 = 1, the amplitude T ( jw) is
constant and above this frequency it is zero. Pass band and stop band are clearly separated at
wo = 1. But since the ideal response can not be achieve . We make the approximation based
on the ideal response.
We make the magnitude T(jw) nearly constant in PB. In the SB, we require sharp
roll – off (n-pole roll –off). Where ‘n ‘ will be large no if abrupt transition from PB to SB is
desired.
Mathematically, we can write,
T(jw) = Re[ T(jw) ] + j Im [ T(jw)]
36
Re[T(jw)] = Real part of T(jw)
Im [T(jw)] = Imaginary part of T(jw).
Where it is to be noted that Re[T(jw)] indicates an even functions.
Where Im[T(jw)] indicates it is an odd function.
Agains,
T*(jw) = T(-jw) = Re[ T(jw)]+jIm [T(jw)] ……….(ii)The functions so obtained is called
conjugate of T(jw)
Thus (i) and (ii) gives
T(jw) T*(jw) = T ( jw) = Re[T(jw)]2 + jIm[T(jw)]2 ……(iii)
2
2
T(jw) T*(jw) = T(s) T*(s) = T (s)
2 2
The function T (s) (or T ( jw) ) is called magnitude squared function.
Example 01: Find the magnitude square function for

T(s) = (s+2) / (s3 + 2s2 + 2s+3)
T(s) = -s+2 / -s3 + 2s2 – 2s +3
2
∴ T (s) = T(s) . T(-s)
= (2+s)/(s3+2s2 +2s+3) × (2-s)/(-s3 +2s2 – 2s+3)
= ………..
The magnitude square function is an even function which can be represented by using a
numerator and denominator polynomial that are both even, i.e
A( w 2 )
T ( jw) =
2
B( w 2 )
A + A2 w 2 + A4 w 4 + ........... + A2 n w 2 n
T ( jw) = 0
2
B0 + B 2 w 2 + B4 w 4 + ........... + B 2 n w 2 n
A0
T ( jw) =
2
B0 + B2 w + B4 w 4 + ........... + B2 n w 2 n
2
Here A2 = A4 = ……A2n = 0 (assumption).

The choice has been made as per our inspection on the roll off that was directly dependent on
the number of poles. This means larger the difference between degree of A and B , we get
the larger roll-off . This will give us a direct n-pole roll off for Tn(jw) or Tn(s) which will be
know as “ All pole” function.
Special case:
We assume ,
B 2 = B4 = 0
B2n = (1/w0)2n. B0 and A0 = B0
Now , putting these assumption in the equation (i) we get,
Ao
T ( jw) =
2
B0 + B 2 n w 2 n
Bo
= 2n
 1 
B0 +   B0
 w0 
37
1
= 2n
 1 
1 +   w 2 n
 w0 
2 1
T ( jw) = 2n
………….(ii)
 w 
1 +  
 w0 
In generalize condition,
wo = 1
2 1
T ( jw) = ………….(iii)
1 + (w )
2n
2 1
T ( jw) = …………(iv)
1 + (w)
2n
From equation (iv) the following property can be written.

1. At w = 0 , i.e T(j0) = 1 for all values of n.
2. At w = 1 (=w0), i.e T(j1) = 0.707 for all values of n.
3. At w = ∞ , i.e T(j ∞ )= 0 for all value of n.
4. For large values of w; Tn(jw) exhibits larger roll off.
5. Butterworth response , also known as, maximally flat response, is all pole functions.
6. Butterworth (BU) response can be expanded in Taylor’s series from as:
2 1
T ( jw) =
1 + (w)
2n
= (1+w2n)-1/2
= 1+ ½ . w2n + (1/2)2. (w2n)2/2! - ………..
≈ 1 – ½. w2n
∴ In Taylor series,
 1 
T ( jw) = 1 − w 2 n  ………..(v)
 2 
Again we know ,
2 1
T ( jw) =
1 + (w )
2n
Putting jw = s
2 1 1 1 1
T (s ) = = = =
s
2n
s 2n
1+ s 2n
1 + (−1) n s 2 n
1 +   1 + 2n
 j j (−1) n
2 1
T (s ) = ………(vi)
1 + (−1) n s 2 n
Which gives the butterworth response in s-domain
Evaluation of T(s) for BU – Response:

(i) For n = 1 equation (vi) becomes
38
2 1
T (s ) =
1− s2
s2 =1
s = ±1
2 NOTE:
∴ T ( s) = 1/(1-s)(1+s)
(i) If sn = -1, then, s=1∠
= 1/(1+s). 1/(1-s) (180+k360)/n, k = 0, 1…..(n-
= T(s) . T(-s) 1) in s domain.
T(s) = 1/(s+1) (ii) If sn =-1, then, S = 1 ∠
k360/n, k =0,1, 2….(n-1)
Date: 2065/5/29
Butterworth transfer function (continued ……)

(ii) For n = 2
Equation (vi) becomes ;
1
T ( s) =
2
1 + (−1) 2 s 4
jw
45
135
225 315
1
=
1+ s4
To get the poles ,
1+s4 = 0
S4 = -1
S = 1 ∠ (180 ˚ + k360 ˚ )/4 , k = 0, 1, 2, 3 [since n = 4]
S = 1 ∠ 45 ˚ , 135˚ , 225˚ , 315˚

The poles that lie on the left half of s-plane are:
S = 1 ∠ 135˚ , 225˚
Or S = -0.0707 ± j0.707 = s1, s2
1
∴ T(s) =
( s − s1 )( s − s 2 )
1
=
( s + 0.707 − j 0.707)( s + 0.707 + j 0.707)
1
=
s 2 + 2s + 1
(iii) For n = 3
39
1
T ( s) =
1 + (−1) 3 s 6
1
=
1− s6
To get the pole
1-s6 = 0
S6 = 1
S = 1 ∠ k360/n , k = 0,1,2 ………(2n-1)
S = 1∠ 0, 60, 120, 180, 240, 300
The poles that lie on left half of s-plane are
S = 1 ∠ 120, 180, 240
Or, = 1∠120, 1∠ 180, 1∠ 240
S1 = -0.5 + j0.866
S2 = -1+j0
S3 = -0.5 - 0.866j
1
∴ T ( s) =
( s − s1 )( s − s 2 )( s − s 3 )
1
=
( s + 1)( s − 0.5 − 0.866 j )( s + 0.5 − 0.866 j )
1
=
( s + 1)( s 2 + s + 1)
jw
120 60
σ
180
240
320
Order and cutoff frequency for Butterworth:

It is to noted that, at w =wp , α = α p = α max
And at w = ws , α = α s = α min
We know that
1
T ( s) =
2
2n
 w 
1 +  
 wo 
Also the attenuation formula is given by ;
α = -20log T (s)
40
 
 
 1 
α = -20log10  2n 
1 +  w  

 w 
  o  
1
2n − 2
 w
α = -20log10 1 +  
 wo 
2n
 w 
α = 10log10 1 +   …………….(i)
 wo 
2n
 w 
α /10 = log10 1 +  
 wo 
2n
α /10  w 
10 = 1 +  
 wo 
2n
 w 
  = 10α /10 -1
 wo 
 w 
  = (10α /10 -1)1/2n
 wo 
w
w= 1
(10 α / 10 − 1) 2 n
Now at w = wp , α = α max
wp
wo = 1
………….(ii)
α max/ 10
(10 − 1) 2n
and at w = ws , α = α min
ws
wo = 1
………………..(iii)
α min/ 10
(10 − 1) 2n
equating (i) and (ii) can be equated as:

wp ws
1
= 1
α max/ 10 α min/ 10
(10 − 1) 2n
(10 − 1) 2n
1
wp (10 α max/ 10 − 1) 2n
= 1
wo
(10 α min/ 10 − 1) 2 n
2n
 wp  (10 α max/ 10 − 1)
  =
w  (10 α min/ 10 − 1)
 o 
Taking log on both sides,
2n
w  (10 α max/ 10 − 1)
20 log  p  = log
 wo  (10 α min/ 10 − 1)
(10 α max/ 10 − 1)  wp 
n = log / 2 log  
(10 α min/ 10 − 1) w 
 o
41
Now let us find expression for transition band ratio , i.e
TBR = ws/wp , where , TBR = Transition band ratio.
Ws/wp = [(10α min/10 – 1)/(10α max/10-1)]1/2n ………….(v)
Example 01: Consider a filter using a butterworth response to realize the following
specifications of LPF.
α max = 0.5 dB
α min = 20 dB
wp = 1000 rad/sec
ws = 2000 rad/sec
Determine the order and cut off frequency for the filter.
Solution:
n = 4.83 ≈ 5
wo = 1234.12 rad/sec
Note: Always choose higher value of ‘n’ ( i.e the order of filter )because it provides larger
roll off which decreases attenuation.
Date: 2065/6/2
2. Chebyshev Approximation Method For LPF :
T(jw) T(jw)
1 1
BU-R
C-R
W W
Wo Wo
Fig (i) (a) Chebyshev response (b) butterworth response
The generalize low pass filter can be represented by

2 1
Tn ( jw) = ……………….(i)
1 + [ Fn ( w)] 2
For Butterworth
Fn(w) = (w/wo)n
With w0 = 1
Fn (w) = wn
Similarly to butterworth we have to determine the function Fn(w) for chebyshev response
for which the concept of Lissagious figure is required.
Lissagious figure:
42
Horizontal plate
Line of axis
sin θ
Vertical plate
Fig (ii) (a) CRO Lissagious figure.

n=1
y
n=2
n=3
n=4
Fig(ii) (b) Lissagious figure for n = 1,2,3 and 4
When adjustable frequency multiple of fixed frequency is applied , stationary figures are
obtained which are know as Lissagious figures.
Analysis:
Let the deflection due to voltage on horizontal plates be
x = coskT …………….(ii)
Where , k = 2 π/T
The deflection due to voltage on vertical plates will be then,
y = cosnkT …………….(iii) Where n is integer and proves the multiple frequencies.
From (ii),
KT = cos-1x
y = cosn cos-1x ……………..(iv)
cn(x) = cosn cos-1x which is the equation for Lissagious figures.
Example: If n = 4
Assume, θ = cos-1x
x = cosθ
Then,
y = cos4θ
θ x 4θ y
0 1 0 1
22.5 0.924 90 0
45 0.707 180 -1
67.5 0.383 270 0
90 0 360 1
43
y
0.383
x=1
-1
0.707
# Analyse the same for n = 3 and 5.
Chebyshev magnitude Response:

We know that ,
2 1
Tn ( jw) =
1 + [ Fn ( w)] 2
Where Fn (w) = ∈ cn(w) ; ∈ ≤ 1
Where cn(w) = cosn cos-1w
Therefore the magnitude square response will be
2 1
Tn ( jw) = …………….(vi)
1+ ∈2 c n ( w)
2
This function (i.e cn(w)) is valid within the range w = ± 1. However , the function must
also be valid for longer value of w for which we should refine our assumption for cn(w).
∴ w > 1,
Let,
Cos-1(w) = jz
w = cosjz
we know that ,
e j ( jz ) + e − j ( jz ) e z + e −z
cosjz = = = coshz
2 2
∴ cosjz = coshz
∴ w = coshz
Z = cosh-1w
∴ w = cosj cosh-1w
∴ cos-1(w) = jcosh-1w
∴ cn (w) = cosn cos-1w

= cosnj cosh-1w
= cosj(ncosh-1w)
= coshn cosh-1w
∴ cn(w) = cosh cosh-1w , w> 1

Cn(w) = cosn cos-1w, w = ± 1
44
Properties of magnitude response for Chebyshev:

We know that,
2 1
Tn ( jw) =
1+ ∈2 c n ( w)
2
1
Tn ( jw) =
1+ ∈2 c n ( w)
2
Where, cn(w) = cosn cos-1w w≤1

= coshn cosh-1w w≥1 and ∈ ≤ 1
1. At w = 0,
Cn(0) = cosn π/2 ; 0,1,2…….
Tn ( jw) = 1 for n = odd
1
= for n = even
1+ ∈2
2. w = 1
cn (1) = 1 for all values of n.
1
∴ Tn ( jw) =
1+ ∈2
1 1
w w
w=1 w=1
Fig (iii) (a) C-R for n = odd (b) C-R for n = even
Order of C-R filter:

We know , the attenuation formula is given by
α = -20log Tn ( jw) dB
1
1  1 2
But, Tn ( jw) = =  
 1+ ∈2 c 2 ( w) 
1+ ∈ c n ( w)
2
2
 n 
1
 1 2
∴ α = - 20log  

