EE 508

Lecture 17

Basic Biquadratic Active Filters

Second-order Bandpass
Second-order Lowpass
Effects of Op Amp on Filter Performance
Review from Last Time
Standard LP to BS Transformation

map ω=0 to ω = ±∞
map ω=0 to ω = 0
map ω=1 to ω = ωA
map ω=1 to ω= -ωB T  j 
map ω= –1 to ω= ωB
map ω= –1 to ω= -ωA 1

-1 1 ω

T  j 
BS
ω=∞
ω=-∞ 1

ω
-1 -ω0 ω0 1
 Review from Last Time
LP to BS Transformation
Pole Q of BS Approximations T  j  T  j 
BP

1 1
BWN

Im 1 ω ωAN ωBN ω
1

ω0BPH,QLBPH BW = ωBN - ω AN

ω0BPL ,QLBPL  ωM  ω ANωBN

Im
 BW 
ω0LP ,QLP  Define:   
ω ω
 M 0LP 
Re Re
It can be shown that
2
Q 4  4  4
QBSL  QBSH  LP 1  2  1  2   2 2
2      QLP
For γ small, 2QLP
QBS 

It can be shown that
  QBS 
2 
ωM  QBS
ω0BS       4
2  QLP  QLP  
 
Note for γ small, QBS can get very large
Review from Last Time
Standard LP to BS Transformation
s  BW
sX  2
s  ωM2

• Standard LP to BS transformation is a variable mapping transform

• Maps jω axis to jω axis in the s-plane

• Preserves basic shape of an approximation but warps frequency axis

• Order of BS approximation is double that of the LP Approximation

• Pole Q and ω0 expressions are identical to those of the LP to BP transformation

• Pole Q of BS approximation can get very large for narrow BW

• Other variable transforms exist but the standard is by far the most popular
Review from Last Time
LP to HP Transformation
(Normalized Transform)

T  j 

-1 1 ω

T  j 
HP

ω=-∞ ω=∞

ω
-1 1
Review from Last Time

Comparison of Transforms
T  j  T  j 
LP to BP 1
LPN
BP

BW
s +1
2

s
s•BW N 1 ω ωL ωH ω
ωM

T  j  T  j 
BS

LP to BS
1 1

s  BWN BWN

s 2
s 1 ωAN ωBN ω
1 ω
1

LP to HP T  j 
LP
T
HP
 j 
1
1
1
s
s
1 ω ω
1
 Filter Design
Process

Establish
Specifications
- possibly TD(s) or HD(z)
- magnitude and phase
characteristics or restrictions
- time domain requirements

Approximation
- obtain acceptable transfer
functions TA(s) or HA(z)
- possibly acceptable realizable
time-domain responses

Synthesis
- build circuit or implement algorithm
that has response close to TA(s) or
HA(z)
- actually realize TR(s) or HR(z)

Filter
 Filter Design/Synthesis
Considerations
There are many different filter architectures that can realize a
given transfer function

Considerable effort has been focused over the years on

“inventing” these architectures and on determining which is
best suited for a given application
 Filter Design/Synthesis Considerations
Most even-ordered designs today use one of the following three basic architectures
Cascaded Biquads

VIN T1(s) T2(s) Tk(s) Tm(s) VOUT

Biquad Biquad Biquad Biquad

T  s   T1T2  Tm

Leapfrog

I1(s) I2(s) I3(s) I4(s) Ik-1(s) Ik(s) VOUT

VIN
Integrator Integrator Integrator Integrator Integrator Integrator

a1 a2

Multiple-loop Feedback (less popular)

αF α1 α2 αm
α0
XIN  T1(s) T2(s) Tm(s)
XOUT
Biquad Biquad Biquad

What’s unique in all of these approaches?

 Filter Design/Synthesis Considerations
Most odd-ordered designs today use one of the following three basic architectures
Cascaded Biquads

VIN VOUT
T1(s) T2(s) Tk(s) Tm(s) Tm+1(s)
First
Biquad Biquad Biquad Biquad Order

T  s   T1T2  Tm
Leapfrog

I1(s) I2(s) I3(s) I4(s) Ik-1(s) Ik(s) VOUT

VIN
Integrator Integrator Integrator Integrator Integrator Integrator

a1 a2

Multiple-loop Feedback (less popular)

αF α1 α2 αm αm+1
α0
XIN  T1(s) T2(s) Tm(s) Tm+1(s)
First XOUT
Biquad Biquad Biquad Order

What’s unique in all of these approaches?

 Filter Design/Synthesis Considerations
What’s unique in all of these approaches?

α2
αk
Tm+1(s) α1
T(s) I(s)
First XIN
α0
 XOUT
Biquad Integrator Order
k
a2s2 +a1s+a0
T  s = 2 I s =
I0 a s+a0
T s = 1 XOUT = α i
s +b1s+b0 s s+b0 i=0

• Most effort on synthesis can focus on synthesizing these four blocks

(the summing function is often incorporated into the Biquad or Integrator)
(the first-order block is much less challenging to design than the biquad)
• Some issues associated with their interconnections

• And, in many integrated structures, the biquads are made with integrators
(thus, much filter design work simply focuses on the design of integrators)
 Biquads
How many biquad filter functions are there?

