
AN-405
a
ONE TECHNOLOGY WAY • P.O. BOX 9106
Increasing the Speed of the Output Response of the AD606
INTRODUCTION
The AD606 is a complete demodulating logarithmic amplifier. As such it incorporates successive detection
stages, a dc feedback section to null out the input offset
voltage, a current summer to add up the logarithmic current outputs of the stages, and a low pass filter for the
log output to remove the carrier from the demodulated
signal. The latter, while offering useful integration for
some lower speed systems, slows down the response of
the logarithmic output in high speed systems. It is therefore desireable in high speed systems to bypass this filter and realize the faster response.
The higher speed response is important when using the
AD606 as a logarithmic detector for systems like medical
ultrasound where it is desired to observe a weak echo
signal that closely follows a strong echo signal. It is imperative that the response of the log detector to the
strong echo to totally decay before the following weak
signal can be detected accurately.
The intent of this application note is to describe how to
realize a speed up of approximately a factor of eight in
the logarithmic output response of the AD606.
The internal block diagram of the AD606 as it appears in
the data sheet is shown in Figure 1. The various blocks
mentioned above are all shown. It can be seen that the
REFERENCE
AND POWER UP
30kΩ
HIGH-END
DETECTORS
2
PRUP
ISUM
COMM
16
30kΩ
1.5kΩ
250kΩ
1.5kΩ
AD606
1
INLO
Figure 1. AD606 Simplified Block Diagram
VPOS
1415
3
12µA/dB
ONE-POLE
2 µA/dB
FILTER
13
4
12 11 9
360kΩ
LOW-PASS FILTER
MAIN SIGNAL PATH
11.15 dB/STAGE
9.375kΩ
9.375kΩ
2pF
5
30pF
30pF
OFFSET-NULL
TWO-POLE
2pF
SALLEN-KEY
FILTER
X2
6
360kΩ
X1
•
by Peter Checkovich
LADJFIL1 FIL2
LMHIINHI
10
FINAL
LIMITER
7
LMLOOPCMVLOGBFINILOGCOMM
APPLICATION NOTE
NORWOOD, MASSACHUSETTS 02062-9106
detected current from all the successive demodulating
stages is low pass filtered, converted to a voltage and
output to VLOG (Pin 6). However, one node called ISUM
is connected to Pin 3 and no other mention of it is made
in the data sheet. However, all applications show this pin
as an N/C, so it is not obvious how to get useful current
from this node.
The key to steering the current available at the ISUM
node (labeled as 12 µA/dB) to Pin 3, is to bias Pin 3 at the
same voltage as the positive supply of the AD606 (VPOS,
Pin 13). This will steer the current away from being internally consumed in the part and direct it off chip. This
current into Pin 3 can then be converted to a voltage by
using an op amp I to V converter.
There is, however, an accuracy penalty that is incurred
by using this technique. The ISUM current is inversely
proportional to the value of a thin film resistor whose
absolute accuracy is ±25%. The thin film resistors used
in the on-chip I to V circuitry match those of the ISUM
circuit by better than 1% so the overall accuracy is
not adversely affected when using the on-chip I to V
converter.
When using the faster off-chip I to V converter, the ±25%
tolerance of the thin film resistors of the AD606 will affect the gain of the logarithmic output. In systems where
lineariity is most important and the system is calibrated
for overall gain variations, this should not present any
problems.
The off-chip I to V converter will also have a slightly
greater variation in the intercept vs. temperature than
the on-chip I to V converter. Once again, the therrmal
tracking properties of the thin film resistors minimizes
the thermal variations in the AD606, while using an offchip I to V converter will have about 1 dB of additional
variation in the intercept.
Two configurations can accomplish the proper biasing
of the ISUM terminal as described, however, each re-
8
quires an additional power supply. Since the AD606 is a
single supply part, the price for the added performance
comes at the expense of added system complexity.
•
617/329-4700

ALL POSITIVE SUPPLY CONFIGURATION
The first configuration uses a +5 V and ground supply
for the AD606 as the part is normally designed to operate. Since ISUM will now want to be operating at +5 V, a
higher supply voltage, nominally +10 V, will be required
for the I to V converter. Figure 2 is a schematic of this
configuration. This configuration lends itself well to battery powered systems, where supplies of only one polarity are available and a dc to dc converter to create a
negative supply would generate too much noise.