+ ∈2 2
 1 c n ( w) 
1
= -10log
1+ ∈2 c n ( w)
2
α = 10 log 1+ ∈2 c n 2 ( w) …………(vii)
∴ α = 10log 1+ ∈2 (cos n cos −1 w) 2 w ≤1
for w > 1,
α = 10 log 1+ ∈2 (cosh n cosh −1 w) 2 ............(ix)
Now ,
45
α max occurs when , cn(w) = 1
∴ equation (vii) reduces to ,

α = α max = 10 log ( 1+ ∈2 .1)………(x)
αmax / 10 = log ( 1+ ∈2 .1)
1+ ∈2 = 10α max/10
1
α max/ 10
∈= (10 − 1) 2
…………….(xi)
Date: 2065/6/7
Here
we know that
w = wnp , then, ∈2 c n 2 ( w) = 1
1
= cosh(n cosh −1 wnp )
c n ( wnp ) = [since wnp >1]
∈
1
Cosh-1(ncosh-1whp) =
∈
1
Cosh-1(ncosh-1whp) =
∈
-1 -1 1
Cosh whp = 1/n. cosh ( )
∈
1
∴ wnp = cosh(1/n. cosh-1( )) ………(xii)
∈
Wnp = cosh [1/n. cosh-1({10αmax/10 -1}1/2)]
Now α = αmin when w = ws
∴ αmin = 10 log10(1+ ∈2 c n 2 ( ws )
α min/10
∈2 c n ( ws ) = 10
2
–1
ε (cosh ncos ws) = 10α min/10-1
2 -1 2
Or, ( cosh ncosh-1ws)2 = (10α min/10-1)/ (10α max/10-1)

n cosh-1ws = cosh-1 [(10α min/10-1)/ (10α max/10-1)]1/2
∴ n = {cosh-1 [(10α min/10-1)/ (10α max/10-1)]1/2}/cosh-1ws …..(xiii)
Example: Given wp = 1 , ws = 2.33 , αmax = 0.5dB , αmin = 22 dB. Calculate ‘n’ for
Butterworth and chebyshev filters which filter would you select.
Solution: For Butterworth filter , the order is given by
n = log10[(10α max/10-1)/(10α min/10 -1)]/ 2 log (wp/ws)
= log[(100. 5/10 -1)/(1022/10 -1)]/2log (1/2.33)
= 4.234 ≈ 5
∴ n for BU = 5
For Chebyshev the order is given by ,
n = cosh-1[(10αmin/10-1)/(10αmax/10-1)]/cosh-1(2.33)
= 2.89 ≈ 3
46
n for chebyshev = 3 .
Since the order of chebyshev filter (i.e n =3) is less then the order of butterworth filter (i.e n
= 5) and both filter provides the same roll- off for the specification, n would choose
chebyshev filter.
Chebyshev poles location and network function:

We know
2 1
T ( jw) = ……..(i)
1+ ∈ c n ( w)
2 2
Substituting s = jw equation (i) becomes,

2 1
T (s ) = ………..(ii)
1+ ∈2 c n ( s / j )
2
To determine the poles,

1+ ∈2 c n ( s / j ) = 0
2
1
cn (s / j) = ± j ……(iii)
∈
Again,
Cn (s/j) = cosn cos-1(s/j)
Let
Cos-1(s/j) = x = u + jv
Then, cn(s/j) = cosnx = cosn (u+jv)
= cosnu. Cosnjv – sinnu. Sin njv
= cosnu coshnv – jsin nu . sinh nv
1
=0 ±j [ from equ. (iii)]
∈
Thus, comparing , we get, [ cosjnv = coshnv
Cosnu . cosh nv = 0 [ sinjnv = jsinhv]
-sinnu. Sinhnv = 0
∴ The minimum value of
Coshnv = 1, coshnv not equal to 0
∴ cosnu = 0
Or cosnuk = cos(2k+1). Π /2, k = 0,1,2………….
Uk = (2k+1) π/2n ……….(v)
Now ,
1
-sinnuk = sinhnvk = ±
∈
But, sin nuk = +- 1
1
∴ +-1 . sinhnvk = ±
∈
1
Or sinhnvk =
∈
-1 1
Nvk= sinh ( )
∈
1
Vk = 1/n. sinh-1( )
∈
47
Again, we know that
Cos-1 (s/j) = x = u +jv
s/j = cosx = cos(u+jv)
in general,
sk = jcos(uk +jv)
= j[cosuk .cosjv – sinuk. sinjv]
= j[cosuk . coshv – jsinuk. sinhv ]

Sk = sinuk. sinhv + jcosuk. coshv ……(vi) , k = 0,1,2….(2n-1)
Again,
Sk = sin[(2k +1) π /2n] sinhv + jcos[(2k+1)π /2n] coshv
Or , sk = σk + jwk ………..(viii)
Where,
Σk = sin[(2k+1) π /2n] sinhv ……….(ix)
Wk = cos[(2k+1)π /2n] coshv ………..(x)
Form euation (ix) ,
σ k2/sin2 hv = sin2[ ( 2k+1)π /2n] ……..(xi)
and from quation (x)
w2k/cos2hv = cos2 [ (2k+1)π /2n] ……..(xii)
Now adding equation (xi) and (xii) we get,
σ k2/sin2 hv+ w2k/cos2hv = 1 ……….(xii)
Which is equation of ellipse . Therefore we can say that the poles of chebyshev filter lie on
the ellipse.
Date: 2065/6/9
Example:01 Obtained the 4th order network function of a low pass chebyshev filter with α
max = 0.75 dB
Solution: n = 4 αmax = 0.75 dB
Now ε = ( 10αmax/10 -1)1/2 whp = cosh (1/n. cosh-1 (1/ε))
= (100.75/10 -1)1/2 = 0.434
And whp = cosh ( 1/n. cosh-1(1/ε)) =
Pole location is given by

48
Sk = sinuk sinhv + jcosuk coshv
Where, uk = (2k+1) .π/2n ; k = 0, 1, ………2n-1
V = 1/n. sinh-1(1/ε)
∴ uo = π/8 u1 = 3π/8 , u2 = 5π/8, u3 = 7π/8 , u4 = 9π/8
u5 = 11π/8 u6 = 13π/8 , u7 = 15π//8
v = 0.393 (adjust calculator in radian)
s0 = 0.154 + 0.996j
s1 = 0.373+ 0.413j
s2 = 0.373 – 0.413j
s3 = 0.154-0.996j
s4 = -0.154 – 0.996j
s5 = -0.373 – 0.413j
s6 = -0.373 + 0.413j
s7 = -0.154 + 0.996j
The transfer function (or n/w function) for forth order chebyshev filter is given by ,
T(s) = 1/(s+s4)(s+s5)(s+s6)(s+s7)
jw
S7 S0
S1
S6
S2
S5
S4 S3
Home Assignment:
Example:02: Determine the network function for 3rd order chebyshev LPF with αmax = 0.75
dB ( =αp ; pass band attenuation)
Date: 2065/6/14
Inverse chebyshev low pass approximation:
49
T (jw) T (jw)
W W
Ideal LPF BU-Response
T (jw) T (jw)
W W
Chebyshev-respone inverse-Chebyshev-response
2
1- T (jw)
W
Fig: intermediate stage to obtain inverse chebyshev response.
2
Tic(jw)
Fig: The reciprocal value of w of intermediate stage give the value of w in I-C response.
We know the response of chebyshev is given by

2 1
T ( jw) =
1+ ∈2 c n ( w)
2
2 1
1- Tc ( jw) = 1-
1+ ∈2 c n ( w)
2
∈2 c n ( w)
2
=
1+ ∈2 c n ( w)
2
Now replace w by 1/w
50
∈2 c n (1 / w)
2
2
TIC ( jw) = ………………(i)
1+ ∈2 c n (1 / w)
2
Where,
2
TIC ( jw) is the magnitude square response for I-C.
We know ,
cn(1/w) = cosncos-1(1/w)
at for w = 1
cn(1) = 1 for all value of n
Thus equation (i) becomes
∈2 .1
2
TIC ( j.1) =
1+ ∈2
∈2 .1
TIC ( j.1) = ……. (ii)
1+ ∈2
We know that ,
α min = -20log TIC ( j.1) dB ……(iii)
Using equation (ii) on equation (iii) , we get,

α = α min = -20log TIC ( j.1) dB
1/ 2
 ∈2 
= - 20log  
 1+ ∈
2

 1+ ∈2 
= 10log  2 
 ∈ 
1
∴ αmin = 10 log [ 1+ }
∈2
1
Or , 10αmin/10 -1 =
∈2
( )
1
α min/ 10 −
∈= 10 − 10
…………. (iv) 2
Again in general, the attenuation formula can be written as:

 ∈2 c n 2 (1 / w) 
α = -10log  
1+ ∈ c n (1 / w) 
2 2
  1
α = 10 log 1 + 
∈2 c n (1 / w) 
2

Now at w = wp α = αmax
Then above equation becomes
 1 
α = αmax = 10 log 1 + 
∈2 c n (1 / w p ) 
2

1
(10α max/10 – 1) =
∈2 c n (1 / w p )
2
2 1 1
c n (1 / w p ) = . α max/ 10
∈ (10
2
− 1)
51
2 (10 α min/ 10 − 1)
c n (1 / w p ) =
(10 α max/ 10 − 1)
(10 α min/ 10 − 1)
c n (1 / w p ) = ……….(v)
(10 α max/ 10 − 1)
c n (1 / w p ) = coshn cosh-1 (1/wp) …………. (vi)
[∴ wp < 1, 1/wp > 1]

Thus equating equation (v) and (vii)
1
α min/ 10
 (10 − 1)  2
Coshn cosh-1(1/wp) =  α max/ 10 
 (10 − 1) 
1
 (10 α min/ 10 − 1)  2
cosh −1  α max/ 10 
 (10 − 1) 
∴ n= ………………(vii)
cosh −1 (1 / w p )
Which gives the required order for the inverse chebyshev filter.
Now , for half power frequency i.e at w = wp
TIC ( j.1) = 1/√2
2
TIC ( j.1) =½
Which means,
∈2 c n (1 / wP ) = 1
2
2 1
c n (1 / wnp ) =
∈2
1
c n (1 / w) =
∈
1
Coshn.cosh-1(1/wnp) =cosh-1
∈
1
n coshn.cosh-1( (1 / wnp ) = cosh-1( )
∈
1
cosh-1( (1 / wnp ) = 1/n. cosh-1( )
∈
1
1/wnp = cosh[1/n. cosh-1( )]
∈
1
wnp = < 1 ……….(viii)
1 1 
cosh  cosh −1 ( )
n ∈
Which gives the desire half power frequency.

Example: 01
Given, α max = 0.5 dB
αmin = 22 dB
wp = 0.9
n=?
wnp = ?
52
Assignment:
Example:02 Differentiate between Butterworth , chevyshev and inverse chebysehev filters.
Pole zero location for inverse chebyshev:

We know that ,
∈2 c n (1 / w)
2
2
TIC ( jw) =
1+ ∈2 c n (1 / w)
2
T(s). T(-s) = z(s).z(-s)/[p(s).p(-s)]

Where, z(s) z(-s) |s = jw = ∈2 c n 2 (1 / w)
P(s) P(-s)|s = jw = 1 + ∈2 c n 2 (1 / w)
For zero location:
∈2 c n (1 / wk )
2
∴ ∈≠ 0 ⇒ c n 2 (1 / wk ) = 0
c n (1 / wk ) = 0
Cosn cos-1(1/wk) = cos(kπ/2) for k = 1,3,5 ………..(i.e odd)
ncos-1(1/wk) = kπ/2
1/wk = cos(kπ/2n) which gives the zero for inverse chebyshev.
Wk = sec(kπ/2n)
For poles:
1+ ∈2 c n 2 (1 / wk ) = 0
The poles location are similar to chebyshev.
Simply replacing wk by 1/wk
i.e if chebyshev poles = pi
Then , inverse chbyshev poles = 1/pi
Fig. Zero location

Fig. Pole location
Example:01
Given,
α min = 18 dB
αmax = 0.25 dB
ws = 1.4 rad/sec
wp = 1 rad/sec
Find out the pole and zero for inverse chbyshev response.
53
Chapter 5
Frequency transformation:
Frequency transformation is important because the prototype LPF with any type of
approximation can be converted into high pass band pass , band stops filters within the same
characteristics easily.
T(jw)
1
0.707
Wc
The effect of frequency transformation are:

