AD846
A simple equation can, therefore, be used to determine the band-
width of an amplifier employing the AD846 in the inverting
configuration.
23
3 dB Bandwidth =
R + 0.05 1+ G
(
)
F
where: The 3 dB bandwidth is in MHz
G is the closed-loop inverting gain of the AD846
RF is the feedback resistance in kΩ.
Figure 39. AD846 Three-Terminal Model
NOTE: This equation applies only for values of RF between
10 kΩ and 100 kΩ, and for RLOAD greater than 500 Ω. For RF =
1 kΩ the bandwidth should be estimated from Figure 41.
Figure 41 illustrates the closed-loop voltage gain vs. frequency
of the AD846 for various values of feedback resistor. For com-
parison purposes, the characteristic of a conventional amplifier
having an 80 MHz unity gain bandwidth is also shown.
Figure 40. Op Amp Three-Terminal Model
A more detailed examination of the closed-loop transfer func-
tion of the AD846 results in the following equation:
−RF
RS
RF
Closed-Loop Gain G(s) =
1+ CCOMP RF + 1+
RIN
s
R
S
Compare this to the equation for a conventional op amp:
−RF
RS
Figure 41. Closed-Loop Voltage Gain vs. Bandwidth for
Various Values of RF
CCOMP
gM
RF
Closed-Loop Gain G(s) =
1+
1+
s
For the case where RF = 1 kΩ and RS = 100 Ω (closed-loop gain
of –10), the closed-loop bandwidth is approximately 28 MHz. It
should also be noted that the use of a capacitor to shunt RF, a
normal practice for stabilizing conventional op amps, will cause
this amplifier to become unstable because the closed-loop band-
width will increase beyond the stable operating frequency.
R
S
where: CCOMP is the internal compensation capacitor of the am-
plifier; gM is the input stage transconductance of the amplifier.
In the case of the voltage amplifier, the closed-loop bandwidth
decreases directly with increasing values of (1 + RF/RS), the
closed-loop gain. However, for the transimpedance amplifier,
the situation is different. At low gains, where (1 + RF/RS) RIN is
small compared to RF, the closed-loop bandwidth is controlled
by the internal compensation capacitance of 7 pF and the value
of RF, and not by the closed-loop gain. At higher gains, where (1
+ RF/RS) RIN is much larger than RF, the behavior is that of a con-
ventional operational amplifier in which the input stage transcon-
ductance is equal to the inverting terminal input impedance of
the transimpedance amplifier (RIN = 50 Ω).
A similar approach can be taken to calculate the noise perfor-
mance of the amplifier. A simplified noise model is shown in
Figure 42.
The equivalent mean-square output noise voltage spectral den-
sity will equal:
RF
R
2
S
V
2 = R I
2 + 1+
[VN2 + R I
2 + 4 kT RP ]
(
)
(
)
ON
F
NN
P
NP
RF
+ 4 kT RF
+1
R
S
–9–
REV. C
ADI [ ADI ]
