Data Sheet AD8315
Rev. D | Page 13 of 22
The intercept need not correspond to a physically realizable
part of the signal range for the log amp. Therefore, the specified
intercept is −70 dBV, at 0.1 GHz, whereas the smallest input for
accurate measurement (a +1 dB error, see Table 2) at this frequency
is higher, being about −58 dBV. At 2.5 GHz, the +1 dB error point
shifts to −64 dBV. is positioning of the intercept is deliberate
and ensures that the V
SET
voltage is within the capabilities of
certain DACs, whose outputs cannot swing below 200 mV.
Figure 33 shows the 100 MHz response of the AD8315; the
vertical axis does not represent the output (at pin VAPC) but
the value required at the power control pin, VSET, to null the
control loop.
1.5
1.0
0.5
0
–70dBV
1.416V @ –11dBV
0.288V @ –58dBV
ACTUAL
IDEAL
S
L
O
P
E
=
2
4
m
V
/
d
B
V
SET
100µV
–80dBV
–67dBm
1mV
–60dBV
–47dBm
10mV
–40dBV
–27dBm
100mV
–20dBV
–7dBm
1V (RMS)
0dBV
+13dBm (RE 50Ω)
V
IN
, dBV
IN
, P
IN
01520-032
Figure 33. Basic Calibration of the AD8315 at 0.1 GHz
CONTROLLER-MODE LOG AMPS
The AD8315 combines the two key functions required for the
measurement and control of the power level over a moderately
wide dynamic range. First, it provides the amplification needed
to respond to small signals in a chain of four amplifier/limiter
cells (see Figure 32), each having a small signal gain of 10 dB and a
bandwidth of approximately 3.5 GHz. At the output of each of
these amplifier stages is a full-wave rectifier, essentially a square
law detector cell that converts the RF signal voltages to a fluctuating
current having an average value that increases with signal level.
A further passive detector stage is added before the first stage.
These five detectors are separated by 10 dB, spanning some 50 dB
of dynamic range. Their outputs are each in the form of a
differential current, making summation a simple matter. It is
readily shown that the summed output can closely approximate
a logarithmic function. The overall accuracy at the extremes of
this total range, viewed as the deviation from an ideal logarithmic
response, that is, the log conformance error, can be judged by
referring to Figure 7, which shows that errors across the central
40 dB are moderate. Other performance curves show how
conformance to an ideal logarithmic function varies with
supply voltage, temperature, and frequency.
In a device intended for measurement applications, this current
is converted to an equivalent voltage, to provide the log (V
IN
)
function shown in Equation 1. However, the design of the AD8315
differs from standard practice in that the output must be a low
noise control voltage for an RF power amplifier not a direct
measure of the input level. Furthermore, it is highly desirable that
this voltage be proportional to the time integral of the error
between the actual input V
IN
and the dc voltage V
SET
(applied to
Pin 3, VSET) that defines the setpoint, that is, a target value for
the power level, typically generated by a DAC.
This is achieved by converting the difference between the sum
of the detector outputs (still in current form) and an internally
generated current proportional to V
SET
to a single-sided, current-
mode signal. This, in turn, is converted to a voltage (at Pin 4,
FLTR, the low-pass filter capacitor node) to provide a close
approximation to an exact integration of the error between the
power present in the termination at the input of the AD8315
and the setpoint voltage. Finally, the voltage developed across
the ground-referenced filter capacitor C
FLT
is buffered by a special
low noise amplifier of low voltage gain (×1.35) and presented at
Pin 7 (VAPC) for use as the control voltage for the RF power
amplifier. This buffer can provide rail-to-rail swings and can
drive a substantial load current, including large capacitors. Note
that the RF power amplifier is assumed to have a positive slope
with RF power increasing monotonically with an increasing
APC control voltage.
CONTROL LOOP DYNAMICS
To understand how the AD8315 behaves in a complete control
loop, an expression for the current in the integration capacitor
as a function of the input V
IN
and the setpoint voltage V
SET
must
be developed (see Figure 34).
3
1
RFIN
4
FLTR
7
VAPC
×1.35
I
SET
= V
SET
/4.15kΩ
SET
V
SET
V
IN
I
DET
= I
SLP
log
10
(V
IN
/V
Z
)
I
ERR
I
DET
C
FLT
SETPOINT
INTERFACE
LOGARITHMIC
RF DETECTION
SUBSYSTEM
01520-033
Figure 34. Behavioral Model of the AD8315
First, the summed detector currents are written as a function of
the input
I
DET
= I
SLP
log
10
(V
IN
/V
Z
) (3)
where:
I
DET
is the partially filtered demodulated signal, whose
exact average value is extracted through the subsequent
integration step.
I
SLP
is the current-mode slope and has a value of 115 µA per
decade (that is, 5.75 µA/dB).
V
IN
is the input in V rms.
V
Z
is the effective intercept voltage, which, as previously noted,
is dependent on waveform but is 316 µV rms (−70 dBV) for a
sine wave input.