1. Magnitude response 2. Network function
3. Location of poles and zeroes. 4. Network elements.
Types of transformation:
1. LP to LP transformation
Transformation
W
W0 Ω0
Old LPF New LPF
Replace s by wo/Ωo .s
i.e
∴ w0 = 1 ( in normalized case)
∴ s → s/Ω0
∴ TLP(new) (s) = TLP(old) (s/Ω0)
For eamaple,
If
TLP(s) = 1/S+1
Then
TLP(old)(s) = 1/s+1
∴ TLP(new)(s) = TLP(old)(s/Ωo) = 1/(s/Ωo)+1 = Ω0/(s+Ω0)
1. For resistor:
- No change.
2. For inductor:
XL = LS
s
Putting s →
Ω0
54
s L
XL’ = Lold = old s = Lnew .S
Ω0 Ω0
∴ Lnew = Lold/Ω0
3. For capacitor:
Xc = 1/cs
s
Putting s →
Ω0
1 1 1
Xc’ = = =
s C old C new .s
C old .s
Ω0 Ω0
∴ Cnew = Cold/s
2 LP to HP Transformation:
Transformation
W
W0 Ω0
LPF with W0 HPF with Ω0
Ω0
In this case we replace s →
w0 .s
Ω0
Or , s → [Since w0 = 1]
s
∴THP(s) = TLP(s)
s=
Ω0
= TLP ( )
Ω0
S
s
Example if TLP(s) = 1/(s+1)

1 s
Then, THP(s) = =
Ω0 Ω0 + s
+1
s
(1) For resistor:
No change
(2) For inductor:
XL = LS
Ω0
Putting s →
s
55
Ω0 1
XL’ = L. =
s  1 
 .s
 LΩ 0 
1
Comparing with 1/CS
 1 
 .s
 L Ω 0 
1
C=
LΩ 0
(3) For capacitor:

Xc = 1/cs
Ω0
Putting s →
s
1 s  1 
XL’ = = =  .S = LS
 Ω0  cΩ 0  CΩ 0 
 .c
 s 
 1 
Comparing  .S with LS
 CΩ 0 
 1 
L =  
 CΩ 0 
Date: 2056/6/15
(3) LP to BP Transformation:
T (jw) TBP(jΩ )
LP
Transmission
W Ω
Wp Ws ΩL ΩU
In this case,
s2 + Ω2
s → w0 .
Ωu − Ω L
Here, Ωu – ΩL = B
And w0 = 1
s2 + Ω2
∴ s →.
Bs
Where Ω02 = ΩL. Ωu
56
(1) For resistor

- no change
(2) For inductor:

XL = LS
The new value of inductive reactance is given by:
 s 2 + Ω0 2 
XL’ = L.  

 Bs 
LΩ 0
2
L L 1
XL’ = .s + = .s +
B Bs B B
.s
LΩ 0
2
∴ The new component are inductor and capacitor in series.

L B
B LΩo2
(3) For capacitor:

The new capacitive reactance form LP to BP is given by :
1 1 1 1
= = = =
s + Ω0 cs + cΩ 0 cΩ 0
2 2 2
2 2
c c 1
c. s+ s+
Bs Bs B Bs B B
.s
cΩ 0
2
The new components (i.e inductor and capacitor) are in parallel as shown in fig. below:
B C
CΩo2 B
LP to BS Transformation:
T (jw) TBS(jΩ )
LP
Transmission
Wp W ΩL ΩU Ω
Ws
57
Bs
In this case s is replaced by .w0
s 2 + Ω0
2
Bs
But w0 = 1, ∴ s→
s + Ω0
2 2
(1) For resistor :

Resistor value remain same.
(2) For inductor:
XL = LS
Bs 1 1 1
XL = L. =. = =
s + Ω0 s + Ω0 Ω
2 2 2
2 2 2
s 1 1
+ 0 s+
LBS LBS LBs LB LB
.s
Ω0
2
The new component (i.e inductor and capacitor ) are in parallel as in figure below:
LB 1
2
Ωo LB
(3) For capacitor:
Xc = 1/cs
s 2 + Ω0 Ω
2 2
1 s2 1 1
Xc = = = + 0 = .s +
Bs CBS CBs CBs CB CB
c. .s
s2 + Ω0 Ω0
2 2
CB 1
2
Ωo CB
1
Example:01: If T(s) = , then change the above function from LP to BP. Given , ΩL =
s +1
10 and Ωu = 20.
Solution:
1
Then, TLP(s) = , ΩL = 10 , Ωu = 20
s +1
We know ,
Ω02 = ΩL. Ωu = 10. 20 = 200
For Lp to BP we replace
s2 + Ω0 s 2 + 2000 s 2 + 200
2
s→ = =
B (20 − 10) s 10 s
58
Thus,
1 10 s
TLP s 2 + 200 = TBP ( s ) = 2 = 2
s + 10 s + 200
s= s + 200
10 s +1
10 s
10( s )
∴TBP(s) = 2
s + 10 s + 200
Example:02: Obtain the transfer function of the 4th order Butter worth HPF with Ω0 = 2π
×104 rad/sec.
1
TLP(s) =
s + 2.61313s + 3.41921s 2 + 2.61313s + 1
4 3
We know that ,
Ω0
s→
s
1
=
 Ω0  Ω  Ω  Ω 
4 3 2
  + 2.61313 0  + 3.41921 0  + 2.61313 0  + 1

 s   s   s   s 
Example:03:The filter shown in the figure below is a 4th order chebyshev low pass filter with
αp = 1 dB and wp = 1. Obtain a bandpass filter from this low pass with Ωo = 400 rad/sec and
B = 150.
A C
V1
+
-
} B } D
} E
V2
Solution:
For LP to BP conversion , we replace
s2 + Ω0
2
s→
Bs
Where, Ωo = 400 rad/sec , B = 150
Now for section A:

L = 1.2817
Which changes to series LC component as shown below:

L B
B LΩo2
∴ The new inductor value is = L/B = 1.2817/150 = 8.54 mH

and the new value of capacitor is = B/L Ω20 = 150/(1.2817×4002) = 731.45 µF.
For section B:
59
C= 1. 9093
Which changes for LP to BP As:
B C
CΩo2 B
New inductor value = B/C Ωo2 = 150/(1.9093×4002)= 491.01 µF

New capacitor value = C/B = 1.9093/150 = 12.72 µF
For section C:
L = 1.4126
B L
LΩo2 B
For section D:
B C
CΩo2 B
For section E:
R=1 R=1
+
8. 54 mH 9. 41 mH
731. 45uF 663. 66uF
V
+
491uH 12.72 uF 1Ω
893. 71mH
- 6. 99 mF
Date: 2065/6/16
60
Doubly Terminated LC-Ladder ckt:

I2
R1 I1 +
1
+ Loss less V2
Vs zi n R2
Ladder
- V1
Fig.1 Doubly Terminated LC ladder ckt.
From figure(i)
I1= Vs/(R1 +Vin) ……………(i)
Where,
Zin = Rin + jxin …………….(ii)
Since the ckt is loss less
Input power = output power
P1 = zin |I1(jw)|2 = |V2(jw)|2/R2 ……………(iii)
From equation (i) and (iii)
zin|Vs(jw)|2/(R1+zin) = |V2(jw)|2/R2
or , |V2(jw)|2/|vs(jw)|2 = zin R2/(R1+zin)2 …………..(iv)
Now for matched source.
R1 = zin
Which means
V1 = vs/2
∴ P1max = |v1(jw)|2/R1 = |vs(jw)|2/4R1
Also it is to remember that ,
P2 = |v2(jw)|2/R2
|(jw)|2 = p2/p1max = [|v2(jw)|2/R2]/ |vs(jw)|2/4R1 = 4R1/R2. |v2(jw)/vs(jw)|2 …………..(vi)
Form equation (iv) and (vi)

|H(jw)|2 = 4R1/R2. {zin R2/(R1+zin)}
= 4R1zin/ (R1+zin)2 = 1- (R1-zin)2/(R1 +zin)2
(R1-zin)2/(R1 +zin)2 = |(jw)|2
= reflection coefficient
( R1 − z in ) 2
ρ ( s).ρ (− s) =
( R1 + z in ) 2
( R1 − z in )
ρ ( s) = ± .....................(vii)
( R1 + z in )
From equation (vii) , we get
1 − ρ (s)
z in = R1 . −−−−−1st zin ……..(viii)
1 + ρ (s)
1 + ρ (s)
z in = R1 . -----------2nd zin
1 − ρ ( s)
Generally we take R1 = 1. Both impedances in equation (viii) are reciprocal impedance.
Synthesis of Doubley Terminated LC ladder with equal terminal (All pass filter)
61
For butterworth response:
1 N ( s) N (− s)
T ( jw = H ( jw) = =
2 2
[since w0 = 1]
1+ w 2n
D( s) D(− s)
1 w 2n
ρ ( s ) = 1 − H ( s ) = 1 − H ( jw) = 1 − =
2 2 2
1 + w 2n 1 + w 2n
w 2n w 2n s n .(− s ) n
ρ ( s).ρ (− s) = = = ……(ix)
1 + w 2n D( s ).D(− s ) D( s ).D(− s )
Now,
For n = 1
D(s) = s+1 [since T(s) = H(s) = 1/S+1]
Form equation (ix)
ρ(s) = sn/D(s)
= s1/s+1 = s/s+1
s
1−
1 − ρ ( s) s +1− s
∴ zin1 = R1 . = 1. s + 1 = .
1 + ρ (s) s s +1+ s
1+
s +1
1
Zin1 = . .............(a)
2s + 1
Zin2 = 2s+1 ……….(b)
Zin2 = 2s+ 1 = Ls + R
i.e L = 2, and R = 1
∴ The ckt will be
R1 2
+ R2 1
vs
-
From equation (a) , zin1 = 1/(2s+1) i.e c = 2, and R = 1

+
1
+ 1 v2
vs
-
2
For n = 2
D(s) = s2+√2s + 1
sn sn
ρ ( s) = = 2
D(s) s + 2 s + 1
1 − ρ ( s) 1 − s 2 /( s 2 + 2s + 1) (s 2 + 2s + 1 − s 2 )
∴ zin1 = = = 2
1 + ρ (s) 1 + s 2 /( s 2 + 2s + 1) (s + 2s + 1 + s 2 )
62
( 2s + 1)
∴ zin1 = ……..(a)
(2s 2 + 2 s + 1)
Similary,
2 s 2 + 2s + 1
Zin2 = ……….(b)
2s + 1
Taking equation (b)
√2 s+1) 2s22+√2 s +1 (√2.s z1(s)

2s +√2 s
1 ) √2.s + 1( √2 s Y2(s)
√2 s
1) 1 (1 z3(s)
1
∴ The ckt will be as follows:

+
1 1.41
+ 1 v2
vs
-
1.41
-
2
1 1.41
+ 1.41 1
vs
-
Home work : For n = 3 and n = 4
Date: 2065/6/17
Synthesis of Doubly Terminated LC - Ladder with unequal termination: ( R1 ≠ R2) :

For R1 ≠ R2 the butter worth response is given by ,
H 2 ( 0)
H ( jw) = = T ( jw)
2 2
1+ w 2n
Generally we take,
R1 ≠ 1 and R1 ≠ R2
R1 I1 I2
+ R2 V2
Zin LC
- Vs
63
V2
From figure, the transform function , T(s) =
Vs
From which we get ,
R2
T(0) =
R2 + R1
Now we know
2
4 R V (s)
H ( s) = 1 . 2
2
R2 Vs (s)
4 R1
H (s) =
2 2
. T (s)
R2
R1
H (S ) = 2 .T ( s )
R2
R1 R R2
H (0) = 2 .T (0) = 2 1 .
R2 R 2 R1 + R 2
R 2 .R1
∴ H ( 0) = 2
R1 + R 2
Example:01: Realize the doubly terminated ladder filter with a Butter worth response for n
= 3, R1 = 1, R2 = 2 .
Solution:
We know, for unequal termination ( i.e R1 ≠ R2) the Butterworth response is given by,
H 2 ( 0)
H ( jw) =
2
1 + w 2n
Here, n = 3, R1 = 1 & R2 = 2
4 R2 .R1 4 .1 .2 8
H2(0) = = =
(R2 + R1 ) 2
(1 + 2) 2
9
8/9
∴ H ( jw) =
2
1 + w 2n
The reflection coefficient function is
ρ ( jw) = 1 − H ( jw)
2 2
8/9 1 + w 2n − 8 / 9 1 / 9 + w 2n
=1 − = =
1 + w2n 1 + w 2n 1 + w 2n
1 / 9 − ( s / j ) 2×3 1 / 9 + ( s / j ) 6
ρ ( jw) =
2
=
1 + w 2×3 1 + w6
1 / 9 − ( s ) 6 (1 / 3) 2 − ( s 3 ) 2 (1 / 3 − s )(1 / 3 + s )
Or, ρ ( s) =
2
= =
1 − s6 1 − s6 1− s6
(1 / 3 + s 3 ) (1 / 3 − s 3 )
ρ ( s ).ρ (− s ) = .
D(s) D(− s)
Where, D(s). D(-s) = 1- s6
64
1/ 3 + s3
∴ ρ ( s) =
D( s)
For n = 3,
D(s) = s3+2s2 + 2s+1 (from table)
∴ The first impedance is ,
1/ 3 + s3
1−
1 − ρ ( s) s 3 + 2 s 2 + 2s + 1
Zin1 = =
1 − ρ ( s) 1/ 3 + s3
1+ 3
s + 2s 2 + 2s + 1
2s 2 + 2s + 2 / 3
Zin1 = 3 ……….(a)
2 s + 2s 2 + 2s + 4 / 3
2 s 3 + 2s 2 + 2s + 4 / 3
Zin2 = …………..(b)
2s 2 + 2s + 2 / 3
Now using continued fraction method for equation (b)
2s2+2s+2/3 ) 2s3+2s2 +2s + 4/3 ( s z1(s)