The positive input of the op amp is biased at +5 V. The
feedback circuit of the op amp will drive the inverting
input to also be biased at +5 V, which is the point it
wants to operate at to use the current available at ISUM
as explained above.
For high speed response, a high speed op amp like one
from Analog Devices XFCB process is required. If filtering of the carrier frequency from the response is also to
be incorporated by using a capacitor in shunt with the
feedback resistor, then the op amp will have to be a voltage feedback type. Also because the op amp is configured with 100% feedback, a unity gain stable op amp is
required. Candidates that meet these criteria and are
cost effective and low power consumption are the
AD8041 and AD8047.
49.9Ω
0.1µF
COMM
30kΩ
15
30kΩ
14
PRUP
REFERENCE
AND POWER UP
VPOS
16
INPUT
+5V
13
The value of the feedback resistor for the op amp will
determine the scaling of the output. The current into the
ISUM node is 12 µ A/dB. In order to produce the same
response as the AD606 (37.5 mV/dB) the feedback resistor should be 37.5 mV/12 µ A or 3.12k. Over the 80 dB
range of the AD606, this will produce a signal with a dynamic range of 80 dB × 37.5 mV/dB or 3.0 V.
Another factor to consider is that with no signal at the
input of the AD606 and ISUM biased at the same potential as the positive supply of the AD606, the standing
current into ISUM is 650 µA dc. This current will create a
voltage of 2.0 V across a 3.12k feedback resistor with the
output more positive than the inverting input. Thus, the
op amp will have a baseline output of 7 V (5 V + 2 V) and
will want to go 3 V higher than this to handle the
dynamic range of the signal. With a positive rail of only
+10 V, this is not possible.
One approach is to reduce the value of the feedback resistor by a factor of two resulting in a value of 1.54 k
(nearest 1% value). This will cut the generated voltages
described above in half and yield half the scale factor for
the log output (18.75 mV/dB). Thus the baseline output
will be 6 V and the dynamic range of the signal will be
1.5 V. This will generate a signal that at its maximum is
+5V
X1
0.1µF
2.0kΩ
12
LOW-PASS FILTER
1000pF
11
30pF
30pF
OFFSET-NULL
LADJFIL1 FIL2
360kΩ
360kΩ
10
9
LMHIINHI
200Ω
INLO
0.1µF
1
1.5kΩ
250kΩ
1.5kΩ
AD606
HIGH-END
DETECTORS
ISUM
2
MAIN SIGNAL PATH
11.15 dB/STAGE
3
12µA/dB
ONE-POLE
FILTER
2 µA/dB
9.375kΩ
9.375kΩ
4
5
2
TWO-POLE
2pF
SALLEN-KEY
FILTER
X2
2pF
VLOGBFINILOGCOMM
68
10 pF
1.54kΩ
+10V
1
0.1µF
7
AD8041
+5V
3
0.1µF
5
4
Figure 2. 0 V, +5 V +10 V Configuration
–2–
FINAL
LIMITER
7
6
LIM OUT
LMLOOPCM
VOUT

2.5 V below the positive rail of the op amp. Another technique for shifting the baseline dc level at the output of
the op amp will be discussed in the next section.
Figure 3 shows the response of the AD8041 output of the
circuit in Figure 2. The input signal is a sinusoidal burst
at a frequency of 10.7 MHz and an amplitude of 10 mV
p-p. The 10 pF capacitor across the feedback resistor
was selected to provide the best compromise between
filtering the carrier ripple and minimizing the rise and
fall times of the output. In comparison Figure 4 shows
the VLOG output of the AD606 as conventionally operated using the same input signal. It can be seen that by
using the speed up technique the rise and fall times
have decreased from approximately 400 ns to 50 ns or
about a factor of eight improvement. Other signal amplitudes of 1 mV, 100 mV and 1 V p-p were also used as an
input with similar results.