T(s)
Biquad

a2s2 +a1s+a0
T  s = 2
s +b1s+b0

a2s2 +a1s+a0
T  s = a0  0, a1  0 , a2  0
s2 +b1s+b0

a0 a2s2 +a0
T s = 2
a0  0 T  s = 2 a0  0, a2  0
s +b1s+b0 s +b1s+b0

a1s a1s+a0
T s = 2
a1  0 T s = 2
a0  0, a1  0
s +b1s+b0 s +b1s+b0

a 2 s2 a2s2 +a1s
T  s = a2  0 T  s = a2  0, a1  0
s2 +b1s+b0 s2 +b1s+b0
 Filter Design/Synthesis Considerations

T  s
Review: Second-order bandpass transfer function XIN XOUT

T2BP  jω 

ω 
s 0 
H Q
H T2BP  s   H
ω 
s2 + s  0  + ω02
2 Q
ω
ω0 ω0
BW = ωB -ω A =
Q
ωA ωB ωPEAK = ω0
 Filter Design/Synthesis Considerations

There are many different filter architectures that can realize a given transfer
function

Will first consider second-order Bandpass filter structures

T  s
XIN XOUT

ω 
s 0 
T  s  H Q
ω 
s2 + s  0  + ω02
Q
ω0
BW = ωB -ω A =
Q
ωPEAK = ω0
 Filter Design/Synthesis Considerations
There are many different filter architectures that can realize a given transfer function

Will first consider second-order Bandpass filter structures

Example 1:
R
VOUT
VOUT 1 s
 T  s 
VIN C L VIN RC  1  1
s2 +s   +
 RC  LC

Second-order Bandpass Filter

3 degrees of freedom
2 degrees of freedom for determining dimensionless transfer function
(impedance values scale)

BW = ?
ω0  ? Q=?
Example 1:
R VOUT
VOUT 1 s
 T  s 
VIN C L VIN RC  1  1
s2 +s   +
 RC  LC

C 1
1 BW =
ω0  Q=R RC
LC L

Can realize an arbitrary stable 2nd order bandpass function within a gain factor

Simple design process (sequential but not independent control of ω0 and Q)

If trimming is necessary, prefer to trim with a single resistor

Can’t trim this filter with single resistor

Example 2:
R2
R1 VOUT
K
C2
VIN C1 R3

VOUT K s
 T  s 
VIN R1C1  1 1 1 1-K  R1+R2
s2 +s  + + +  +
 R1C1 R3C1 R3C2 R2C1  R1R2R3C1C2

Second-order Bandpass Filter

6 degrees of freedom (effectively 5 because dimensionless)

Denote as a +KRC filter

ω0  ? Q=? BW = ?
Example 2:
R
R VOUT
K Equal R, Equal C Realization
C
VIN R VOUT K s
 T  s 
C
VIN RC  4-K  2
s2 +s   +
 RC  RC 
2

ω0  ? Q=? BW = ?
 Example 2:
R
R VOUT
K Equal R, Equal C Realization
C
VIN R VOUT K s
 T  s 
C
VIN RC  4-K  2
s2 +s   +
 RC  RC 
2

2 2 4-K
ω0  Q= BW =
RC 4-K RC
3 degrees of freedom (effectively 2 since dimensionless)

• Can satisfy arbitrary 2nd=order BP constraints within a gain factor with this circuit
• Very simple circuit structure

• Independent control of ω0 and Q but requires tuning more than one component

• Can actually move poles in RHP by making K >4

Example 2:
R
R VOUT
1 Unity Gain, Equal R
C2
VIN R VOUT 1 s
C1  T  s 
VIN R C1   1  2 1  2
s2 +s     +   + 2
  R   C1 C2   R C1C2

ω0  ? Q=? BW = ?
Example 2:
R
R VOUT
1 Unity Gain, Equal R
C2
VIN C1 R

VOUT 1 s
 T  s 
VIN R C1   1  2 1  2
s2 +s     +   + 2
  R   C1 C2   R C1C2

 1 2 1 
ω0 
2
Q 2
C2 1 C1
+ BW =     
R C1C2 C1 2 C2  R   C1 C2 

Can’t trim this filter with resistor

Example 3: C1
VOUT
-K
R1 C2
VIN R2

VOUT K s
 T  s  
VIN 1+K  R1C1  1 1 1  1  1
s2 +s   + +   +
  R1C1 R2C2 R2C1  1+K   1+K  R1R2C1C2

Second-order Bandpass Filter

5 degrees of freedom (4 effective since dimensionless)