3 2
2s +2s +2/3.s
4/3.s+4/3 ) 2s2 +2s+ 2/3 (3/2. s Y2(s)
2s2+2s
2/3 ) 4/3.s +4/3(2s z3(s)
4/3.s
4/3 ) 2/3 (1/2 Y4(s)
2/3
1 2
+ 1
-
2/3 2
Home Assignment:
• Try it for n = 1, 2, 3 and 4 , for unequal terminal i. e R1 = 1 and R2 = 2.
[ for n = 4, D(s) = s4+2.16s3+3.14s2+2.6s+1]
• Review of ideal and non ideal properties of operational amplifiers, GBP, CMRR,
Inverting and non inverting A/F.
65
Fundamental of Active filter circuit:-

Ideal & Non-ideal properties of op-amp.
Gain Bandwidth product( GBP)
CMRR & its importance.
The main advantage of Active filter:-

Small in size
Provide grater amplification
Cheaper than passive filter.
The limitation area:-

Extra Vcc is required
Sensitive to temperature
Low gain at high temperature
Low gain at high frequencies
CMRR should be high
Certain important configuration of op-amp:-

Rf
R
-
+ Vo
Rf
Vo = − .Vi
R
(2) Non-investing:-
Rf
R
-
+ Vo
Vi
 RF 
Vo = 1 + Vi
 R 
(3)Integration:-
66
c
R
-
Vi + Vo
1 1  1
Vo = − .Vi =  − Vi
RCS RC  S 
If R1=1 & C = 1, then
Vo 1
=− I.e. Integrator always contributes polo.
Vi S
(4) Differentiator:-
Rf
-
c + Vo
Vi
Vi − O Vo − O
=
1 R
CS
Vo = −(CRS )Vi
If Ro= 1& Co= 1, Then
Vo
= −S
Vi
(5) Summer:-
Rf
R1
V1
-
V2 + Vo
R1
Vo = −
RF
(V1 + V2 )
Ri
(6) Subtract or (Difference A/F)
67
Rf
R1
V1 -
V2 + Vo
R1
Rf
Vo =
RF
(V2 − V1 )
Ri
Design of Active filters (op-amp based):-

(1) Investing type:-
Z2
V1 Z1 -
+ V2
From fig.
V1 ( S ) − Z 2
R(S) = =
V1 ( S ) Z1
(a) T(S) = -K/S
Since, the above T(S) contributes polo we can reduce the T(S) with T(S) of integrator
−K
I.e. T (S ) = −
1
=
RCS S
1
⇒K=
RC
If R=1, then,
C=1/K
⇒ If C=1, then, R=1/K
1/k
R =1
-
+ V2
Thus the design will be

(b) T(S)= -KS (Do yourself)
(c) T(S) = -K(S+a1)
We can compare with the general T(S) of investing type ie.
Z2
T(S) = −
Z1
68
= k (S + a1 )
Z2
∴
Z1
1
y y
Or 2 = 1 = k ( S + a1 )
1 y2
y1
If y 2 = 1, then,
Y1 = KS+Ka1
1
∴Y2
Y1
1/ka1
1
-
+ V2
1/ka1
Fig:- Design for R(s) = - (s + a1)
K
(d) T(S) = −
S + P1
1/k
Let we can write,
Z2 K
=
Z 1 S + P1 k/p1
y1 K -
= + V2
y 2 (S + P1 )
1
y1 1
= fig: Design for T(S)= -K/(S+P1)
y 2 (S + P1 )
K
y1=1, then
S + P1 S P1
y2 = = +
K K K
− ks
(e) T(S) = 1
s + p1
1/k k/p1
Z2 1 -
= V1
Z1 1 P + V2
+ 1
K KS
If Z2=1, then,
69
1 P
Z1 = + 1
K KS
S + q1
(f) T (S ) = − K
S + P1
Z2  s + q1 
= K  
Z1  s + p1 
y1  s + q1 
= k  
y2  s + p1 
Let y1= ks + kq1
Then, y2 = s + p1
1
k
1/p1
V1 -
+ V2
1/ka1
− k (s + q1 )
Fig: Design for T(S) =
(s + p1 )
# 2nd approach of above problem Do Yourself)
(Do Yourself
(2)Non-investing type:-
V1 -
V2
+
Z2
Z1
a1-p1/p1
-
+ V2
k (s + q1 )
(a) T(S) = Where, q1>p1
(s + p1 )
Comparing, 1/a-p1
z2  s + q1  1
1+ = k  
z1  s + p1 
70
z2  s + q1 
= k   −1
z1  s + p1 
ks + kq1 − s − p1
=
s + p1
z2 s (k − 1) + (kq1 − p1 )
=
z1 (s + p1 )
For, k = 1
z2 q1 − p1 k (s + q1 )
= T(s) = for k = 1
z1 s + p1 (s + p1 )
y1 1
=
y2 s p1
+
q1 − p1 q1 − p1
If y1 = 1, then
s p1
y2 = +
q1 − p1 q1 − p1
For, k ≠ 1
z 2 s (k − 1) + (kq1 − p1 )
=
z1 (s + p1 )
We assume, 1200
Kq1 = p1
K = p1/q1
s ( p1 − q1 )
s (k − 1)
-
z
∴ 2 = =
q1
V1 + V2
z1 (s + p1 ) (s + p1 )
z2 p −q
= 1 1
z1 q1
1 + p1 / s
p1 − q1 k (s + q1 )
If z2 = Design for T(s) =
q1 (s + p1 )
Then, z1 = 1 + p1 / s for k ≠ 1 & p1 >q1
y1
# 2nd approach =? (Do Yourself)
y2
Example:- 01
Realize 1st order inverting which satisfy the following T(s)
1000
T(s) =
s + 1000
We know that,
− z2 − 1000
=
z1 s + 100
71
y1 1000 1
= =
y2 s + 1000 s
+1
100
If y1=1 then,
s 1
y2 = + 1, c =
1000 1000
Example:-02
Realize the 1st order Non- inverting filter with following T(s)
s+4
T (s ) =
s+8
soln:- For the case given,
z2 s + 4
T(S) = 1 + =
z1 s + 8
z2 s + 4 − s − 8 − 4
Or, = =
z1 s+4 s
∴ The direct approach does not provide the required design; we go in the following manner.
Here, P1 =8
Q1=4
Ie, p1=q1
p1 8
Let, k = = =2
q1 4
2(s + 4 ) 1  (s + 4 )   1 
∴T ( s) = . = 2  − (− 1)
(s + 8 ) 2  (s + 8 )   2 
= T1 ( s ).T2 (s ).T3 (s )
2( s + 4 )
For, T1(S) =
(s + 8)
z2 2 s + 8 − s − 8
1+ =
z1 (s + 8 )
z2 s
=
z1 s + 8
y1 s
=
y2 s + 8
If y1= S, then y2 = S + 8
72
1 1
- 1
- -
+ + V2
1 2 +
1 1/8
RC-CR Transformation:-
1 ohm
z2 1000
T (S ) = =
z1 S + 1000 1 ohm 1mf
-
It is law pass response, + V2
1 1
If, Ri = = = 1kΩ
Ci 1mf 1F
1 1
& Ci = = = 1F
Ri 1
1000
V1 - V2
z s +
T(s) = 2 =
z1 s + 1000
It is the transformation by which a low pass filter can be converted into a high pass filter by the
simple change in the component Value i.e. In this case Ri is replace by Ci and Ci=1/Ri and Ci is replaced by
Ri and Ri=1/Ci.
CHAPTER:- 8
Biquad circuits:-
Consider a 2nd order filter as shown below:-
L - At low freq, c behaves as line open cut so, V2 = V1
R
- At high freq. c behave as line short cut V2= O
Hence, it is a low pass filter.
+
-
c V2
1
v (s ) cs
T(s) = 2 =
v1 (s ) 1
R + LS +
cs
1
= Lc
R 1
s2 + s +
L Lc
73
R 1
To get poles, S 2 + S+ =0
L LC
&, for loss less ckt , ie, if R= 0,
Then, S2+ 1/Lc =0
1 1
Or, s = ± j = ± jwo where, wo =
LC Lc
∴ Poles are imaginary and conjugate,
Quality factor:-(Q)
WoL
Q=
R
It is defined as the ratio of inductive reactance at frequency Wo to the resistance.
Now,
1 L 1 L
Q= =.
LC R R C
Wo R
Also, =
Q L
Wo 2
∴ T (s ) = …………………………. (i)
Wo
S2 + S + Wo 2
Q
This is the standard from & the design parameter is Wo & Q.
V1 Wo1 .Q V2
Wo
To get the actual poles:- S 2 + .S + Wo 2 = 0
Q
Let, the poles be, − ε ± jβ then,
D(S) = (S + α + jβ )(s + α − jβ )
( )
D(S) = S 2 + 2αβ + α 2 + β 2 = 0 …………………………… (ii)
Comparing equation (i) and (ii)
Wo
2α = & α 2 + β 2 = Wo 2
Q
Wo
∴α =
2Q
1
& β = Wo 1 −
4Q 2
Typing Biquad current:-
74
2
Gw0
A typical Biquad ckt can be represented as, T(s) = ±
w
s 2 + 0 s + w0
2
Q
Where, G = Gain & ± choice of inverting and non inverting.
In normalized case, i.e. for w0 = 1
±G
T (s ) = ……….. (i)
s
s + +1
2
Q
Equation (i) can be implemented if G & Q ate given,
Let us go for inverting type of design
−G
i. e. T (s ) =
s
s2 + +1
Q
v2 G
=−
v1 s
s2 + +1
Q
 s 
− Gv1 =  s 2 + + 1v2
 Q 
  1 
− Gv1 =  s s 2 +  + 1 v 2
  Q 
Gv1 v2
v2 = − −
 1  1
s s +  s s + 
 Q  Q
v2 v2
v2 = − −
 1  1
 s +  s s + 
 Q  Q
 
 
− vs Gv1   1 
v2 =  − − [− 1]………. (ii)
 1  1    s 
  s +   s +  
 Q  Q 
The equation (ii) is cascade Realization costing of 3 steps:
Stage:-1
 
 
1  −G 
− v + v1 ………… ii (a)
1 2  1 
s+  s + 
Q  Q  

Stage:-2
1
− ……….. ii (b)
s
75
1
-
+
Stage:-3
(-1) = ii(c)
-
+
For stage 1, we need more analysis:
Z3
V1 Z2
-
Z1 + V3
From figure
z3 z
v3 = − .v1 − 3 .v2
z1 z2
 − 1   −1 
v3 = z 3  v2 +  v1  ……….. (iii)
 z 2   z1  
From equation ii (a)
v3 =
1
1
[− 1v2 + (− G )]v1
s+
Q
  
1  −v  1 
∴ v3 =  2 +  −  v1 …………. (iV)
1 1
s+   1 
 
Q   G 
Comparing eqn (iii) & (iV)
1
z3 =
1
s+
Q
z 2 = 1 (a resistor)
1
z2 = − (a resistor)
G
76
1
For , z 3 =
1
s+
Q
1
Or, y3 = s +
Q
∴ The ckt for Z3 will be
1
Q
1
1/Q
V1 Q
-
V2 + V3
1
The overall ckt will be,

1
1 1
1/Q Q
1 1
- -
+ -
1 + + V2
Fig: This is ring of 3 ckt and is popularly known as two Thomas Biquid.
Two Thomas Biquid:-

C1 C2
R5
R3 R1
V1 R4 R5
- -
V2 + -
V3 + V4 + V2
R2
Fig:-General Two Thomas Biquid.