900mV
10mV
DUAL SUPPLY CONFIGURATION
If the baseline output of the op amp is desired to be at or
near ground, then the supplies for the AD606 can be
shifted down by 5V so that it operates with ground as its
most positive rail and -5V as its lower rail. All other pins
of the AD606 that are normally directly connected to a
power supply must also be shifted down by 5V from the
standard connections. This will change the location of
the bypass capacitors. The pins that are newly assigned
to -5V require bypass capacitors (to ground), while those
newly assigned to ground do not. In single-ended circuits, an undriven input to the AD606 can still have its
capacitor connected to ground in this configuration.
With the top rail of the AD606 at ground, ISUM also
wants to be biased at ground. So besides the ground
and -5V supplies, an additional +5V supply is required to
power the op amp I to V converter. Fig. 5 is a schematic
of this configuration. The non-inverting input of the op
amp is biased at ground which forces the inverting input
(and ISUM) to also be at ground.
The dc standing current into ISUM will create a baseline
voltage of 1V (650 µA x 1.54K) at the output of the op
amp. More operating headroom in the op amp can be
obtained if this voltage can be shifted down. Providing
current into the summing node can accomplish this voltage shift.
100mV/div
TRIG'D
–100mV
Figure 3. V
6.25mV
5mV/div
100ns/div
Response of the Circuit in Figure 2
OUT
10mV
If it is desired to shift this voltage down by 1V (for a
baseline voltage at ground), then a current of 1V/1.54k or
650 µA will have to be input into the summing node. If
the +5V supply is used to provide this current, then R1
should be 5V/650 µA or 7.68k (nearest 1% value). Fig. 5
shows the location of this resistor in the circuit. This assumes that the +5V supply is accurate enough for this
application. For greater accuracy, a voltage reference
can be used and similar calculations performed to obtain the proper component values. This same technique
for level shifting the baseline output of the op amp can
be used in the all positive supply configuration described above.
If the intrinsic 37.5 mV/dB scaling of the AD606 is desired, then the feedback resistor can be increased to
3.16k. As mentioned above, this will create a signal with
a 3V dynamic range which requires more headroom.
The AD8041 can handle a 0 to 3V single with a +5V positive rail, but to use op amps with less headroom like the
AD8047, the baseline can be shifted further negative to
provide more operating headroom. In general, downstream circuitry will dictate the desired dynamic range
and dc operating point.
TRIG'D
–43.75m
Figure 4. V
200ns/div
Response of AD606
LOG
–3–

CONCLUSIONS
The methods presented above increase the speed of the
logarithmic output response of the AD606 by approximately a factor of eight over the response of the integrated output circuit. A high speed op amp I to V
converter replaces the intrinsic op amp in the AD606 to
provide the improved response. The slope and intercept
variations are affected, but can be compensated in some
systems. Two power supply configurations are described along with techniques for varying the scale factor and the dc operating point of the output.
INPUT
49.9Ω
0.1µF
INLO
0.1µF
–5V
16
COMM
30kΩ
1.5kΩ
250kΩ
1.5kΩ
AD606
1
–5V
0.1µF
15
PRUP
REFERENCE
AND POWER UP
30kΩ
HIGH-END
DETECTORS
ISUM
2
0.1µF
14
3
12µA/dB
R1
7.68kΩ
13
VPOS
X1
9.375kΩ
ONE-POLE
FILTER
2 µA/dB
45
+5V
0.1µF
2.0kΩ
12
30pF
30pF
OFFSET-NULL
LOW-PASS FILTER
MAIN SIGNAL PATH
11.15 dB/STAGE
2pF
9.375kΩ
2pF
10 pF
1.54kΩ
1
2
AD8041
3
0.1µF
4
–5V
1000pF
11
LADJFIL1 FIL2
360kΩ
360kΩ
TWO-POLE
SALLEN-KEY
FILTER
X2
VLOGBFINILOGCOMM
68
–5V
+5V
0.1µF
7
6
5
0.1µF
10
FINAL
LIMITER
7
LMHIINHI
0.1µF
VOUT
9
LMLOOPCM
200Ω
000000000
LIM OUT
0.1µF
–5V
Figure 5. –5 V, 0 V, +5 V Configuration
–4–
PRINTED IN U.S.A.