REV. D
AD745
–11–
DESIGN CONSIDERATIONS FOR I-TO-V CONVERTERS
There are some simple rules of thumb when designing an I-V
converter where there is significant source capacitance (as with
a photodiode) and bandwidth needs to be optimized. Consider
the circuit of Figure 18. The high frequency noise gain
(1 + C
S
/C
L
) is usually greater than five, so the AD745, with its
higher slew rate and bandwidth is ideally suited to this applica-
tion.
Here both the low current and low voltage noise of the AD745 can
be taken advantage of, since it is desirable in some instances to
have a large R
F
(which increases sensitivity to input current noise)
and, at the same time, operate the amplifier at high noise gain.
AD745
I
S
R
B
C
S
C
L
R
F
INPUT SOURCE: PHOTO DIODE,
ACCELEROMETER, ECT.
Figure 18. A Model for an l-to-V Converter
In this circuit, the R
F
C
S
time constant limits the practical band-
width over which flat response can be obtained, in fact:
f
B
≈
f
C
2π R
F
C
S
where:
f
B
= signal bandwidth
f
C
= gain bandwidth product of the amplifier
With C
L
≈ 1/(2 π R
F
C
S
) the net response can be adjusted to a
provide a two pole system with optimal flatness that has a corner
frequency of f
B
. Capacitor C
L
adjusts the damping of the circuit’s
response. Note that bandwidth and sensitivity are directly traded
off against each other via the selection of R
F
. For example, a
photodiode with C
S
= 300 pF and R
F
= 100 kΩ will have a maxi-
mum bandwidth of 360 kHz when capacitor C
L
≈ 4.5 pF.
Conversely, if only a 100 kHz bandwidth were required, then
the maximum value of R
F
would be 360 kΩ and that of capaci-
tor C
L
still ≈ 4.5 pF.
In either case, the AD745 provides impedance transformation,
the effective transresistance, i.e., the I/V conversion gain, may
be augmented with further gain. A wideband low noise amplifier
such as the AD829 is recommended in this application.
This principle can also be used to apply the AD745 in a high
performance audio application. Figure 19 shows that an I-V
converter of a high performance DAC, here the AD1862, can
be designed to take advantage of the low voltage noise of the
AD745 (2.9 nV/冑Hz) as well as the high slew rate and band-
width provided by decompensation. This circuit, with component
values shown, has a 12 dB/octave rolloff at 728 kHz, with a
passband ripple of less than 0.001 dB and a phase deviation of
less than 2 degrees @ 20 kHz.
0.1F
AD745
0.1F
+12V
–12V
100pF
2000pF
10F
+
DIGITAL
COMMON
0.01F
–12V
AD1862
20-BIT D/A
CONVERTER
3k
TOP VIEW
3 POLE
LOW
PA S S
FILTER
OUTPUT
0.01F
ANALOG
COMMON
+12V
DIGITAL
INPUTS
+12V
0.01F
–12V
0.01F
1F
+
16
15
14
13
12
11
10
9
1
2
3
4
5
6
7
8
Figure 19. A High Performance Audio DAC Circuit
An important feature of this circuit is that high frequency en-
ergy, such as clock feedthrough, is shunted to common via a
high quality capacitor and not the output stage of the amplifier,
greatly reducing the error signal at the input of the amplifier and
subsequent opportunities for intermodulation distortions.
INPUT CAPACITANCE – pF
40
30
0
10 1k100
RTI NOISE VOLTAGE – nV/ Hz
20
10
BALANCED
2.9nV/ Hz
UNBALANCED
Figure 20. RTI Noise Voltage vs. Input Capacitance
BALANCING SOURCE IMPEDANCES
As mentioned previously, it is good practice to balance the
source impedances (both resistive and reactive) as seen by the
inputs of the AD745. Balancing the resistive components will
optimize dc performance over temperature because balancing
will mitigate the effects of any bias current errors. Balancing
input capacitance will minimize ac response errors due to the
amplifier’s input capacitance and, as shown in Figure 20, noise
performance will be optimized. Figure 21 shows the required
external components for noninverting (A) and inverting (B)
configurations.