From figure,
1
− R1
c1 s  v2 v1 
v3 =  +  ……… (i)
1  R2 R3 
R1 +
c1 s
1
v4 = − v3 ……… (ii)
R4 c 2 s
v2 = −v4 ……… (iii)
From eqn (ii) & (iii)
1
v2 = − v3 ……… (iv)
R4 c 2 s
Again, from eqn (i) & (iv)
77
− R1
R4 c2 s.v2 =
c1 s  v2 v1 
+
1  R2 R3 
R1 +
c1 s
1
−
∴T (s ) =
v2 R3 R4 c1c2
= ………… (V)
v1 1 1
s + 2
s+
Rc1 R2 R4 c1c2
But the standard form of Biquid is
− Gw0
2
T (s ) = 2 =
v
……… (Vi)
v1 w0
s + .s + w0
2 2
Q
Comparing eqn (V) & (Vi)
1
w0 =
2
R2 R4 c1c2
1
∴ w0 =
2
……… (Vii)
R2 R4 c1c2
1
= Gw0
2
Also,
R2 R4 c1c2
1 1
or, = G.
R2 R4 c1c2 R2 R4 c1c2
R2
or, G = …….(Viii)
R3
Finally,
w0 1
=
Q R1c1
1 1
or, =
R2 R4 c1c2 R1c1
Q
2
R1 c1
⇒Q= ……… (ix)
R2 R4 c1c 2
With,
c1 = c 2 = 1
& R2 = R4 = 1
We get,
W0 = 1
1
G=
R3
1
R3 =
G
Q = R1
78
⇒ R1 = Q
The important property of the Biquid ckt is that it can be orthogonally turned. It means
(a) R2 can be adjusted to a specified Value of w0.
(b)R1can then be adjusted to give specified of Q without changing w0, which has been already adjusted.
(3) Finally, R3can be adjusted to give the desired Value of G fir the ckt without changing w0 & Q which has
already been set.
These three steps are known as tuning algorithm.
Sallen – key Biquad circuit:-

C1
R1 R2 Va
-
+ V2
V1 -
+
C2
RA RB
Fig: Sallen-key Biquad

From fig (i),
v2 R
= 1 + A = k ………………. .(i)
v1 RB
v2
⇒ Va = ………………… (ii)
k
Applying Nodel Analysis at mode a,
V a − Vb V a − 0
+ = 0 …………… (iii)
R2 1
S
C2
Applying Nodal analysis at node b,
v2
vb −
vb − v1 k + vb − v 2 = 0
+
R1 R2 1
c1 s
 1 1 1  v2 v
Or, Vb  + +  − − 1 − v 2 c1 s = 0
 R1 R2 R3  R2 k R1
 1 1  v  1 
Or, Vb  + + c1 s  − 1 − v 2  + c1 s  = 0
 R1 R2  R1  R2 k 
Similarly, rearranging eqn (iii)
v2 v2
k − v3 + k = 0
R2 R2 1
c2 s
 1 c s v
 + 2 v2 − b = 0
 R2 k k  R2
79
 1 c s
Or, vb = R2  + 2 v 2 ……….(Vi)
 R2 k k 
Thus from eqn (V) & (Vi)
 1 c s  1 1  v  1 
R2  + 2  + + c1 s v 2 − 1 − v 2  + c1 s  = 0
 R2 k k  R1 R2  R1  R2 k 
  1 c s  1 1   1  v
Or,  R2  + 2  + + c1 s  −  + c1 s  v 2 = 1
  R2 k k  R1 R2   R2 k  R2
1
k
⇒ T (s ) =
v2 R1 R2 c1c 2
=
v1  1
s 2 +  +
1
+
(1 − k ) s + 1

 R1c1 R2 c1 R2 c 2  R1 R2 c1c 2
2
Gw0
T (s ) = ………. (Viii)
 w0 
s +   s + w0
2 2
Q
Comparing eqn (Vii) and (Viii)
G=k
1
w0 =
R1 R2 c1c 2
w0 1 1 1− k
= + +
Q R1c1 R2 c1 R2 c 2
Design – I (equal elements Values):-

In this case,
R2 = R1 = R = 1
& C1 = C2 = C = 1
For which,
W0 = 1
1 R
K = 3− = 1+ B
Q RA
Now, let us take,
RA = 1 then,
RB = 2- 1/Q
∴ In this case the final ckt will be,
C1= 1
R1= 1 R2= 1
- V2
+
V1 -
+
C2 = 1
RA=R
1
B = 2-1/Q
80
Design –II (unity gain design):-
In this case k = 1
This is turn require d the non – inverting ckt to be replaced by a Voltage followers.
We keep,
R1 = R2 = 1
We know,
w0 1 1 1− k
= + +
Q R1c1 R2 c1 R2 c 2
w0 1 1
⇒ = +
Q c1 c1
w0 2
Or, =
Q c1
1 1
Also, W0 = =
R1 R2 c1c 2 c1c 2
But, we take, in normalized case,
W0 = 1
⇒ C1C2 = 1
C1= 1/C2
2Q
Thus, C1 = 2Q
C2 = 1/2Q
1 1
∴ The final ckt will be, -
+
V2
-
+ V1
1/2Q
Design –II (equal capacitance of equal feed bake):-

In this case,
C1 = C2 = C = 1
RA = RB = R = 1
∴ K= 1+RB/RA = 2
∴ W0 = 1
w0 1 1 1 1− 2
= = + +
Q Q R1 R2 R2
⇒ R1 = Q
1
Also, W0 = =1
2
R1 R2 c1c 2
⇒ R2 = 1/Q
∴The final ckt will be,
81
1
Q 1/Q
- V2
+
V 1 -
+
1
1 1
Example:- 01
Design a 4th order Butterworth filter using equal element of Sallen Key ckt. Then Let w0 =2 π 1000
rad/sec & use capacitor of 0.1µF
n
Sol .:-
The 4th order Sallen Key in blocks can be represented by
1st 2nd
1st order 2nd orderr
From table,
Q1 = 0.54 and Q2 = 1.31
st
For 1 stage:-
Wo = 1, Q1 = 0.54
& for equal element design in Sallen key,
R1 = R2 = 1
& C1 = C2 =1
RA = 1,
RB =2 – 1/Q = 2 -1/0.54
RB = 0.148
nd
For 2 stage:-
WO = 1
Q2 = 1.31
R1 = R2 = R = 1
C1 = C2 = C = 1
RA = 1
RB =1/Q2 = 2-1/1.31 = 1.236
∴ The design should be for,
Ω = 2π 1000
& C = 0.1 ϑf
We can apply both magnitude and frequency scaling at once.
Now, we know,
C old 1
C new = = = 235.54
C new .kf 0.1 × 10 × 2π 1000
−6
For 2nd stage,

RBnew = Km RBold = 1591.54*1.236 = 1967.14
The ckt will be,
82
0.1µ F 0.1µF
1.6k 1.6k 1.6k

- -
+ 1.6k +
V2
V1 -
+
0.1µF 23 6
2k
1.6k 1 1.6 k
Fig: 4th order butter worth active Salleney biquad with equal element design
for W = 2π 1000 rad / sec & C = 0.1µF .
Gain adjustment (EqualiZation in Sallen key:-

KWo 2
T (S ) = ………………….. (i)
 Wo 
S + 
2
.S + Wo 2
 Q 
In Butterworth,
T (jo) = 1 or 0 dB
But in equation (i) T (jo) = k (k>/1) which needs to be equalized.
Ra
R1
-
V1 Vb
- + Rb
+ V1
Fig:- i(a) fig:- i(b)
If ‘H’ is considered to be the gain provided by fig i(b) which is such that,
H. k = 1
Also it is to noted that, in Sallen key,
G=K
Also, H. G = 1
H =1/G
Vb Rb
Now, T (S) = =
V1 Ra + Rb
Also, we should remember that,
Ra Rb
= R1
Ra + Rb
Now, solving the above equation by setting R=1, we get Ra = 1/H
In term of ‘G’ the Value of Ra & Rb is
Ra = G
G
Rb =
G −1
83
In term ‘Q’ Ra and Rb can be expressed as,
1
Ra = 3 −
Q
3 −1
∴G = K =
Q
3Q − 1
Rb =
2Q − 1
Gain Enhancement (Increment) in Sallen key:-

C1
R1 R2
-
+ V2
V1 -
+
Ra
C2 RB RA
Rb
We have, gain,
1 R
K = 3− = 1+ A
Q RB
But, sometimes for given ‘Q’ the Value of gain will be Very small and amplification to our need. Although
the separate ckt for gain enhancement can be used, the Sallen key ckt itself can be modified to compensate
the gain, using additional arrangement of two resistor as, shown in the fig (ii)
Let, C1 = C2 = C
& R1 = R2 = R then,
T(S) = of sallen key will be,
k
T(S) = R C2 2
 3 − µk  1
S2 + .S + 2 2
 RC  R C
Rb
Where, µ =
Ra + Rb
Wo 3 − µk
∴ =
Q RC
1
Q=
3 − µk
For a given Value of ‘Q’ the gain ‘k’ can be increased to our requirement by proportionally decreasing the
new factor µ.
84
High pass sallen key:-

In this case,
GS 2
THP(S) =
 Wo 
S 2 +  .S + Wo 2
 Q 
Applying RC-CR transformation in active low pass Sallen key biquid, in the non-inverting terminals we get
the following final ckt THP(S) as,
1/C1
1/R1 1/R 2
-
+ V2
V1 -
+
1/C2 RA R B
KS 2
THP(S) =
 1 1 K −1 1
S 2 +  + + .S +
 R2 C 2 R2 C1 C1 R1  R1 R2 C1C 2
1
Where, WO =
R1 R2 C1C 2
G=k
Wo 1 1 k −1
= + +
Q R2 C 2 R2 C1 R1C1
1/C1
Band pass Sallen key Biquad:-
R1 C2
-
In this case, + V2
V 1 -
+ C1 R3
RA R B
 k 
 .S
 R1C1 
TBP(S) =
 1 1 1 1− k R1 + R2 
S 2 +  + + + .S + 
 R1C1 R3 C 2 R3C1 R2 C1 R1 R2 R3C1C 2 
Where,
R1 + R2
WO =
R1 R2 R3C1C 2
WO
Q=
1 1 1 1− k
+ + +
R1C1 R3C 2 R3C1 R2 C1
85
k
R1C1
G=
1 1 1 1− k
+ + +
R1C1 R3C 2 R3C1 R2 C1
Also, in standard from,
W 
G O .S
TBP(S) =  Q 
W 
S 2 +  O .S + WO 2
 Q 
Where, W1 = WO – Bw/2
W2 = WO + Bw/2
WO WO
And, Q = =
Bw W2 − W1
Band stop Sallen key:-
TBS(S) =
(
G S 2 + WO
2
)
W 
S 2 +  O .S + WO
2
 Q 
R1 R2
V1 -
+ V2
C1 C2
R A RB
Assuming,
R2 = R1 = R
C1 = C2 = C
R3 = R/2
We obtained,
 1 
K S 2 + 2 2 
TBS(S) =  R C 
4(1 − k )S 1
S2 + + 2 2
RC R C
1
WO =
RC
1
Q=
4(1 − k )
RB
G = K = 1+
RA
86
# Use equal amount design to obtain Bond pass Sallen key Biquad with WO =1 & Q =10. Also find the
upper band & lower band frequency. When WO =1000Hz
WO =1
Q = 10
W1, W2 if WO = 1000Hz
R1 + R2
WO =
R1 R2 R3C1C 2
WO
Q=
1 1 1 1− k
+ + +
R1C1 R3C 2 R3C1 R2 C1
k
R1C1
G=
1 1 1 1− k
+ + +
R1C1 R3C 2 R3C1 R2 C1
R2
V1 R1 C2
-
+ V2
C1 R3
RA R B
R1 = R2 = R3 = R
C1 = C2 = C =1
From which, we get,
2R 2
WO = B
=
R R
But,
WO = 1
∴ RW = 2
1 2 2 2
Also, 10 = ⇒ 10 = ⇒ 4−k = ⇒ K = 4−
4−k 4−k 10 10
R
From which,
R 2
K = 4− = 4− = 3.86
10 10
Again,
k 3.86
G= = = 27.28
4 − k 4 − 3.86
∴G = −27.28
∴ The required gain (K=3.86) for design parameter WO = 1 & Q =10 is less than the gain (G = 27.28), so
gain must be reduced. For this, we need the two resisters (Ra & Rb). Sampling by replacing R1 so, that,
Ra =G =27.28
87
G 27.28
Rb = = = 1.04
G − 1 27.28 − 1
Again, We know,
K = 1+RB/RA
For, RA = 1
RB = k-1 + 3.86 – 1 = 2.86
∴ The required final ckt will be:-
1/2
V1 27.28
-
+ V2
1 1/2
1.04
1 2.86
Now,
New frequency = 1000HZ = Ώ0
Ω0
∴ Frequency scaling factor kf = = 1000
w0
∴ Cnew = cold/kf
Or, Cnew = 1/1000 = 1mf
C1new = C2new = 1mf
Now for upper band and lower band frequency Ώ0 = Q/BW
⇒ BW = Q/ Ώ0 = 10/1000 = 0.01
Ώ1 = Ώ0 – BW/2 = 1000-0.01/2 = 999.999HZ
Ώ2 = Ώ0 + Bw/2 = 1000+0.01/2 = 1000.005HZ
Question:-.1 Design a 4th order butterworth active Sallen key law pass filter with unity gain. Realise it with
practical components.
Question:-.2 Design a 5th order butterworth active Sallen key law pass filter with equal feed back resistance
and equal capacitance Values. Then use, w0 = 2 π 1000rad/sec and C = 1µF .
Question:-.3 Design a 4th order buttterworth active Sallen Key low pass filter with equal element design.
Multiple feed back Biquad current:-
R2 C2
-
V1 R1 R2 + V2
C1
Fig(i) low pass MFB
88
1
−
v R1R3c1c2
T(s) = 2 = ………(i)
v1 1 1 1 1  1
s +  +
2
+ s +
c1  R1 R2 R3  R2 R3c1c2
In standard form,
− Gw0
2
T (s ) = ……….(ii)
 w0 
s +   s + w0
2 2
Q
Comparing eqn (i) & (ii)
1
w0 =
2
R2 R3c1c2
1
⇒ w0 =
R2 R3c1c2
w0 1  1 1 1 
And, =  + + 
Q c1  R1 R2 R3 
and, G = R2/R1
Equating (i) can be modified in the form,
T (s ) =
v2 Gb0
=− 2 ………..(iii)
v1 s + b1s + b0
Where, b0 = W02
Design in terms of bo & b1

