OP176 Datasheet, PDF (16/21 Pages) Analog Devices – Bipolar/JFET, Audio Operational Amplifier

OP176
Passive Component Selection for Active Filters
The passive components suitable for active filters deserve more
than casual attention. Resistors should be 1%, low TC, metal-
film types of the RN55 or RN60 style. Capacitors should be 1%
or 2% film types preferably, such as polypropylene or polysty-
rene, or NPO (COG) ceramic for smaller values.
Active Filter Circuit Subtleties
In designing active filter circuits with the OP176, moderately
low values (10 kΩ or less) for R1 and R2 can be used to
minimize the effects of Johnson noise when critical. The
practical tradeoff is, of course, capacitor size and expense. DC
errors will result for larger values of resistance, unless compen-
sation for amplifier input bias current is used. To add bias
compensation in the HP filter section of Figure 42a, a feedback
compensation resistor equal to R2 can be used. This will
minimize bias current induced offset to the product of the
OP176’s I_OS and R2. For an R2 of 25 kΩ, this produces a typical
compensated offset voltage of 50 µV. Similar compensation is
applied to Figure 42b, using a resistance equal to R1+ R2.
Using dc compensation, filter output dc errors using the OP176
will be dominated by its V_OS, which is typically 1 mV or less. A
caveat here is that the additional resistors can increase noise
substantially. For example, a 10 kΩ resistor generates ~ 12 nV/
√Hz of noise and is about twice that of the OP176. These
resistors can be ac bypassed to eliminate their noise using a
simple shunt capacitor chosen such that its reactance (X_C) is
much less than R at the lowest frequency of interest.
A more subtle form of ac degradation is also possible in these
filters, namely nonlinear input capacitance modulation. This
issue was previously covered for general cases in the section on
minimizing distortion. In active filter circuits, a fully compen-
sating network (for both dc and ac performance) can be used to
minimize this distortion. To be most effective, this network
(Z_COMP) should include R1 through C2 as noted for either filter
type, of the same style and value as their counterparts in the
forward path. The effects of a Z_COMP network on the THD + N
performance of two 1 kHz HP filters is illustrated in Figure 46.
One filter (A) is the example shown in Figure 44a (Curves A1
and A2), while the second (B) uses RC values scaled 10 times
upward in impedance (Curves B1 and B2). Both filters operate
with a 2 V rms input, ± 18 V supplies, 100 kΩ loading, and
analyzer bandwidth of 80 kHz.
1
0.1
0.010
0.001
0.0001
20
B1
A1
B2
A2
100
1k
FREQUENCY – Hz
10 k 20k
Figure 46. THD + N (%) vs. Frequency for Various 1 kHz HP
Active Filters Illustrating the Effects of the Z_COMP Network
Curves A1 and B1 show performance with Z_COMP shorted,
while curves A2 and B2 illustrate operation with Z_COMP active.
For the “A” example values, distortion in the pass band of
1 kHz–20 kHz is below 0.001% compensated, and slightly
higher uncompensated. With the higher impedance “B” net-
work, there is a much greater difference between compensated
and uncompensated responses, underscoring the sensitivity to
higher impedances. Although the positive effect of Z_COMP is seen
for both “A” and “B” cases, there is a buffering effect which
takes place with lower impedances. As case “A” shows, when
using larger capacitance values in the source, the amplifier’s
nonlinear C-V input characteristics have less effect on the
signal.
Thus, to minimize the necessity for the complete Z_COMP com-
pensation, effective filter designs should use the lowest capaci-
tive impedances practical, with an 0.01 µF lower value limit as a
goal for lowest distortion (while lower values can certainly be
used, they may suffer higher distortion without the use of full
compensation). Since most designs are likely to use low relative
impedances for reasons of low noise and offset, the effects of
CM distortion may or may not actually be apparent to a given
application.
–16–
REV. 0