User`s guide

10
For the zoom FFT mode, the
unambiguous range is further
reduced by a factor of two to
0±(.25/fmod). If the absolute
delay is beyond these limits, it
will be displayed incorrectly.
For example, if the absolute
delay of a DUT and test setup
was 273 ns, and the zoom FFT
mode was used with a
modulation frequency of 1 MHz,
the program would measure
–227 ns (273 ns - .5/1 MHz). If
the direct demodulation mode
was used (zoom FFT off), the
program would measure the
absolute delay correctly, since
the maximum delay range in
this case would be ±500 ns.
The maximum measurable
delay range must be considered
when measuring the absolute
delay of a DUT if a normalized
measurement is done with
significant reference delay. The
reference delay and the delay of
the DUT must be less than the
unambiguous range. For
example, if the unambiguous
range is ±500 ns, and the
reference sweep had 373 ns of
delay, then the DUT delay must
be less than 127 ns. Beyond
this, the delay value will be
negative (until it wraps around
a second time).
The only way to increase the
unambiguous measurement
range is to lower the modulation
frequency. However, this
reduces the aperture as well,
which increases measurement
noise. If the measured absolute
delay of your device changes
significantly with decreasing
modulation frequency
(assuming normalization is
done), then the electrical length
of the device is too long for a
valid measurement at that
modulation frequency. If you
know the approximate electrical
length of your device and can
therefore infer how many phase
wraps occurred, then an invalid
absolute group delay
measurement can be corrected.
If direct demodulation (zoom
FFT off) was used, add N ×
(1/fmod); if the zoom FFT was
used, add N × (.5/fmod), where
N is an integer which accounts
for the number of times the
measurement wrapped around.
Negative delay
Negative values of group delay
can occur when the
measurement wraps around as
described above, or when the
reference path length is longer
than the measurement path.
The program has a flag which
controls whether negative delay
numbers are corrected or not.
The default is set to show
negative numbers (no
correction). This allows
normalization to work correctly
when the reference path is
longer than the measurement
path. For example, if the
reference measurement had -4
ns of absolute delay and the
DUT measured 56 ns, then the
normalized measurement would
correctly show 60 ns of delay (56
ns - (–4 ns)).
When the group delay flatness
of the cables is not a significant
contribution to measurement
error, the normalization process
should be done with both the
input and output cables
present. If cable contribution is
likely to be a significant factor
over one or both of the
frequency ranges used in the
measurement, then the
normalization can be done with
only one of the cables. The other
cable can then be measured
separately over its intended
frequency range, using the
direct phase measurement
(either with a vector network
analyzer or the HP 71500A),
and its delay contribution
manually removed from the
normalized test data. The group
delay program currently does
not incorporate a two-part
normalization such as this.
Measuring the second cable in
this way is necessary for an
absolute measurement, but may
not be needed for relative
measurements if its
contribution would not be a
significant factor in the
measurement.
Unambiguous delay range
The nature of the modulation
technique for measuring group
delay is such that absolute
group delay values are valid
only for 0±(.5/fmod), where fmod
is the modulation frequency
used for the measurement. This
corresponds to 0±180° of phase
shift. After +180°, the
measurement will wrap around
to –180°, which changes the
sign of the group delay from
positive to negative. This is
equivalent to subtracting
1/fmod worth of delay.