Using State Variable Filter to Create Lead/Lag Filters

Rayal Johnson

April 29, 2004

Introduction the denominator without the use of feed-forward
zeros. In other words:
The state variable filter serves as an exam-
ple of a low-pass, high-pass and band-pass fil- Θ(N um) ≤ Θ(Den) (4)
ter. The filter is unique because it is based on is possible using only integrators.
the idea that any transfer function can be cre-
ated using only integrators and gain blocks as
its basic components. This filter also serves as Modifying SVF for Lead/Lag
a starting point to creating other filters such as Compensators
lead and lag filters used in compensators for con-
trol systems. The lead compensator is a very useful filter in
control theory. This filter is used to add phase
margin to a system close to instability. The lead
State-Variable Filter (SVF) compensator transfer function is realized with
one pole and one zero. The lead zero is at a
The state-variable filter uses two integrators lower frequency than that of its pole, creating a
and one gain block to create a filter that exhibits phase bump at a frequency equalling the geomet-
a low-pass, band-pass or high-pass characteristic ric mean of the pole and zero. The lead transfer
based on where the output is realized. The ba- function is:
sic state-variable filter is presented in Figure 1. τ1 s + 1
s2Vo sVo H= , τ1 > τ2 (5)
τ2 s + 1
Vi AB A B B Vo
- - s s
In order to realize the ‘lead’ transfer function,
the low-pass transfer function from the SVF is
C
analyzed. Some modifications are however nec-
essary. First, the ‘s2 ’ in the denominator needs
to disappear, and secondly, there needs to be a
Figure 1: State-Variable Filter ‘s’ as the second term in the numerator. The
The low-pass, band-pass and high-pass out- latter can be realized by adding a feed-forward
s2 Vo
puts are represented by Vo , sV
B and AB respec- path to the SVF as shown in Figure 2.
o

tively, which from here on in, will be referred to

as Vo , V2 and V1 . The respective transfer func- D
tions are represented by Equations 1-3. +
Vi A B Vo
Vo AB - - s s
= 2 (1)
Vi s + CAs + AB
C

Vo sA
= 2 (2)
V2 s + CAs + AB Figure 2: State-Variable Filter, Modified
The resulting transfer function of this config-
Vo s2
= 2 (3) uration is:
V1 s + CAs + AB
Vo A(Ds + B)
= 2 (6)
Vi s + A(C + D)s + AB
These three equations show that it is possi-
ble to realize transfer functions with the order of The next step is to now eliminate the ‘s2 ’ term.
the numerator less than or equal to the order of It is clear that ‘A’ multiplies all terms except the

1
‘s2 ’ term, therefore to eliminate this term, take Beginning with Figure 3, the next step is
the limit as ‘A’ approaches infinity, then divide to assign voltages and currents to each line in
by ‘B’ to make it the same form as Equation 5. the block diagram. The most convenient as-
signments, if OTAs will be used in place of
D operational amplifiers, are shown in Figure 4.
Vo Bs + 1
lim = C+D (7)
A→∞ BVi
B s+1 i
D

Since the function is now in the same form as the v1 +

Vi 1 i Vo
lead function, the time constants can be deter- C
B
s v2
i
mined.
D
τ1 = (8)
B
Figure 4: Lead/Lag Filter with Variables as Cur-
C +D
τ2 = (9) rents and Voltages
B
There is one thing to note, however, namely
τ1 > τ2 is required for a lead compensator. In
Circuit Implementation of
order for this to be true, ‘C’ has to be negative. Lead/Lag Filter
D C +D Since the error voltage changes to a current, it
> (10) is clear that an OTA is used there. One thing to
B B
notice in the block diagram is that at the output
of the ‘ C1 ’ block is sampled and not split into two
D >C +D (11)
currents. The error voltage therefore needs to
be changed to a current twice, one for the gain
0>C (12) block and one for the integrator. The implemen-
tation of this block diagram is shown in Figure 5.
If ‘C’ is made to be positive, then a lag compen-
sator can be realized with the same topology. +
V i
The final block diagram is shown in Figure 3. in
-
G 1-
G v 2 1
+ +
C G3 -
v2 - G3
D + + Vout
i −
G1
-
+ +
Vi 1 B Vo
C s

Figure 5: Lead/Lag Filter Implementation

Figure 3: Lead/Lag Filter In this configuration, any OTAs that have the
From this block diagram, it is easy to see how same subscript have the same bias currents. A
a lead or lag filter can be constructed with the short hand-analysis of this circuit will show the
use of one integrator and two gain stages. Basic transfer function for the lead/lag filter.
analog building blocks can now be used to realize
this filter. A more convenient way to synthesize i = G1 (Vin − Vout ) (13)
this filter is to just use transconductance ampli- i G1 (Vin − Vout )
fiers (OTA) and capacitors, and no resistors. V1 = = (14)
G2 G2

2
 i G1 (Vin − Vout ) C
V2 = − =− (15) τ1 = (18)
Cs Cs G2

G3 (V1 − V2 ) (G1 + G2 )C
Vout = = V1 − V2 (16) τ2 = (19)
G3 G1 G2

In this configuration, the circuit is a lag fil-

After a little algebra, the transfer function is ter, because τ2 > τ1 . To make this circuit into a
realized as: lead filter, the inputs to the G1 filter need to be
Cs
switched, specifically, Vin connects to the nega-
Vout G2 + 1 tive terminal and Vout connects to the positive
= (G +G )Cs (17)
Vin 1 2
+1 terminal. Two birds with one stone!
G G
1 2