Let, C1 = 1F
1
⇒ bo =
R 2 R3 C 2
R2
⇒G=
R1
1 1 1
& b1 = + +
R1 R2 R3
Solving,
−1
 2C 2 bo 
R2 =  
 b1 + b1 2 − 4C 2 bo (1 + G ) 
 
Example: 01
Design a biquad ckt for G = 5, b1 = 1.2 & bO =1
89
Here, given,
G=5
b1 = 1.2
bO =1
bo = 1 ⇒ WO = 1
& we know,
Wo
b1 =
Q
WO 1
⇒Q= = = 0.833
b1 1.2
Also,
R2
=5
R1
Let, C1= 1F (Choose higher)
C2 = 0.05F (Choose lower Value)
For which,
−1
 2 × 0.05 × 1 
R2 =  
1.2 + 1.2 − 4 × 0.05 × 1(1 + 5) 
2
=16.89
≈ 17
∴ R2 = 17
But,
R2
5=
R1
R2 17
⇒ R1 = =
5 5
∴ R1 = 3.4
Again,
1
bo =
R2 R3C2
1
Or, R3 =
bo R2C2
1
=
1 × 17 × 0.05
R3 = 1.18
∴ The final design will be
90
16.89 0.05
-
V1 3.4 1.18 + V2
1
It is to be noted that,
11 1 1   1 1 1 
 + +  = 1 + +  = 1.2 = b1
C1  R1 R2 R3   3.4 16.89 1.18 
High pass MFB:-
C2 R2
C3
-
V1 + V2
R1
In this case,
C1 2
.S
C2
T(S) =
C + C2 + C3 1
S2 + 1 +
R2C2C3 R1R2C2C3
The standard form is,
− GS 2 − GS 2
T(S) = = 2 …………….. (ii)
 WO  S + b1 s + bO
S + 
2
.S + WO 2
 Q 
With, C1 = C3 = 1F
G = C1 /C2 = 1/C2
∴ C2 = 1/G
Also,
1
bO =
R1 R2 C 2 C3
C1 + C 2 + C 3
b1 =
R2 C 2 C 3
Now, bo can rewritten as,
1
2+
b1 = G
R2
G
 1
 2 + S
= R2 = 
R2 b1 G
G b1
91
 2G + 1 
R2 =   …………………… (iV)
 b 
Similarly,
1 1 G
bO = = =
Rl R2 C 2 C 3
R1
(2G + 1) . 1 .1 R1 (2G + 1)
b1 G
Gb1
∴ R1 = ………..(V)
b0 (2G + 1)
Equation (iii), (iV) & (V) show that the component Value (with C1 = C2 = 1 ) can be adjusted from the
design parameters G, b0 and b2
92
Band pass MFB:-
C2 R2
-
V1 + V2
R1 C1
1
− .s
R1c 2
T(s) =
 1 1  1
s 2 +  +  s +
 R2 c1 R2 c 2  R1 R2 c1c 2
Where,
1
w0 = = b0
2
R1 R2 c1c 2
w0 1 1 R1 R2
= + = b0 ⇒ Q =
Q R2 c1 R2 c 2 c2 c
+ 1
c1 c2
R2 c1
G=
R1 (c1 + c 2 )
w0 w0
Q= =
Bw w2 − w1
Bw
Where, w1 = w0 −
2
Bw
And w2 = w0 +
2
* Band step filter for MFB (cfkm} ug]{)
93
KHN Biquad (kerwin – Huelsman-Newcomb):-

The general KHN Biquad ckt is given by;
R1
Rf C C
-
R R
R2 - -
+ + + V2
V1
R3
Fig (i) universal KHN-Biquid

If we consider only high pass, the ckt will be as follows: Rf
V1 R1
Fig (ii) high passes KHN Biquad -
+ V HP
The o/p for fig (ii) will be, V2 R2
V3
R3  Rf  R2  Rf  w0    w  2 R3
VHP = 1 + v1 + 1 +  − VHP  − RF   0  VHP 
R2 + R3  R1  R2 + R3  R1  s   s  
 
 1 w0 1  1 
2
 w0 
2

sin ce, VBP = − .VHP − VHP & VLP = − V BP =  −  VHP =  −  VHP 
 Rcs s Rcs  Rcs   s  
Thus in general, blocks,
VHP -Wo VHP (-W o)2V HP

S S
High pass Band pass Low pass

Filter (V HP) filter(VBP) Filter(V LP)
But the standard form of KHN is giVen by, (for the 1st stage)
2
1 w0 w 
VHP = kv1 − VHP −  0  V HP
Q s  s 
Where,
Rf/R1 = 1
R3/R2 = 2Q-1
K = 2- 1/Q
W0 = 1/Rc
KHN Biquad ckt is also called universal biquad ckt because from its Various stages as shown in fig (i), low
pass prototype, high pass prototype and band pass prototype can be achieve from a single ckt.
94
Chapter:-9
Sensitivity:-
Let us consider the following two fig (i) & (ii) with
L2
L1 = 0.9956H
C1 = 0.91097F
R1 = 1Ω, - C1 R1
V1 +
K=1Ω
Fig:- (i)
R3 = 1.0143 Ω
R4 = 8.9422 Ω
C2 = 0.1F R3 R4 -
C3 =1F
V1 + V2
Fig:- (ii)
1.10251
The transfer function of both the ckts are same which is T(s) =
s + 1.09735 + 1.10251
2
At W = 0, i.e. T(yo) = 1 which indicates that both gives the Butterworth response.
Now, let us assume that all elements are increased by 1%
For passive |T(y0)| = 0.99168 → gain reduced 0.83%
& for-active, |T (y0)| = 0.98308 → gain reduced 1.7%
Thus, it proves that passive filters are less sensitive to element changing than active filters.
Definition of sensitivity:-
If x is the element & y is the design parameter for example R may be element & w0 may be the
∆y
x ∆y
y y
%changeiny y
design parameter then sensitivity is denoted by ∫
x
defined by, ∫ =
x
=
%changeinx ∆ x
= .
y ∆x
x
y
x dy
∫ = y dx
x
⇒ If S = 2 then 1% change in x, result 2% change in y.

⇒ If S = 0.1 than 1% change in x, result 0.1% change in y.
* Sensitivity is known ⇒ single parameter sensitivity
⇒ First order sensitivity
⇒ Differential sensitivity
⇒ Classical sensitivity
⇒ Bode sensitivity
* Properties of first order sensitivity:-
y y
x dky x dy
(i) ∫ =
y dx ∫x
= =
x
ky dx
95
ky y
∴∫ = ∫ Where, k = constant
x x
x d (k + y )
y+k
x dy x dy y
(ii) ∫
x
=
y+k dx
= =
y + k dx y dx y + k
y+k y
y
∴ ∫ x
=
y + k ∫x
1
y
dy 1 dy 1 dy
(iii) ∫ = x. = . = . −1
y  1  xy d ( x )−1
xy dx
1 d  .dx
x
 x dx
y
1 dy 1 1 dy x
Or, ∫ = . − 2 = − −1
xy dx − x x y dx
=−
dy
1 y.
x dx
y y
∴ ∫ = −∫
1 x
x
1
y y
(iV) ∫ = − ∫
x x
y1 y 2 y1 y2
(V) ∫ x
= ∫+
x
∫
x
y y
1
(Vi) ∫ = ∫
xn
nx
y1
y2 y1 y2
(Vii) ∫= ∫− ∫
x x x
ln y y
1
(ix) ∫x
=
ln y ∫x
exp ( y ) y
(x) ∫ x
= y∫
x
* Derive all the properties
Sensitivity of passive ckt (Biquad):-
V2
96
We know,
2
Gw0
T(s) =
w 
s 2 +  0  s + w0
2
Q
We need to find out,
w w Q
SR 0 , SL 0 , SR , S Q L , S Qc
from figure,
1
T (s ) = Lc
R 1
s2 + +
Ls Lc
1 1
1 1 − −
Comparing, w0 = = 1 1
= L .C 2 2
Lc 2 2
LC
w0 R
Or, =
Q L
1 1
− − 1 1
L.w0 L.L .C 2 2 −
⇒Q= = = L2 C 2 R −1
R R
R d (w0 ) d  − 2 − 2 
1 1
R
= =0
w
(1) S R 0 = . . L C
w0 dR −
1
2
−
1
2
dR  
L C
 1 −1 
d  L2 C 2 R −1 
R d (Q )
= 1 1 −1 .   = −1
R
= .
Q
(2) S R
Q dR − R dR
L2 C 2
1
(3) S W0 Q = −
2
1
(4) S Q L =
2
1 1
=− (6) Sc = −
w0 Q
(5) Sc
2 2
Sensitivity of active Biquad ckt:-
(1) Two Thomas Biquid
In this case,
1 −
1
−
1
−
1
−
1
w0 = = R2 2 R4 2 c1 2 c2 2
R2 R4c1c2
2
R1 c1 −
1
−
1 1
−
1
Q= = R1R2 2 R4 2 c1 2 c2 2
R2 R4 R 2
−1
G = R2 R3
−1
1) S RlWO =
2
97
−1
2) S RW4O =
2
−1 −1
3) SCW1O = S RW2O, R4 , C1 ,C 2 =
2 2
−1
4) SCW2O =
2
Similarly,
5) S RQ1 = 1
1
6) S RQ2 = −
2
1
7) S RQ1 = −
2
1
8) SCQ1 = +
2
1
9) SCQ2 = −
2
10) S RG2 = 1
11) S RG3 = −1
1 R2 d (WO )
(1) S RW2O = − = .
2 WO d (R2 )
 −1 −1 −1 − 1 
d  R22 .R42 .C1 2 .C2 2 
= 2.  
R
WO d (R2 )
3 1 1 1
R2 1 − − − −
= . − .R2 2 .R4 2 .C1 2 .C2 2
WO 2
3 1 1 1
1 − − − −
R2 . − .R2 2 .R4 2 .C1 2 .C2 2
= 2
1 1 1 1
− − − −
R2 2 .R4 2 .C1 2 .C2 2
−1 −2
3
.R2
−1
= 2 3 =
− 2
R2 2
(2) Sallen key Biquid ckt:-
In this case,
1
WO =
R1R2C2C1
1
R1R2C1C2
Q=
1
+
1
+
(1 − K )
R1C1 R2C1 R2C2
98
RB
K =1+
RA
Equal Element design Equal capacitance and
Sensitivity 1 feedback resistance
K = 3− ……. (1)
Q (K = 2) ………… (2)
1) S RQ1 −
1
+Q
1
− +Q
2 2
2) S RQ2 +
1
−Q +
1
−Q
2 2
3) SCQ1 −
1
+ 2Q
1
+Q
2 2
4) SCQ2 1
− 2Q −
1
−Q
2 2
5) S kQ 3Q-1 2Q
6) S RQA 1-2Q -1
7) S RQB 2Q-1 1
8) S Rk A (− 2Q − 1) −
1
(3Q − 1) 2
9) S RkB (− 2Q − 1) 1
(3Q − 1) 2
High sensitive Moderately sensitive
(1) Design (1) is the simplest implementation interim of element Values out it’s disadvantage is that it is
highly.
(2) Design (2) is less sensitive them design (1) in sensitivity is achieved at the expense of large
resistance Value spread.
(3) Design (3) is the least sensitivity is achieved at the expense of large capacitor Value spread.
1
Q. Derive S RQ1 = − + Q in Sallen key equal amount design.
2
Multiparameter sensitiVity:-
Let,
Y = f (x1, x2, x3 ……….. xn)
Then,
∆y ∆xi
n y
= ∑S
y i =1 xi xi
Let, R1, R2 ………… Rn be the receptivity group and C1, C2, Cn be the capacitive group and
µ1 , µ 2 ,.............. µ n be the capacitive gains.
Then,
99
∆y  m  ∆R  n y  ∆C  k y  ∆µ
y
= ∑S  + ∑ S  + ∑ S  ………… (i)
y  i=1 Ri  R  i=1 Ci  C  i =1 µi  µ
Thus, we can define the multiparameter sensitivity as the combined effect of all the individual sensitivity
in a particular ckt and is generally express as shown in eqn (i).
100
Chapter:-10
Higher order Active filter:-
The higher order active filter, in terms of transform function, can be defined by,
am s m + am−1s m−1 + ......... + a1s + ao
T (S ) =
bn s n + bn−1s n−1 + ......... + b1 + bo
If ‘n’ is even, then, the higher order active filter in cascade realization may be expressed as:
 a2i s 2 + a1i s + aoi
n/2

