Application Notes
(Continued)
The pedestal voltage at 25C is subtracted from the output
voltage of each curve. Variations of the pedestal voltage
over temperature are thus included in the error.
The pedestal voltage variation itself consists of 2 error
sources. One is the variation of the reference voltage V
REF
.
The other is an offset current I
that is generated inside the
detector. This depicted in
Figure 4
. Depending on the mea-
surement strategy one or both error sources can be elimi-
nated.
The error sources of the pedestal voltage can be shown in a
formula for V
OUT
:
V
OUT
= V
REF
+ (I
OS
+ I
DET
) * R
FB
Where I
represents the intended detector output signal.
In the absence of RF input power I
equals zero. The
formula for the pedestal voltage can therefore be written as:
V
PEDESTAL
= V
REF
+ I
OS
* R
FB
For low input power levels, the pedestal variation V
PEDESTAL
is the dominant cause of error. Besides temperature varia-
tion of the pedestal voltage, which limits the lower end of the
range, the pedestal voltage can also vary from part-to-part.
By applying a suitable measurement strategy, the pedestal
voltage error contribution can be significantly reduced or
eliminated completely.
POWER MEASUREMENT STRATEGIES
This section describes the measurement strategies to re-
duce or eliminate the pedestal voltage variation. Which strat-
egy is chosen depends on the possibilities for a factory trim
and implementation of calibration procedures.
Since the pedestal voltage is the reference level for the
LMV232, it needs to be calibrated/measured at least once to
eliminate part-to-part spread. This is required to determine
the exact detector output signal. Because of process toler-
ances, the absolute part-to-part variation of the output volt-
age in the absence of RF input power will be in the order of
5 - 10%.All measurement strategies discussed eliminate this
part-to-part spread.
Strategy 1: Elimination of Part-to-Part Spread at Room
Temperature Only
In this strategy, the pedestal voltage is determined once
during manufacturing and stored into the memory of the
phone. At each power measurement this stored pedestal
level is digitally subtracted from the measured output signal
of the LMV232 during normal operation. The procedure is
thus:
Measure the detector output in the absence of RF power
during manufacturing.
Store the output voltage value in the cell phone memory
(after it is analog-to-digital converted).
Subtract the stored value from each detector output read-
ing.
The advantage of this strategy is that calibration is required
only once during manufacturing and not during normal op-
eration. The disadvantage is the fact that this method neither
compensates for the residual temperature drift of the refer-
ence voltage V
nor for offset current variations. Only
part-to-part variations at room temperature are eliminated by
this strategy. Especially the residual temperature drift nega-
tively affects the measurement accuracy.
Strategy 2: Elimination of Temperature Spread in V
REF
If software changes need to be reduced to a minimum and
the baseband chip has a differential ADC, strategy 2 can be
used to eliminate temperature variations of the reference
voltage V
REF
. One pin of the ADC is connected to FB and
one is connected to OUT (
Figure 6
).
The power measurement is independent of the reference
voltage V
REF
, since the ADC reading is:
V
OUT
-V
FB
= (I
OS
+ I
DET
) * R
FB
20127805
FIGURE 4. Pedestal Voltage
20127806
FIGURE 5. Strategy 1: Room Temperature Calibration
20127807
FIGURE 6. Strategy 2: Differential Measurement
L
www.national.com
11