Voltage-Control Voltage-Source (VCVS)

Low-Pass Filters
A general filter circuit which may be used
to obtain second order low-pass, high pass Y3

and band-pass active filters is shown in Y1 Y2

b
a
A
Figure 9.
Y4
The circuit is called a voltage controlled
VI
VO
voltage source (VCVS) filter because of the
presence of the ideal op amp VCVS with
gain A>0. This type of filter was first
Figure 1 – Generalized 2nd Order Filter
introduced by Sallen and Key in 1955.
For the circuit shown, the node equations at (a) and (b) are given by
( ) …………… (23)

( ) …………… (24)

where and there is no current into the op amp. Eliminating VA and solving for the

transfer function yields

( )( )
…………… (25)

To obtain a low-pass filter Y1 and Y2 must C2

be constants and thus for a quadratic R1 R2

denominator, Y3 and Y4 must represent

capacitors.
R4
4
VI C1 VO
Hence,
R3

giving Figure 2 – Low-Pass Filter

( )
( )( ) ( )

…………… (26)
( )

17
For a low-pass second order filter, using a non-inverting gain, the transfer function is

…………… (27)

where k is the gain and b0 and b1 determine the type of low-pass filter (Butterworth or
Chebyshev). Comparing this T.F with that above, gives the following coefficients:
…………… (28)

( ) ( )…………… (29)

…………… (30)

Butterworth values R3 and R4 are arbitrary except that their ratio must be fixed. However, for
best operation they should be chosen to minimize the DC offset of the op amp. To achieve this,
the differential voltage into the op amp must be minimized and for this,
…………… (31)
For this circuit, there must be no capacitive coupling at the input since there would be no DC
return to ground from the non-inverting input. Such a path is necessary to provide input bias
current. If A=1,

( )
…………… (32).
( )

( ) ( )
…………… (33)

where s=jω and cut-off frequency, ωc=1. This yields eqns. (28), (29) and (30). The transfer
function of a low-pass 2nd order filter with cut-off frequency, ωc is given by

( ) ( )
…………… (34)

where s=jω and cut-off frequency, ωc=ω0. This yields

…………… (35)

( ) ( )…………… (36)

…………… (37)

A straightforward design procedure is then to let ω0=1, and design the filter. The resulting
capacitance values C1 and C2 must then be scaled down to and to move the filter cut-off

frequency from ωc=1 to ωc=ω0. (Though in practice, it is the capacitor values that are scaled,

18
from the equations above, it is clear that the shifting of ωc can also be achieved by scaling R1
and R2).
Once b0, b1 and A are determined, there remain 6 unknowns and therefore certain components
will have to be chosen since a unique solution is not possible. Thus, if C2=1 then, with ω0=1
…………… (38)

( ) ( )…………… (39)

Design Procedure
We can now lay down a design procedure for the VCVS 2nd order low-pass filter.
1. Select C2=1F.
2. For the type of filter (Butterworth or Chebyshev) and the specific filter requirements find the
constants b0 and b1 from respective table. The constant A=k of the active filter circuit, which
represents the circuit gain must be known or chosen depending on the active filter circuit
application. Also, the cut-off frequency must be known and if the active filter is a
Chebyshev filter, the ripple wave factor RWdB must be known.
3. Determine C1, R1 and R2 from eqns. (38) and (39).
From (16),
…………… (40)

Sub (40) in (39) yields

( )
( )…………… (41)

Solving for R2 we get

( )
( ) …………… (42)

which yields

√ ( )
…………… (43)

From (40), R2 can have two values. For feasible design R2 has to be real, hence the
following condition must be met

…………… (44)

19
 C1 must be selected such that eqn. (44) is satisfied which would yield a positive value under
the radical in eqn. (43). R1 is then determined from eqn. (40).
4. Next we calculate R3 and R4 for gain k and minimum offset voltage. From before, for min.
offset,
…………… (45a)
( )
…………… (45b)

Hence, from eqn. (45a),

( )…………… (46)
and from (45b),
( )
…………… (47)

where k≠1. If k=1, then R4=R1+R2 and R3=∞ i.e. an open circuit.
5. Finally, we denormalize the capacitance values by dividing these values by ω0. Additionally,
it can be seen from eqns. (28) to (30) and eqns. (35) to (37) that the active filter will retain
its response if all the register values are multiplied by an Impedance Scaling Factor, ISF, and
at the same time all the capacitors are divided by the ISF. This step is useful if capacitor
values are too large and at the same time the resistor values are too small. Hence
Resistors →Resistors ×ISF
Capacitors→ Capacitors÷ (ω0×ISF)
ISF=fc÷20π; Rule of Thumb
Consider the T.F. eqn. (34) for a low-pass 2nd order filter:

( ) ( )
For k=1 and using a 2nd order Butterworth response,

( ) ( )
…………… (48)
√ ( ) √ ( )

In the T.F. eqn. (32), when k=A=1 then

( )
Now let R1=R2=R and C2=2C1=2C, then

20
Remembering s=jω

( ) ( )
…………… (49)

Let , then
√

…………… (50)
( ) √ ( )

It is clearly seen that eqns. (48) and (50) are identical. Hence, a simplified design procedure for
a unity gain 2nd order low-pass Butterworth filter is as follows:
1. Choose the cut-off frequency ω0 or f0.
2. Choose C1=C, usually a convenient value between 100pF and 0.1µF.
3. Determine C2 from C2= 2C1=2C.
4. Calculate .
√

5. Choose R1=R2=R and from above, R4= R1+R2=2R.

Examples

1. Design a VCVS 2nd order low-pass Butterworth filter having unity gain cut-off frequency of
105 rad/s. (f0=15915Hz)
Solution

a. (using simplified procedure)

Choose C1=C=0.01µF and hence C2=2C=0.02µF.
Determine
√
Hence R1=R2=707Ω and R4= R1+R2=2R=1.4kΩ.
b. (Using Full Design Procedure)
Select C2=1F
2nd Order Butterworth, therefore b0=1 and b1=√2.