T ( S ) = ∏  
i =1  b2 i s + b1i s + boi
2

Similarly, if ‘n’ is odd, then,
au s + ao1 (n+1)/ 2  a2i s 2 + a1i s + aoi 
T (S ) =
b11s + bo1
∏  b s 2 + b s + b 
 2i 1i o1 
Example 01:-
Design a low pass Butterworth active Sallen key filter with unity Voltage gain. The design filter ckt
must meet the following specification.
α max = 0.5d 3
α min = 10d 3
wp = 1000rad/sec
ws = 200rad/sec
Choose appropriate element Values so that the filter can be practically realized.
Soln:- The order of Butterworth filter is given by
 max 
log 10α − 1
n=  10 
= 4.83
=5
∴ The filter is of 5th order (i.e. n = 5)
2nd order 2nd order

1st order
Q1 = 0.5 Q2 = 0.62 Q3 = 1.62

Also w0 = 1
For given condition
wp
Ω0 =
(10)
For 1st stage:-
R
-
+
C
101
1
Here, T(s) = Rc
1
s+
Rc
∴ w0 =1
1/Rc =1
Or, Rc = 1
Let, Cnew = 0.1 µF
Again we also need to perform frequency scaling with
Ω0
Kf = = 1263.2
w0
∴Applying both magnitude and frequency scaling
cold
Cnew =
kf .km
1
Or, 0.1 µF =
1263.2 × km
⇒ km = 7616.40
Rnew = km Rold
= 7916.40 X 1 = 7.916k
2nd stage:-
-
+
In unity Voltage gain of Sallen key

W0 = 1
& R 1, R 2 = 1
C1 = 2Q
C2 = 1/2Q
Where, Q = 0.62
∴C1 = 1.24F
C2 = 0.806F
Applying magnitude and frequency scaling
Let, C1new = 0.1 µF
Or, C1new = C1 old/kmkf
⇒ Km = 9816.34
c2old 0.806
∴ C2new = = = 0.065µF = 65.03nF
kmkf 9816.34 × 1263.2
Similarly,
102
R1new = R2 new = kmRold = 9816.34 × 1 = 9.8
For 3rd stage:-
wo = 1
R1 = R2 = 1
C1 = 2Q, C2 =1/2Q
For, Q = 1.62
C1 = 2Q = 2 X 1.62 = 3.24F
C2 = 1/2Q = 1/2X1.62 = 0.308F
For let C1new = 0.1 µ F
Cold 3.24
C1new = = = 25649.14
kf .km 0.1 × 10 − 6 × 1263.2
C2old 0.308
C2 new = = = 9.51nF
km.kf 25649.14 × 1263.2
∴ R1new = R2 new = kmR1old = 25649 × 1 = 25.64k

Therefore the final ckt will be,
0.1µ F 0.1µ F
7.961k 9.8k 9.8k

- - 25.64k
-
+ + 25.64k +
65.04µF
9.51µ F
Fig: - 5th order low pass Butterworth active Sallen key Biquad with unity Voltage gain.
Ex: 02; Design a 5th order low pass Butterworth filter with F0 = 1 kHz and capacitance of
Of 0.1 µ F. Implement this ckt in MFB.
Example:- 03
implement the same in two Thomas.
Example:- 04
In some application filter ckt must meet the following specification.
α max = 0.5dB
α min = 20dB
ws
=2
wp
The design must be highest sensitivity low pass Butterworth active Sallen key filter with unity Voltage gain.
Chose appropriate element Values so that the filter can be realized.
Also calculate the sensitivity of Q with 1% incorrect in Values of filter elements due to some
reason.
103
Chapter: - 11
Simulation of passive n/w: -
Tellegen proposed a mode called gyrator where.

V1 = kI2............(i)
V2 = kI1………..(ii)
For the fig (i) shown below
I1 I2
IL
+ +
V1 Greater V2 ZL
- -
Here, k is a real constant

Now, let ZL be the load, then,
V2 = ZLIL
Or, V2 = -ZLKL
From equation (i) and (ii)
v 
V1 = k  2 
 ZL 
 (− kI1 ) 
Or, V1 = k  −  (from eqn (ii))
 Z L 
v1 k 2
Or, =
I1 Z L
K2
∴ Zin = = Z i1
ZL
If ZL is capacitor, then,
1
ZL=
cs
∴ Z in = k2cs = Leq .s
Where,
Leq = k2c
I1 I2
+
V1 Greater V2
-
Leq
I1 K I2
V1 V2 C
Fig:-(iii) symbol for Gaygrator
104
GIC (General Impedance Converter):-
It was developed by Antoniou, So also called Antoniou GIC.
+
-
V3 Z3 I2
Z1 Z2 Z4 V2
V1 I 1
-
+
Fig (i) General impedance converter
I1 I2
V1 V2
GIC
Fig:- (ii) symbol for Gic
V1 = V2
& I1 = -I2
From fig (i)
v1 − v3
I1 =
z1
v2 − v4
I2 =
z4
v1 − v4
= [∴ v1 = v2 ]
z4
Also, I1 = -I2
v3 − v1 v −v 
Or, = − 4 1 
z2  z3 
I1 z1 I1 z4
Or, − =
z2 z3
− z2 z4
Or, I1 = .I 2
z1 z3
We know,
v1 − z1 z3 v2
Z in = = . [∴ v1 = v2 ]
I1 z2 z4 I 2
v2
=− k
I2
Where,
105
z1 z3
k=
z2 z4
Let, ZL be the lead, then,
v2 = Z L I L = − Z L I 2
v2
Or, = −Z L
I2
Thus,
Zin = kZL
z1 z3
Z in = .Z L = Leq.S
z2 z4
Always,
ZL = R L
Z1 = R1
Z3 = R3
1
Now, if z2 = and Z4 = R4 then,
c2 s
R1R3c2
zin = .RL .s = Leq.S
R4
R1R3C4
In this case, k =
R2
− z2 z 4
Also, I1 = I2
z1 z3
1
Or, I1 = − .I 2
ks
Or, I 2 = − ksI1
⇒ − I 2 : I1 = ks : 1
Thus the representation will be
I1 I2
KS:1
RL
Leq
Gic
It is to be noted that this is the case of Grounded inductor simulation.
Example:- 01
Simulate the following ladder ckt with GIC.
106
1.618 1.618 1.618
0.618 1
+ 0.618
-
V1
Soln.
We need to simulate L2 &L4 with GIC
For, L2
We know that,
L2eq = KRL
Let, Z2 =1/c2s
& Z1 = R1
Z3 = R3
Z4 = R4
If R1 = R2 = R4 = 1
& C2 = 1, then,
L2eq = RL
∴ RL = L2eq = 0.618
∴ RL = 0.618
Similarly, for L4
RL = 0.618
∴ The final ckt will be,
1 1.618 0.5 1.618
1 1 1
+
-
V1
1
1
1
1
0.618
0.618
Fig: Equivalent ckt with GIC.
Floating Inductor simulator:-
107
L
3 4
V2 1 2
V1 GIC GIC
1’ 2’
3’ 4’
3 4
1 KS:1 1:KS 2
V1 V2
V1’ V2 ’
1’ 2’
3’ 4’
From fig 1(a)

V1 − V2
= Leq .s ………………….. (1)
I1
From fig (a) (c),
V 1 = V 11
V2 = V21
Also,
I11 = − ksI1
I 21 = − ksI 2 ……………. (ii)
I11 = − I 21
I1 = − I 2 …………….. (iii)
From eqn (ii) & (iii)
I 21 = KSI1
Also,
V11 − V21
=R
I 21
V1 − V2
Or, =R
KSI1
V1 − V2
Or, = KRS …………..(iV)
I1
From eqn (i) & (iV),
Leq = K.R
Leq
R=
K ……………. (V)
Where,
108
R1 R2C 2 R1 R3C4
K= or ,
R4 R2
Example:- 01
Simulate the following ladder ckt with GIC.
L2 L4
R1
V1 C1 C3 C5 RL
Ladder Design with frequency dependent –Ve resistor (FDNR):-

We know,
Z1Z 3
Z in = .Z L ……………….. (i)
Z2Z4
If,
1
Z1 =
c1s
1
Z3 =
C3 S
ZL = R L
Z2 = R2
Z4 = R4
Then, eqn (i) becomes,
1 1
.
c1s c3 s
zin = .RL
R2 R4
RL
=
R2 R4c1c3 s 2
1
∴ zin = …………(ii)
Ds 2
Where,
R2 R4c1c3
D=
RL
Put, s = jw
−1
Then, zin = ………(iii)
Dw2
Equation (iii) define FDNR
Process,
Let us consider the following simple RLC ckt.
109
L
R
C
+
-
V1
We scale this ckt by 1/s i.e

R R.1/S 1/R
L LS.1/S L
1/CS . 1/S C=D

C
∴ The final ckt will be,

1/R L 1/R
C1
C=D R2
+-
R4
RL
Here,
R2 R4c1c3
D=
RL
If, R2 = R4 = 1 & c3 = c3 = 1
Then,
1 1
RL = =
D c
Example:-02
R1
C3 C5 RL
+
-
V1
Realize the above ckt
110
Leapfrog simulation of ladders:-
Consider a ladder circuit,
L2 L4 L6
R1
C3 C5 RL
+
-
V1 V2
I1
I5
+I 3 +
V1 Y1 Y3 Y5
+
- ZV2 V2 Z4
V4
Z6
V6
- -
Fig :-1(b)
From figure, (b)

I1 = y1 (V1-V2)
V2 = Z2 (I1-I3)
I3 = y3 (V2-V4) __________(i)
V4 = Z4 (I3-I5)
I5 = y5 (V4-V6)
V6 = Z6I5
Recall that,
Y = Gx can be represented in the form
(2) Y = X1+X2
+ or, X1
Y
X Y
X2
X2
(3) y = m1x1+m2x2
m1
m1 + or, X1
X1 Y
Y
m2
X2
m2
X2
111
Modifying eqn (i) set,
I1 = y1 (V1+ (-V2))
V2 = Z2 (I1+ (-I2))
I3 = y3 (V2 + (-V4))
V4 = Z4 (I3+ (-I5)) ………….(ii)
I5 = y5 (V4+ (-V6))
V6 = Z6I5
+ y1 I1
V1
+ Z2 V2
I1
I1 + Y3 I3
-V4
I1 + Z4 V4
V4 + Y5 I5
-V6
I5 Z6 V6
-1 -1 -1
V1 + y1 + z2 + y3 + z4 y5 z6 v6
V2 V4 +
-1 -1
112
In active ckt, we can not realize the current, so we replace all the currents by their respective Voltages.
i.e, we replace,
I → V1
Y → Ty
Z → TZ
From eqn (iii)
= (v1 − v2 )
I1 y1
y y
or, VT 1 = Ty1 (v1 − v2 )
Also, v2 = z2 (I1 − I 3 )
=
z2
(I1 − I 3 )z
z1
v2 = Tz2 (vI 1 − vI 3 )
Thus eqn (ii) can be rewritten as,
VI1 = Ty1 (V1-V2)
V2 = TZ2 (VI1-VI3)
VI3 = Ty3 (V2-V4)
V4 = TZ4 (VI3-VI5) ……….(iii)
VI5 = Ty5 (V4-V6)
V6 = TZ6VI5
-1 -1 -1
V1 + Ty1 + Tz2 + Ty3 + Tz4 Tz6

