HP 8753D - Page 303

result
in the
group delay
data. These
errors
can
be
signicant
for
long
delay
devices
.
Y
ou
can
verify
that 1

is

180

by
increasing the
number
of
points
or
narrowing
the
frequency
span
(or
both)
until the
group
delay
data
no
longer
changes
.
Figure
6-21. Rate
of Phase
Change V
ersus Frequency
When
deviations
from
linear
phase
are
present,
changing
the
frequency
step
can
result
in
dierent
values
for
group
delay
.
Note
that
in
this
case
the
computed
slope
varies
as
the
aperture
1f
is
increased
(see
Figure
6-22
).
A
wider
aperture
results
in
loss
of
the
ne
grain
variations
in
group
delay
.
This
loss
of
detail
is
the
reason
that
in
any
comparison
of
group
delay
data,
it
is
important
to
know
the
aperture
that
was
used
to
make the
measurement.
Figure 6-22. V
ariations in Frequency Aperture
In determining the group delay aperture, there is a tradeo between resolution of ne detail
and the eects of noise. Noise can be reduced by increasing the aperture, but this will tend to
smooth out the ne detail. More detail will become visible as the aperture is decreased, but the
noise will also increase, possibly to the point of obscuring the detail. A good practice is to use a
smaller aperture to assure that small variations are not missed, then increase the aperture to
smooth the trace.
6-36 Application and Operation Concepts