User`s guide

Group Delay vs. Zoom Size
RF=1 GHz RF=20 GHz
Zoom = off +/- 1 ns +/- 3 ns
Zoom = 8 +/- 300 ps +/- 1.2 ns
Zoom = 16 +/- 275 ps +/- 1.1 ns
Zoom = 32 +/- 250 ps +/- 1.0 ns
Zoom = 64 +/- 200 ps +/- 750 ps
Zoom = 128 +/- 150 ps +/- 500 ps
Table 1: Typical group delay
accuracy versus zoom size with no
averaging. Excludes mismatch error.
Phase and Amplitude vs.
Zoom Size
RF = 1 GHz
Phase Ampl.
Zoom = off +/- 1° –
Zoom = 8 +/- .5° +/- .1 dB
Zoom = 16 +/- .4° +/- .1 dB
Zoom = 32 +/- .3° +/- .05 dB
Zoom = 64 +/- .2° +/- .05 dB
Zoom = 128 +/- .2° +/- .05 dB
Table 2: Typical phase and
amplitude accuracy versus zoom
size with no averaging, at 1 GHz.
Excludes mismatch error.
Phase and Amplitude vs.
Zoom Size
RF = 20 GHz
Phase Ampl.
Zoom = off +/- 5° –
Zoom = 8 +/- 2° +/- .2 dB
Zoom = 16 +/- 1° +/- .15 dB
Zoom = 32 +/- .8° +/- .15 dB
Zoom = 64 +/- .7° +/- .1 dB
Zoom = 128 +/- .5° +/- .1 dB
Table 3: Typical phase and
amplitude accuracy versus zoom
size with no averaging, at 20 GHz.
Excludes mismatch error.
17
Accuracy
Considerations
Typical Accuracy
Tables 1, 2, and 3 show typical
measurement
repeatability/accuracy versus
zoom size at 1 and 20 GHz.
These numbers do not take into
account mismatch errors, or
frequency normalization error,
so they represent best case
conditions (mismatch is
discussed in more detail on page
18). Each data point in the table
was derived from two non-
frequency translating
measurements of a cable, each
with a 100 MHz span centered
at the carrier frequency. The
first measurement was used as
the reference trace for a
normalized measurement.
The measurement was then
repeated with normalization on,
to determine how much trace
noise was present. These values
are typical of what would be
achieved if normalizing to the
DUT output frequency, which
optimizes for relative group
delay measurements. The RF
carrier power was set to 0 dBm
(the maximum allowable input),
and no averaging was used.
Noise Reduction
Averaging
The results in Tables 1, 2 and 3
can be improved by averaging,
at the expense of measurement
time. N averages increase the
accuracy and measurement
speed by a factor of N. The
accuracy improvement
expressed in dB is 10*log(N).
A 3 dB improvement requires
doubling the number of averages.
Smoothing
Smoothing may also be applied
to the trace to reduce the
excursions caused by noise. The
smoothing function is
implemented as a variable-
length median smoothing filter.
The number of trace points used
to compute the median is set by
the smoothing value. Smoothing
tends to work best when the
number of trace points is large.
The smoothing function is
accessed via the MENU keys,
under the Traces menu. Press
avg,hld, followed by smooth
ON|OFF until ON is underlined.
Enter the smoothing value
desired. The results of the
smoothing can be compared to
the original trace by toggling
between on and off. While
smoothing can be a useful tool
to reduce displayed trace noise,
too much smoothing affects the
desired data, by reducing
non-noise variations caused by
the DUT itself.