ADA4938-1/ADA4938-2
Rev. A | Page 19 of 28
THEORY OF OPERATION
The ADA4938-x differs from conventional op amps in that it
has two outputs whose voltages move in opposite directions.
Like an op amp, it relies on open-loop gain and negative
feedback to force these outputs to the desired voltages. The
ADA4938-x behaves much like a standard voltage feedback op
amp and makes it easier to perform single-ended-to-differential
conversions, common-mode level shifting, and amplifications of
differential signals. Also like an op amp, the ADA4938-x has
high input impedance and low output impedance.
Two feedback loops are employed to control the differential and
common-mode output voltages. The differential feedback, set
with external resistors, controls only the differential output
voltage. The common-mode feedback controls only the common-
mode output voltage. This architecture makes it easy to set the
output common-mode level to any arbitrary value. It is forced,
by internal common-mode feedback, to be equal to the voltage
applied to the VOCM input, without affecting the differential
output voltage.
The ADA4938-x architecture results in outputs that are highly
balanced over a wide frequency range without requiring tightly
matched external components. The common-mode feedback
loop forces the signal component of the output common-
mode voltage to zero, which results in nearly perfectly balanced
differential outputs that are identical in amplitude and are
exactly 180° apart in phase.
ANALYZING AN APPLICATION CIRCUIT
The ADA4938-x uses open-loop gain and negative feedback to
force its differential and common-mode output voltages in such
a way as to minimize the differential and common-mode error
voltages. The differential error voltage is defined as the voltage
between the differential inputs labeled +IN and IN (see
Figure 57). For most purposes, this voltage can be assumed
to be zero. Similarly, the difference between the actual output
common-mode voltage and the voltage applied to VOCM can also
be assumed to be zero. Starting from these two assumptions,
any application circuit can be analyzed.
SETTING THE CLOSED-LOOP GAIN
The differential-mode gain of the circuit in
Figure 57 can be
determined by
G
F
dm
IN
dm
OUT
R
V
=
,
This assumes the input resistors (RG) and feedback resistors (RF)
on each side are equal.
ESTIMATING THE OUTPUT NOISE VOLTAGE
The differential output noise of the ADA4938 can be estimated
using the noise model in
Figure 58. The input-referred noise
voltage density, vnIN, is modeled as a differential input, and the
noise currents, inIN and inIN+, appear between each input and
ground. The noise currents are assumed to be equal and produce a
voltage across the parallel combination of the gain and feedback
resistances. vn, cm is the noise voltage density at the VOCM pin.
Each of the four resistors contributes (4kTR)1/2. Table 9 summarizes the input noise sources, the multiplication factors, and the output-
referred noise density terms.
ADA4938
+
RF2
VnOD
VnCM
VOCM
VnIN
RF1
RG2
RG1
VnRF1
VnRF2
VnRG1
VnRG2
inIN+
inIN–
06
59
2-
0
05
Figure 58. ADA4938 Noise Model
Table 9. Output Noise Voltage Density Calculations
Input Noise Contribution
Input Noise Term
Input Noise
Voltage Density
Output
Multiplication Factor
Output Noise
Voltage Density Term
Differential Input
vnIN
GN
vnO1 = GN(vnIN)
Inverting Input
inIN
inIN × (RG2||RF2)
GN
vnO2 = GN[inIN × (RG2||RF2)]
Noninverting Input
inIN+
inIN+ × (RG1||RF1)
GN
vnO3 = GN[inIN+ × (RG1||RF1)]
VOCM Input
vn, cm
GN(β1 β2)
vnO4 = GN(β1 β2)(vnCM)
Gain Resistor, RG1
vnRG1
(4kTRG1)1/2
GN(1 β1)
vnO5 = GN(1 β1)(4kTRG1)1/2
Gain Resistor, RG2
vnRG2
(4kTRG2)1/2
GN(1 β2)
vnO6 = GN(1 β2)(4kTRG2)1/2
Feedback Resistor, RF1
vnRF1
(4kTRF1)1/2
1
vnO7 = (4kTRF1)1/2
Feedback Resistor, RF2
vnRF2
(4kTRF2)1/2
1
vnO8 = (4kTRF2)1/2