Choose C1=0.5F.
√ ( ) √ ( )
√

R4= R1+R2=2.828Ω and R3=∞.

To denormalize capacitance values,

21
 C2=0.01µF

R1=707Ω R2=707Ω

Using ISF = 500,

R4=1.4 k Ω
VI
VO
C1=0.01µF

Figure 3 – Low-Pass Filter design by Example 1

2. Design a VCVS 2nd order low-pass Chebyshev filter having a gain of 2, a cut-off frequency
of 10 kHz and a ripple width RWdB=0.5dB.
Solution

Select C2=1F for convenience

2nd order Chebyshev with RWdB=0.5dB, b0=1.516 and b1=1.426

Choose C1=1F
√ ( ) √ ( )

( ) ( )
( ) ( )

To denormalize capacitance values

R1=701Ω R2=941Ω

Using ISF = 103,

R4=3284 Ω
VI
VO
C1=0.015µF

R3=3284 Ω

Figure 4 – Low-Pass Filter design by Example 2

22
Higher Order (n>2) Low-Pass Filters
A higher order active filter is one in which the transfer function is greater than two (n>2). That is
the transfer function G(s) is one whose numerator and/or denominator polynomials are of degree
greater than two. We shall consider the case of a denominator with degree n>2 and numerator
with degree m≤n.
There are two general methods of obtaining higher order transfer functions. One way is to
attempt to find a single circuit which realizes the function. The second general method is to
factor the transfer function G(s) into first and second order functions G1 ( s) and G2 ( s) , then
realize each factor with a sub-network and cascade the sub-networks to obtain the overall
network.
The cascaded network has the advantage that the mathematics is relatively simple requiring only
the factoring procedure and the use of familiar sub-networks with which to match the
coefficients. Another advantage is that each sub-network can be tuned separately to the required
Gi (s) transfer function. The primary disadvantage is the sensitivity of the filters to component
changes.
In this section, the cascaded network approach to higher order filters will be treated.

First Order Filter

Considering first the 1st order filter shown in Figure 13
the transfer function is easily shown to be K

…………… (51) VI C VO

C2
The general first-order transfer function for a low-pass =0
.0
filter is given by 15
µ st
…………… (52) FFigure 5 – 1 Order Low-Pass Filter

Where ω0 is the cut-off frequency. Comparing (51) and (52), it follows that
…………… (53)

…………… (54)
The transfer function for the circuit shown in Figure 14 is

23
 C

…………… (55) R2

R1
Hence
…………… (56)
VI
…………… (57) VO

In an actual design, equations (53) and (56) may be

used directly on ω0 set to 1 after which the values Figure 6 – 1st Order Low-Pass Filter

are denormalized.
The approach employed here to synthesizing higher-order filters is to cascade first and/or second
order filters approximately. Thus if a 4th order filter is required, 2 second-order filters can be
cascaded. If an odd-order filter is required, at least one first-order filter must be cascaded with
one or more second–order filters. For example, to achieve a 5th order filter, two second-order
filters must be cascaded with a single 1st order active filter circuit.
C1

R2 C3
C5
R1
R3 R4
R5 R6

VI
RA
C2 RB
C4 VO

Figure 7 – 5th Order Low-Pass Filter

Example

Design a 5th order low-pass Butterworth active filter with a gain of -10 at a cut-off frequency of
10 kHz.
Solution

1. ( ) ( ) ( ) ( )

From table 7.2 for n=5: b0'=1, b1''=0.6180, b0''=1, b1'''=1.6180 and b0'''=1.

2. We will now find the passive components that fit the 1st order T.F given by ( ) ,

where and .

24
 For convenience, choose C1=1F. Then and , where k0=10.

Denormalizing, we get

Using ISF=104,

3. We now turn to the 1st of the two 2nd order sections. ( ) . From the procedure

developed for 2nd order filters, choose C3=1F for convenience. Then, with k1=1,

Hence choose C2=0.05F. Then

√ ( ) √ ( )

Denormalizing gives

Using ISF=103,

4. We repeat the steps of 3 to synthesize the 2nd order T.F. ( ) . Choose C5=1F

and k2=1.

Hence choose C4=0.5F. Then

√ ( ) √ ( )

25
 Denormalizing gives

Using ISF=103,

Exercise

Repeat for 0.1 dB Chebyshev.

Low-Pass to High-Pass Transformation

As previously discussed, an all pole normalized low pass function is given by the general T.F.
( ) …………… (58)

Where bi determines the type of filter and k is the gain. The transformation yields

( ) ( )

( ) …………… (59)

where , i=0,1,2…n and bn=1.

The T.F. in (59) corresponds to an nth-order high-pass filter. Practically this transformation from
low-pass to high-pass requires the replacement of filter resistors by capacitors, and the
replacement of circuit capacitors by resistors.

26