
Application Notes
(Continued)
The reading of the ADC obviously doesn’t contain the refer-
ence voltage source V
REF
anymore, but the contribution of
the offset current remains present. This measurement is
performed during normal operation. Therefore, it eliminates
voltage reference variations over temperatures, as opposed
to strategy 1. Also offset variations in the op amp are elimi-
nated in this strategy.
Strategy 3: Complete Elimination of Temperature
Spread in Pedestal Voltage
The most accurate measurement is strategy 3, which elimi-
nates the temperature variation of both the reference voltage
V
and the offset current I
. In this strategy, the pedestal
voltage is measured regularly during operation of the phone,
and stored in the phone memory. For each power measure-
ment, the stored value is digitally subtracted from the
(analog-to-digital converted) detector output signal. Since it
measures the pedestal voltage itself for calibration it com-
pensates both for the reference voltage V
as well as for
the offset current variation I
. The frequency of the ‘calibra-
tion measurement’ can be significantly lower than those of
power measurements, depending on how fast the tempera-
ture of the device changes.
The calibration measurement procedure can be explained
with the aid of
Figure 1
, which depicts a typical power
measurement setup using the LMV232. In normal operation,
the two PA’s in the setup will never be active at the same
time. One PA will produce the required transmit power, while
the other one is off, (disabled) and produces no power. The
pedestal voltage should be measured in the absence of RF
power. This can be achieved by switching the Band Select
(BS) pin such that the LMV232 input is selected where the
disabled PAis connected to. The pedestal voltage at no input
power can be read at the output pin.
Using the Band Select (BS) control pin of the LMV232:
Select the RF input that is connected to the disabled PA,
by the BS pin.
Measure the detector output.
Store the result in the phone memory.
Subtract the stored value from each detector power read-
ing, until a new update is performed.
Important advantages of this approach are that no factory
trim is required and the temperature drift of the pedestal can
be cancelled almost completely as well as the part-to-part
spread. The remaining error is determined by the resolution
of the ADC. A slight disadvantage is that on average more
than one detector reading is required per power measure-
ment. This overhead though can be made almost negligible
in normal circumstances.
20127808
FIGURE 7. Strategy 3: Calibration during normal
operation
L
www.national.com
12