Denote as a -KRC filter

ω0  ? Q=? BW = ?
Example 3: C
VOUT
-K
R C
VIN R

Equal R, Equal C Realization

VOUT K s
 T  s  
VIN 1+K  RC  3  1  1
s2 +s     +
  RC  1+K   1+K RC 
2

3 degrees of freedom
ω0  ? Q=? BW = ?
 Example 3: C
VOUT
-K
R C
VIN R

Equal R, Equal C Realization

VOUT K s
 T  s  
VIN 1+K  RC  3  1  1
s2 +s     +
  RC  1+K   1+K RC 
2

1 1+K 3
ω0  BW =
Q=
RC 1+K
3 RC 1+K 
3 degrees of freedom (2 effective since dimensionless)

• Can satisfy arbitrary 2nd=order BP constraints within a gain factor with this circuit

• Very simple circuit structure

• Simple design process (sequential but not independent control of ω0 and Q,
requires tuning of more than 1 component if Rs used)
 Observation:

R2
R1 VOUT C1
K VOUT
C2 -K
VIN C1 R3 R1 C2
VIN R2

These are often termed Sallen and Key filters

Sallen and Key introduced a host of filter structures

Sallen and Key structures comprised of summers,
RC network, and finite gain amplifiers
These filters were really ahead of their time and appeared long before
practical implementations were available
1955
 R2

Example 4:
R1 C2 C1
VIN

VOUT

VOUT 1 s
 T  s  
VIN R1C1  1  1 1  1
s2 +s  C + C   + R R C C
 R2  1 2  1 2 1 2

Second-order Bandpass Filter

4 degrees of freedom (3 effective since dimensionless)

Denote as a bridged T feedback structure

 R2
Example 4:

R1 C C
Equal C
VIN implementation

VOUT

VOUT 1 s
 T  s  
VIN R1C  2  1
s2 +s   + 2
 CR2  R1R2C

3 degrees of freedom (2 effective since dimensionless)

ω0  ? Q=? BW = ?
 R2
Example 4:

R1 C C
Equal C
VIN implementation

VOUT

VOUT 1 s
 T  s  
VIN R1C  2  1
s2 +s   + 2
 CR2  R1R2C

1 2
ω0  Q=
1 R2 BW =
C R1R2 2 R1 R 2C

Simple circuit structure

More tedious design/calibration process for ω0 and Q (iterative if C is fixed)

Resistor ratio is 4Q2
 R2
Example 4:
R1 C C
VIN

VOUT

Some variants of the bridged-T feedback structure

R2 C2

R1 C2 C1 R1 R2
VIN VIN
VOUT C1
K
VOUT
C2

R1 R2
VIN
C1 VOUT
K
Example 5:

RQ R1

C1 C2 RA
VIN R0 R2 RA

VOUT

VOUT 1 s
 T  s  
VIN R 0 C2  1  1
s2 +s   +
 RQC2  R1R2C1C2

Second-order Bandpass Filter

8 degrees of freedom (effectively 7 since dimensionless)

Denote as a two-integrator-loop structure

 Example 5:

RQ R

C C RA
VIN R R RA Equal R Equal C
(except RQ)

VOUT

VOUT 1 s
 T  s  
VIN RC  R  1  1
s2 +s     +
 R RC  RC 2
 Q  

3 degrees of freedom (effectively 2 since dimensionless)

ω0  ? Q=? BW = ?
 Example 5:

RQ R

C C RA
VIN R R RA Equal R Equal C
(except RQ)

VOUT
VOUT 1 s
 T  s  
VIN RC  R  1  1
s2 +s     +
 R RC  RC 2
 Q  
3 degrees of freedom (effectively 2 since dimensionless)

1 R R 1
ω0  Q= Q BW =  
RC R  RQ  RC

Simple design process (sequential but not independent control of ω0 and Q with Rs,
requires more tuning more than one R if Cs fixed )

Modest component spread even for large Q

Example 5:
RQ R1

C1 C2 RA
VIN R0 R2 RA

VOUT

RQ R1

C1 C2 R3
R0 R2 R3
VIN
VOUT
INT1 INT2

Two Integrator Loop Representation

 Example 5: R1
RQ

Two Integrator Loop Representation C1

R2
C2 R3
VIN R0 R3

VOUT
INT1 INT2

I0 I
XIN  s
0
s
α
I
XOUT 0
s
Inverting and Noninverting Integrator Loop

I0 XOUT
XIN  s+

XOUT Integrator and Lossy Integrator Loop

I0 I
XIN  s+
0
s

Integrator and Lossy Integrator Loop

 How does the performance of these bandpass filters compare?
R2
R
R VOUT
C K
R1 C
C
VIN VIN R
C

VOUT

R
VOUT

VIN C L

C
VOUT RQ R
-K
R C C C RA
VIN R R R RA
VIN

VOUT
Ideally, all give same performance (within a gain factor)
How does the performance of these bandpass filters compare?

R2
R VOUT

R1 C C R
VIN C L R VOUT
VIN K
C
VIN C R
VOUT

• Component Spread
• Number of Op Amps
• Is the performance strongly dependent upon how DOF are used?
• Ease of tunability/calibration (but practical structures often are not calibrated)
• Total capacitance or total resistance
• Power Dissipation
• Sensitivity
• Effects of Op Amps

C
VOUT RQ R
-K
R C
VIN R C C RA
VIN R R RA

VOUT
End of Lecture 17