AD844
Table I.
I
SIG
R1
Gain
–1
–1
–2
–2
–5
–5
–10
–10
–20
–100
+100
R1
1 kΩ
500
Ω
2 kΩ
1 kΩ
5 kΩ
500
Ω
1 kΩ
500
Ω
1 kΩ
5 kΩ
5 kΩ
R2
1 kΩ
500
Ω
1 kΩ
500
Ω
1 kΩ
100
Ω
100
Ω
50
Ω
50
Ω
50
Ω
50
Ω
BW (MHz)
35
60
15
30
5.2
49
23
33
21
3.2
9
GBW (MHz)
35
60
30
60
26
245
230
330
420
320
900
C
S
AD844
R
L
V
OUT
C
L
Figure 3. Current-to-Voltage Converter
Circuit Description of the AD844
Response as an I-V Converter
The AD844 works well as the active element in an operational
current to voltage converter, used in conjunction with an exter-
nal scaling resistor, R1, in Figure 3. This analysis includes the
stray capacitance, C
S
, of the current source, which might be a
high speed DAC. Using a conventional op amp, this capacitance
forms a “nuisance pole” with R1 which destabilizes the closed
loop response of the system. Most op amps are internally com-
pensated for the fastest response at unity gain, so the pole due
to R1 and C
S
reduces the already narrow phase margin of the
system. For example, if R1 were 2.5 kΩ a C
S
of 15 pF would
place this pole at a frequency of about 4 MHz, well within the
response range of even a medium speed operational amplifier.
In a current feedback amp this nuisance pole is no longer deter-
mined by R1 but by the input resistance, R
IN
. Since this is about
50
Ω
for the AD844, the same 15 pF forms a pole 212 MHz
and causes little trouble. It can be shown that theresponse of
this system is:
V
OUT
=
–
Isig
K R1
(1
+
sTd
)(1
+
sTn
)
A simplified schematic is shown in Figure 4. The AD844 differs
from a conventional op amp in that the signal inputs have
radically different impedance. The noninverting input (Pin 3)
presents the usual high impedance. The voltage on this input is
transferred to the inverting input (Pin 2) with a low offset
voltage, ensured by the close matching of like polarity transis-
tors operating under essentially identical bias conditions. Laser
trimming nulls the residual offset voltage, down to a few
tens of microvolts. The inverting input is the common emitter
node of a complementary pair of grounded base stages and
behaves as a current summing node. In an ideal current feed-
back op amp the input resistance would be zero. In the AD844
it is about 50
Ω.
A current applied to the inverting input is transferred to a
complementary pair of unity-gain current mirrors which deliver
the same current to an internal node (Pin 5) at which the full
output voltage is generated. The unity-gain complementary
voltage follower then buffers this voltage and provides the load
driving power. This buffer is designed to drive low impedance
loads such as terminated cables, and can deliver
±
50 mA into a
50
Ω
load while maintaining low distortion, even when operat-
ing at supply voltages of only
±
6 V. Current limiting (not
shown) ensures safe operation under short circuited conditions.
7 +V
S
where
K
is a factor very close to unity and represents the finite
dc gain of the amplifier,
Td
is the dominant pole and
Tn
is the
nuisance pole:
I
B
K
=
R
t
R
t
+
R
1
+IN 3
2 –IN
TZ
5
6 OUT
Td
=
KR1C
t
Tn
=
R
IN
C
S
(assuming
R
IN
<<
R1)
Using typical values of
R1
= 1 kΩ and R
t
= 3 MΩ, K is 0.9997;
in other words, the “gain error” is only 0.03%. This is much less
than the scaling error of virtually all DACs and can be absorbed,
if necessary, by the trim needed in a precise system.
In the AD844, R
t
is fairly stable with temperature and supply
voltages, and consequently the effect of finite “gain” is negli-
gible unless high value feedback resistors are used. Since that
would result in slower response times than are possible, the
relatively low value of R
t
in the AD844 will rarely be a signifi-
cant source of error.
I
B
4 –V
S
Figure 4. Simplified Schematic
–8–
REV. D
AD [ ANALOG DEVICES ]
