Specifications

17
Error correction can be applied to the measure-
ments discussed in this note to reduce the meas-
urement uncertainty. A full two-port calibration
was used for the measurement examples (except
where noted) to provide the best measurement
accuracy of both transmission and reflection meas-
urements of two-port devices. When a full two-port
calibration is applied, the dynamic range and accu-
racy of the measurement is limited only by the sys-
tem noise and stability, connector repeatability,
and the accuracy to which the characteristics of
the calibration standards are known.
In some instances it may be more convenient to
perform a response calibration to remove the fre-
quency response errors of the test setup for trans-
mission only measurements when extreme accura-
cy is not a critical factor. Likewise, an S
11
one-port
or S
22
one-port calibration to remove directivity,
source match and frequency response errors may
be more convenient for reflection only measure-
ments when the AUT is well-terminated.
Transmission measurements
For a gain measurement, the three major sources
of error are the frequency response error of the
test setup, the source and load mismatch error
during the measurement, and the dynamic accuracy.
A simple response calibration using a thru connec-
tion significantly reduces the frequency response
error which is usually the dominant error in a
transmission measurement. For the greatest accu-
racy, a full two-port calibration can be used which
also reduces the uncertainty in the measurement
caused by the source and load match.
Dynamic accuracy is a measure of the receiver’s
performance as a function of the incident power
level and has an effect on the uncertainty of a gain
measurement. This is because the receiver detects
a different power level between calibration and
measurement. The effects of dynamic accuracy on
a gain measurement are negligible (less than 0.5 dB)
as long as the network analyzer is operating below
the specified 0.1 dB compression level.
A gain drift measurement is subject to the same
errors as a gain measurement. Another factor that
could be significant is the transmission tracking
drift of the system. This drift is primarily caused
by the change in the temperature of the test setup
between calibration and measurement. To mini-
mize this effect, allow the instrument to stabilize to
the ambient temperature before calibration and
measurement.
A reverse isolation measurement is subject to the
same errors as a gain measurement. In addition, if
the isolation of the AUT is very large, the transmit-
ted signal level may be near the noise floor or
crosstalk level of the receiver. To lower the noise
floor, a decreased IF bandwidth may be necessary.
When crosstalk levels begin to affect the measure-
ment accuracy, a response and isolation calibration
or a full two-port calibration (including the isola-
tion part of the calibration) removes the crosstalk
error term. When performing the isolation part of
the calibration it is important to use the same
averaging factor and IF bandwidth during the cali-
bration and measurement.
For deviation from linear phase measurements, the
phase uncertainty is calculated from a comparison
of the magnitude uncertainty (already discussed
for gain measurements) with the test signal magni-
tude.
Appendix B—Accuracy considerations