Datasheet
AD823A Data Sheet
Rev. | Page 16 of 20
WIDEBAND PHOTODIODE PREAMP
–
+
V
OUT
V
B
C
D
C
M
C
M
AD823A
R
SH
= 10
11
Ω
C
S
I
PHOTO
C
F
R
F
09439-055
Figure 42. Wideband Photodiode Preamp
The AD823A is an excellent choice for photodiode preamp
application. Its low input bias current minimizes the DC error
at the preamp output. In addition, its high gain bandwidth
product and low input capacitance maximizes the signal
bandwidth of the photodiode preamp. Figure 42 shows the
AD823A as a current-to-voltage (I/V) converter with an
electrical model of a photodiode.
The transimpedance gain of the photodiode preamp can be
described by the basic transfer function:
FF
FPHOTO
OUT
RsC
RI
V
+
×
=
1
(1)
where I
PHOTO
is the output current of the photodiode, and the
parallel combination of R
F
and C
F
sets the signal bandwidth (see
the I to V gain curve in Figure 43). Note that one should set R
F
such that the maximum attainable output voltage corresponds
to the maximum diode current I
PHOTO
. This allows one to utilize
the full output swing.
The signal bandwidth that is attainable with this preamp is a
function of R
F
, the gain bandwidth product (f
u
) of the amplifier,
and the total capacitance at the amplifier summing junction,
including C
S
and the amplifier input capacitance C
D
and C
M
. R
F
and the total capacitance produce a pole with loop frequency (f
p
).
SF
p
CR
f
π
2
1
=
(2)
With the additional pole from the amplifier’s open loop
response, the two-pole system results in peaking and instability
due to an insufficient phase margin (Figure 43(A), Without
Compensation).
Adding C
F
creates a zero in the loop transmission that compensates
for the effect of the input pole. This stabilizes the photodiode
preamp design because of the increased phase margin. It also sets
the signal bandwidth (Figure 43(B), With Compensation). The
signal bandwidth and the zero frequency are determined by
FF
z
CR
f
π2
1
=
(3)
Setting the zero at the frequency f
x
maximizes the signal
bandwidth with a 45° phase margin. Since f
x
is the geometric
mean of f
p
and f
u
, it can be calculated by
upx
fff ×=
(4)
Combining Equation 2, Equation 3 and Equation 4, the value of
C
F
that produces f
x
is defined by
uF
S
F
fR
C
C
××
=
π
2
(5)
The frequency response in this case shows about 2 dB of
peaking and 15% overshoot. Doubling C
F
and cutting the
bandwidth in half results in a flat frequency response with
about 5% transient overshoot.
B