AD8002
Data Sheet
THEORY OF OPERATION
An analysis of the AD8002 can put the operation in familiar
terms. The open-loop behavior of the AD8002 is expressed
as transimpedance, ΔVOUT/ΔI−INx, or TZ. The open-loop
transimpedance behaves just as the open-loop voltage gain
of a voltage feedback amplifier, that is, it has a large dc value
and decreases at roughly 6 dB/octave in frequency.
Because the value of RIN is proportional to 1/gm, the equivalent
voltage gain is just TZ × gm, where the gm in question is the
transconductance of the input stage. This results in a low open-
loop input impedance at the inverting input. Using this
amplifier as a follower with gain (see Figure 48) basic analysis
yields the following result:
VOUT = G ×
TZ (s)
VIN
TZ (s) + G × RIN + R1
where:
TZ(s) implies the transimpedance as a function of the frequency.
G = 1 + R1/R2.
RIN = 1/gm ≈ 50 Ω.
R1
R2
RIN
VOUT
PRINTED CIRCUIT BOARD (PCB) LAYOUT
CONSIDERATIONS
As expected for a wideband amplifier, PCB parasitics can affect
the overall closed-loop performance. Of concern are stray
capacitances at the output and the inverting input nodes. If a
ground plane is to be used on the same side of the board as the
signal traces, leave a space (5 mm minimum) around the signal
lines to minimize coupling. Additionally, make signal lines
connecting the feedback and gain resistors short enough so that
their associated inductance does not cause high frequency gain
errors. Line lengths of less than 5 mm are recommended. If long
runs of coaxial cable are being driven, dispersion and loss must
be considered.
POWER SUPPLY BYPASSING
Adequate power supply bypassing can be critical when optimiz-
ing the performance of a high frequency circuit. Inductance in
the power supply leads can form resonant circuits that produce
peaking in the response of the amplifier. In addition, if large
current transients must be delivered to the load, bypass capaci-
tors (typically greater than 1 μF) are required to provide the best
settling time and lowest distortion. A parallel combination of
4.7 μF and 0.1 μF is recommended. Some brands of electrolytic
capacitors require a small series damping resistor ≈4.7 Ω for
optimum results.
VIN
DC ERRORS AND NOISE
Figure 48. Small Signal Schematic
Recognizing that G × RIN << R1 for low gains, the amplifier
can be seen to the first-order that the bandwidth for it is
independent of gain (G).
Considering that additional poles contribute excess phase at
high frequencies, there is a minimum feedback resistance below
which peaking or oscillation may result. This fact is used to
determine the optimum feedback resistance, RF. In practice,
parasitic capacitance at the inverting input terminal also adds
phase in the feedback loop; thus selecting an optimum value for
RF can be difficult.
Achieving and maintaining gain flatness of better than 0.1 dB
at frequencies above 10 MHz requires careful consideration of
several issues discussed in the following sections.
CHOICE OF FEEDBACK AND GAIN RESISTORS
The fine scale gain flatness varies to some extent with feedback
resistance. Therefore, it is recommended that as soon as
optimum resistor values are determined, use 1% tolerance
values if it is desired to maintain flatness over a wide range of
production lots. In addition, resistors of different construction
have different associated parasitic capacitance of the character-
ization. It is not recommended to use leaded components with
the AD8002.
There are three major noise and offset terms to consider in a
current feedback amplifier. For offset errors, refer to Equation 1.
For noise error, the terms are root-sum-squared to give a net
output error. In Figure 49, the terms are input offset (VIO), which
appears at the output multiplied by the noise gain of the circuit
(1 + RF/RI), noninverting input current (IBN × RN), also multiplied
by the noise gain, and the inverting input current, which when
divided between RF and RI and subsequently multiplied by the
noise gain, always appears at the output as IBN × RF.
The input voltage noise of the AD8002 is a low 2 nV/√Hz. At low
gains, though, the inverting input current noise times RF is the
dominant noise source. Careful layout and device matching
contribute to a better offset and drift specifications for the
AD8002.Use the typical performance curves in conjunction
with Equation 1 to predict the performance of the AD8002 in
any application.
VOUT
= VIO
× 1 +
RF
RI
± IBN
× RN
× 1 +
RF
RI
× I BI
× RF
(1)
RF
RI
IBI
RN
IBN
VOUT
Rev. E | Page 14 of 21
Figure 49. Output Offset Voltage