
REV. A
AD8391
–10–
VO
BIAS
VN
VP
Figure 3. Simplified Schematic
G = 1
IT = IIN
CT RT
IIN
VOUT
RG
RF
RIN
+
VIN
–
VO
+
–
Figure 4. Model of Current Feedback Amplifier
Feedback Resistor Selection
In current feedback amplifiers, selection of the feedback and
gain resistors will impact distortion, bandwidth, noise, and gain
flatness. Care should be exercised in the selection of these resistors
so that the optimum performance is achieved. Table I shows the
recommended resistor values for use in a variety of gain settings for
the test circuits in TPC 1 and TPC 19. These values are only
intended to be a starting point when designing for any application.
Table I. Resistor Selection Guide
Gain
RF ( )RG ( )
–1
909
–2
909
453
–3
909
303
–4
909
227
–5
909
178
GENERAL INFORMATION
Theory of Operation
The AD8391 is a dual current feedback amplifier with high
output current capability. It is fabricated on Analog Devices’
proprietary eXtra Fast Complementary Bipolar Process (XFCB) that
enables the construction of PNP and NPN transistors with fT’s
greater than 3 GHz. The process uses dielectrically isolated
transistors to eliminate the parasitic and latch-up problems caused
by junction isolation. These features enable the construction of
high frequency, low distortion amplifiers.
The AD8391 has a unique pin out. The two noninverting inputs
of the amplifier are connected to the VMID pin, which is internally
biased by two 5 k
resistors forming a voltage divider between
+VS and –VS. VMID is accessible through Pin 7. There is also a
10 pF internal capacitor from VMID to –VS. The two inverting pins
are available at Pin 1 and Pin 8, allowing the gain of the amplifiers to
be set with external resistors. See Page 1 for a connection diagram
of the AD8391.
A simplified schematic of an amplifier is shown in Figure 3.
Emitter followers buffer the positive input, VP, to provide low
input current and current noise. The low impedance current
feedback summing junction is at the negative input, VN. The
output stage is another high gain amplifier used as an integrator
to provide frequency compensation. The complementary common-
emitter output provides the extended output swing.
A current feedback amplifier’s bandwidth and distortion perfor-
mance are relatively insensitive to its closed-loop signal gain,
which is a distinct advantage over a voltage-feedback architecture.
Figure 4 shows a simplified model of a current feedback amplifier.
The feedback signal is an error current that flows into the inverting
node. RIN is inversely proportional to the transconductance of
the amplifier’s input stage, gmi. Circuit analysis of the pictured
follower with gain circuit yields:
V
GTz s
Tz s
R
G
R
OUT
IN
F
IN
=
×
()
() ++ ×
where:
G
R
F
G
=+
1
Tz s
R
sC
R
F
TT
() =
+
1(
)
R
g
IN
mi
=
1
125
Recognizing that G
× RIN << RF , and that the –3 dB point is set
when Tz(s) = RF, one can see that the amplifier’s bandwidth
depends primarily on the feedback resistor. There is a value of
RF below which the amplifier will be unstable, as the amplifier
will have additional poles that will contribute excess phase shift.
The optimum value for RF depends on the gain and the amount
of peaking tolerable in the application. For more information
about current feedback amplifiers, see ADI’s high speed design
techniques at www.analog.com/technology/amplifiersLinear/
designTools/evaluationBoards/pdf/1.pdf.