LTC5596
20
5596f
For more information www.linear.com/LTC5596
applicaTions inForMaTion
High Accuracy Power Measurement
The power measurement accuracy achieved using a power
detector is not only determined by the performance of the
power detector device itself, but also by the approach/
methods used to interpret the DC power detector output
signal. This can be understood by considering Figure 8.
response using linear regression over a suitable power
range (where the detector response is close to linear).
Better accuracy/smaller errors are obtained if SLOPE and
PINTERCEPT are determined for:
• Each detector device individually
• Each operating temperature
• Each operating frequency
To achieve the best accuracy, it is recommended to de-
termine SLOPE and PINTERCEPT for each individual unit,
requiring a 2-point factory calibration. When temperature
drift effects are to be included, SLOPE and PINTERCEPT need
to be determined at different operating temperatures and
the system needs to incorporate a temperature sensor to
determine which parameter values to use for the current
operating temperature.
The LOG-linearity error curves in the Typical Performance
Characteristics section were obtained using linear regres-
sion, applied to the response of the individual detector
devices at T = 25°C. For frequencies up to 28GHz, the
input power range from –37dBm to –5dBm was used. The
resulting LOG-linearity error tends to have larger negative
values than positive values. To center the error curves
within the ±1dB range, an additional 0.5dB was added
to the PINTERCEPT parameter. This slightly increases the
measurement error at T = 25°C, but results in a smaller
error over the full temperature range. The calculated
LOG-slope and LOG-intercept numbers are displayed in
the tables on page 3 and 4.
A better measurement accuracy is achieved if the inter-
preter uses the actual detector response at T = 25°C as
model for the detector, instead of the perfect linear-in-dB
response described above. The resulting measurement
error, the temperature drift error, equals:
Temperature Drift Error = [VOUT(T) – VOUT(25°C)]/SLOPE
A system that achieves this measurement error should
store the full output voltage vs input power response of
the detector with suitable resolution. The error curves
displayed on page 10 and 11 represent the achieved power
measurement accuracy using this configuration.
PMEAS
VOUT
ACT
SIGNAL
POWER
DETECTOR
INTERPRETER
(ESTIMATOR)
LF DETECTOR
OUTPUT VOLTAGE
MEASURED
INPUT POWER
Systems for accurate power level measurements on RF
signals can conceptually be thought to consist of two
elements:
• A high accuracy power detector (like the LTC5596),
converting the power level of an RF signal into a DC
voltage or current;
• An interpreter (also called an estimator), translating
the DC output voltage or current of the power detector
back to a power level.
In Figure 8, PMEAS represents the power level measured
by the system, i.e. the power level the system thinks is
present at its input, while PACT represents the actual power
level present at the detector input. The power measurement
error thus equals the difference: PERR = PMEAS – PACT.
The more the interpreter knows about the operating
conditions and transfer of the detector, the smaller the
measurement error that can be achieved. For example,
the interpreter may assume that the detector response is
perfectly linear in dB, such that the relationship between
input power and output voltage is a straight line:
VOUT = SLOPE • (PMEAS - PINTERCEPT)
This results in a power measurement error equal to:
LOG-Linearity Error = VOUT/SLOPE +PINTERCEPT – PACT
The parameters SLOPE and PINTERCEPT, the LOG-slope and
LOG-intercept, are best obtained from the actual detector
Figure 8. Power Measurement Concept