Datasheet

AD8132
Rev. I | Page 23 of 32
OTHER β2 = 1 CIRCUITS
The preceding simple configuration with β2 = 1 and its gain of 2
is the highest gain circuit that can be made under this condition.
Because β1 was equal to 0, only higher β1 values are possible.
The circuits with higher values of β1 have gains lower than 2.
However, circuits with β1 equal to 1 are not practical because
they have no effective input and result in a gain of 0.
To increase β1 from 0, it is necessary to add two resistors in a feed-
back network. A generalized circuit that has β1 with a value higher
than 0 is shown in Figure 69. A couple of different convenient
gains that can be created are a gain of 1, when β1 is equal to 1/3,
and a gain of 0.5, when β1 equals 0.6.
With β2 equal to 1 in these circuits, V
OCM
serves as the refer-
ence voltage that measures the input voltage and the individual
output voltages. In general, when V
OCM
is varied in circuits with
unmatched feedback networks, a differential output signal is
generated that is proportional to the applied V
OCM
voltage.
VARYING β2
Though the β2 = 1 circuit sets β2 to 1, another class of simple
circuits can be made that sets β2 equal to 0. This means that
there is no feedback from +OUT to −IN. This class of circuits
is very similar to a conventional inverting op amp. However,
the AD8132 circuits have an additional output and common-
mode input that can be analyzed separately (see Figure 71).
With −IN connected to ground, +IN becomes a virtual ground
in the sense that the term is used for conventional op amps. Both
inputs must maintain the same voltage for equilibrium operation;
therefore, if one is set to ground, the other is driven to ground.
The input impedance can also be seen to be equal to R
G
, just as
in a conventional op amp.
In this case, however, the positive input and negative output are
used for the feedback network. Because a conventional op amp
does not have a negative output, only its inverting input can be
used for the feedback network. The AD8132 is symmetrical,
therefore, the feedback network on either side can be used to
produce the same results.
Because +IN is a summing junction, by an analogy to conven-
tional op amps, the gain from V
IN
to −OUT is −R
F
/R
G
. This holds
true regardless of the voltage on V
OCM
, and because +OUT
moves the same amount in the opposite direction from −OUT,
the overall gain is −2(R
F
/R
G
).
V
OCM
still governs V
OUT, cm
; therefore, +OUT must be the only
output that moves when V
OCM
is varied. Because V
OUT, cm
is the
average of the two outputs, +OUT must move twice as far, and in
the same direction as V
OCM
, to create the proper V
OUT, cm
. Therefore,
the gain from V
OCM
to +OUT must be 2.
With β2 equal to 0 in these circuits, the gain can theoretically be
set to any value from close to 0 to infinity, just as it can with a
conventional op amp in the inverting mode. However, practical
real-world limitations and parasitics limit the range of acceptable
gain to more modest values.
β1 = 0
There is yet another class of circuits where there is no feedback
from −OUT to +IN. This is the case where β1 = 0. The differential
amplifier without a resistor described in the Differential Amplifier
Without Resistors (High Input Impedance Inverting Amplifier)
section meets this condition, but it was presented only with the
condition that β2 = 1. Recall that this circuit had a gain equal to 2.
If β2 decreases in this circuit from unity, a smaller part of +V
OUT
is fed back to −IN and the gain increases (see Figure 68). This
circuit is very similar to a noninverting op amp configuration,
except for the presence of the additional complementary output.
Therefore, the overall gain is twice that of a noninverting op
amp or 2 × (1 + R
F2
/R
G2
) or 2 × (1/β2).
Once again, varying V
OCM
does not affect both outputs in the
same way; therefore, in addition to varying V
OUT, cm
with unity
gain, there is also an effect on V
OUT, dm
by changing V
OCM
.
ESTIMATING THE OUTPUT NOISE VOLTAGE
Similar to the case of a conventional op amp, the differential
output errors (noise and offset voltages) can be estimated by
multiplying the input-referred terms, at +IN and −IN, by the
circuit noise gain. The noise gain is defined as
⎟
⎟
⎠
⎞
⎜
⎜
⎝
⎛
+=
G
F
N
R
R
G 1
To compute the total output-referred noise for the circuit of
Figure 64, consideration must be given to the contribution of
resistors, R
F
and R
G
. See Table 11 for estimated output noise
voltage densities at various closed-loop gains.
Table 11. Recommended Resistor Values and Noise
Performance for Specific Gains
Gain R
G
(Ω) R
F
(Ω)
Bandwidth
−3 dB (MHz)
Output
Noise
AD8132
Only
(nV/√Hz)
Output
Noise
AD8132
+ R
G
, R
F
(nV/√Hz)
1 499 499 360 16 17
2 499 1.0 k 160 24.1 26.1
5 499 2.49 k 65 48.4 53.3
10 499 4.99 k 20 88.9 98.6