V4 +
V2 Ty5 v6
-1 -1
Modifying eqn set (iii)

-VI1 = -Ty1 (V1-V2)
-V2 = TZ2 (-VI1+VI3)
-VI3 = -Ty3 (-V2+V4)
V4 = TZ4 (VI3-VI5) ……….(iv)
-VI5 = -Ty5 (V4-V6)
-V6 = TZ6 (-VI5)
113
-1
V1 + Ty1 + Tz2 + Ty 3 + Tz4 Tz 6 v6

-VI1 -V2 VI3 V4 + Ty5
VI5
Certain important op-Amp configuration:-

R
(1) R
(2) R
V1
R R
- -
V1 + V2 R 1.18 + V3
V2
(3) C (4) R1
R
-
V1 + V2 R C2
-
V1 + V2
Example:-:01
1
Design a 4th order low pass Butterworth filter with power frequency of 104 rad/sec. The filter must be
2
implemented on leapfrog active filter simulation.
0.7654 1.848
1
y1 y3
+
- z2 1.848
z4 1
0.7654
Soln:-
Iy1 Tz2 Ty3 Tz1
From fig (i)

Z1 = R1+L1S
114
1
1 L1
∴ y1 = =
R1 + L1S s + R1
L1
1
z2 =
c2 s
1
y3 =
L3s
1
1 RL c4
y4 = c4 + ⇒ z4 = =
RL RL c4 s + 1 s + 1
c4 RL
RL c4 s + 1
=
RL
Now,
1 −1
L1 L1
y1 = ⇒ −Ty1 =
R R
s+ 1 s+ 1
L1 L1
 −1 
= (− 1)
1 1
z2 = ⇒ Tz2 = 
c2 s c2 s  c2 s 
1 1
y3 = ⇒ −Ty3 = −
L3 s L3s
1  1 
 − 
c4 c4
z4 = ⇒ Tz4 = −1 
1  1 
s+  s+ 
c4 RL  c4 RL 
-
+ R1
- - C2
+ - -
+ + -
+ +
Ty3 Ty4
Tz2
-Ty1
Comparison
115
1 1
−
L1 R1c2
⇒ −Ty1 = ≡ [R1 = R2 = 1]
RL 1
s+ s+
L1 R2c2
 1 
 = (− 1) −
1 
Tz2 =
1
≡ (− 1) −  [R = 1]
c2 s  c2 s   Rcs 
1  1 
− Ty3 = − ≡ −  [R = 1]
L3 s  Rcs 
 1   1 
   − 
Yz4 = (− 1)  ≡ (− 1) R1c2
c4  [R1 = R2 = 1]
 1   1 
s+ c R  s+ Rc 
 4 L   2 2 
Now, for frequency and magnitude scaling

W0 = 1rad/sec
Ω 0 = 104rad/sec
∴ kf = 104
Let, km =104
Now scale.
In summary, leapfrog simulation can be done in the following steps.
(1) Choose a suitable low pass prototype which meets the following specifications.(see table 13.1 of van
valkenbutg)
(2) Perform freq transformation if necessary.
(3) Identify the various y & z in the form of block diagram.
(4) Select the leapfrog block diagram to simulate the ckt.
(5) Find the active ckt that realize each of the blocks.
(6) Arrange the ckt with necessary components.
(7) Scale the ckt to meet the actual requirements.
116
Chapter:-12
The Mos switch
S-
Preferred for voice frequency filters

If VGS<VT, switch off, behave as open ckt between S & D.
If VGS>VT, switch on, behave as short ckt between S & D.
100M
OFF
10k
ON
ON OFF
TC
TC is the time period.

1
fc = = Switching frequency
TC
The equivalent representation for such case is
S O
Fig:- SPST Switch
(single pole single throw switch)
Again, consider the two phase clock.
117
TC
Note that, the frequency do not overlap and when φ1 , is OFF, φ2 is ON, and vice versa.
The representation for these cases will be
P1 P2
1’ 2’
1 2
1’ 2’
Fig:- SPDT( SIngle pole double through switch)
DPDT (Double pole Double through) switch:-

1 2
1 2
1 2
1’ 2’
1’ 2’
2’
1’
Let us consider the following specification ckt for DPDT switch.
118
a b
1 2
C
Vc
V1
1’ 2’ V2
The equivalent ckt for S/W is

R
When S/W is brought to position b,

V2 = -VC
V2 =-V1
This show that DPDT S/w acts as an inverter
Q. How can you use DPDT as an Inverter?
Simulation of resistor by switched capacitor

1+ +2
V1 V2
1’ - 2’
Let us assume SPDT switch as shown in fig (i) let V1 (t) be the i/p voltage, if s/w is at
position ‘a’ then the eqn ckt will be:
R1 a b
CR
The capacitor will get charge for

τ 1 = R1CR
V1
0.63V1
V2
I1
Now let us more the switch to position ‘b’.
The charge transferred will be
q = CR (V1-V2)
∴ The current in this is
119
i (t ) = = CR (v1 − v2 )
q
t
C R (v1 − v2 ) (v1 − v2 )
=
Tc Rc
1
∴ RC =
fcCR
Switched capacitor for op-amp based angle operation:-

(1) Integrator
C C
R R
- -
V1 + V2 V1 + V2
From fig
1
R=
fxC R
v2 1 C
∴ =− = − fc. R .s
v1 1 c
.c.s
fcCR
∴ T (s ) = − fc.
CR
.s
c
(2) loosy Integrator
R1
C3
R C2
- C2
V2 R
V1 + -
V1 + V2
Here,
1 1
R1 = & R3 =
fcc1 fcc3
c1
− fc
v c2
∴ 2 =
v1 s + fc c3
c2
(3) Adder Integrator
120
C2
C2
R0
V0 V0 C0
-
R1 + V2 -
V1 + V2
V1
C1
−1 1
v2 = v0 − .v1
R0c2 s R1c2 s
c0 1 c 1
= − fc . .v0 − fc 1 . .v1
c2 s c2 s
(4) Inverting
Rf Cf
R1 C1
-
V1 + V2 V1 -
+ V2
1 1
−
v2 Rf fccf c v2 cfs c1 v2 c
=− = =− 1 = = ∴ =− 1
v1 R1 − 1 cf v1 1 cf v1 cF
fcc1 c1s
(5)
R1 C2
V0 R1
- R1
-
R2 + V2 -V 1 + V2
V1
R1
 1 
⇒ v2 = (− v1 ). −
v2 1
= 
v1 R1c2 s  R c s
1 2 
Thus, the ckt using switched capacitor will be,
C2
-
C1 + V2
V1
v2 c
= fc. 1 .s
v1 c2
(6)
121
C2
c1
R1 V1 -
V1 -
V0
V0 C1 +
V2 +
R1 c2 V2
v2
=
1
(v1 − v2 )
v1 R1c2 s
Example:-01
Realize the given circuit by switched capacitor.
C 2= 0.10µ F
R1=10k
-
V1 + V2
C 2= 0.10µ F C2= 0.10 µF
R1=10k R 1=10k
- -
V1 + V2 V1 + V2
v2 1
=−
v1 R1c2 s
1
But, R1 =
fcc1
v2 c 1
∴ = − fc. 1 .
v1 c2 s
⇒ τ = R1c2 = 10 × 103 × 0.01 × 10 −6 = 10 −4
1 c1
or, R1c2 = . = 10 − 4
fc c2
1 c1 − 4
∴ . 10
fc c2
Let, fc = 10 kHz
1 c1
∴ , = 10 − 4
10 × 10 0.01 × 10 − 6
3
c1 = 0.01µF
First order filter:-
122
V1
V1 V2 C 2 V2
C2 C1
1 c1
fc
T (s ) = T (s ) =
R1c2 c2
1 c1
s+ s + fc
R1c2 c2
R4
C4
R3
C2
V1 - R C2
+ V2 C3 -
C1 + V2
C1
 1   c 
s+   s + fc. 3 
T (s ) = − 1  T (s ) = − 1 
c R3c1  c c1 
c2  s + 1  c2  s + fc. c4 
 R4c2   c2 
  
 1 1 
sin ce : R3 = & R4 = 
 fcc3 fcc4 
123
Example:- 01
Design a switched capacitor filter from the following plot.
3 4 5
102 10 10 10
w
-20dB
From the plot,
T (s ) =
(s + a )(s + b )
( s + c)(s + d )
=
(s + 10 )(s + 10 )
3 4
(s + 10 )(s + 10 )
2 5
 (s + 10 )3    s + 104 
= −   −
2   

5 
 s + 10    s + 10 
= T1 (s ) T2 (s )
for,
T 1(s ) = −
(s + 10 ) ……….(i)
3
(s + 10 )
2
Comparing eqn (i) with

 c 
 s + fc 4 
T (s ) = − 1 
c c1 
c2  s + fc c4 
 c1 

Take, fc = 10 kHz
C1 = C2 =10 pf
∴ C3 = 1 pf
∴ C4 = 0.1pf
For,
− (s + 104 )
T2(s) =
(s + 105 )
Comparing,
 c 
 s + fc 3 
−c
T (s ) = 1 
c1 
c2  s + fc c4 
 c1 

C1 = C2 =10 pf
∴ C3 = 10 pf
C4 = 100 pf
124
C4’
C2 R C2’
R - -
C3’ + V2
V1 +
V2
C1 C1’
Example:-02
dB Gain
6dB
0dB
w
120 220 400
T (s ) =
(s + 100)(s + 400)
(s + 200)2
125

Filter Desing Complete

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Filter Desing Complete

Uploaded by

Copyright:

Available Formats

Downloaded from www.jayaram.com.

Gain and Attenuation:

Then the voltage gain in db is given by ,

Av = 20log10 T ( jw) dB …………….(1)

Or in term of power , the power gain is given by,

Now, the voltage attenuation is given by ,

α = -20log T ( jw) dB…………….(2)

From equation 1 and 2 ,we can write,

Types of filters: ( According to the function)

Non- ideal Characteristics:

3. Band between PB and SB so defined are known as transition bands. (TB).

When , m = n = 1, then the T(s) of equation (i) will be bilinear , i.e

Here, G = a1/b1 = Gain

Realisation of filter with passive elements:

Taking laplace transform of equation (i) and (ii)

Fig. 2. Magnitude plot

For phase plot:

θ (jw) = tan-1(o/w0) – tan-1(w/wo)

Above figure can be modified as:

For magnitude plot ,

θ (jw) = tan-1(-w/wo) - tan-1( w/wo)

From fig (i)

ω > ω 01 -wo1 wo2

For Magnitude plot:

For Phase plot,

Pole plot (θp)

45 pole zero plot

Insertion Gain and insertion loss:

Normalization and Renormalization:

1. Impedance (Magnitude or amplitude) scaling: In this scaling, the magnitude of the

If Km . 1, then it is called scale up.

Let, Rold = old value of Resistor.

The new values of R, L and C are given by

Rnew = Km Rold ………….(i)

Note: Generally we assume new value of capacitor 1µF or 10 µF.

Fig(ii) scaled ckt.

The transfer function for fig. (i) ,

The resistance is unaffected by frequency scaling , i.e

Or, XL = j(wkf) ( Lold/kf) since, L = Lold

Lold = Lold / Kf ………….(vi)

3. Both magnitude and Frequency scaling:

Rnew = KmRold …………….(Viii)

These three equations are know as element scaling equations.

Fig (ii): after frequency scaling.

Perform frequency scaling with Ωo = 1 Ω

Perform magnitude and frequency scaling separately with wo = 3 and Ω0 = 300.

Again for frequency scaling,

Perform magnitude scaling to the ckt given.

One port and two port passive network:

I(s) I1(s) I2(s)

V(s) 1- port V1(s) 2- port

One port network: Let us suppose of fig of 1(a),

Date: 2065/ 5/10

Properties of Passive n/w.

Design of LC Ckt . (Loss less ckt):

Therefore from equation (i) and (ii).

(ii) Z(s) = z(s2+1)(s2+9)/s(s2+4)

In this case , F(s) = Y(s) , then equation (i) becomes

The ckt can be realize as :

z(s) C1 =16/3 F C2 = 16/5 F

Example 02: Design Foster parallel n/w for the function

Fig. Foster’s parallel n/w of LC ckt.

Continued Fraction method or cauer method for LC Ckt