Agilent Technologies 4395A - Page 392

Network
Measurement
Basics
A
veraging
(Sweep
A
veraging)
A
veraging
computes
each
data
point
based
on
an
exponential
average of
consecutive sweeps
weighted
by
a
user-specied
averaging
factor
. Each
new sweep
is averaged
into the
trace
until
the
total
number
of
sweeps
is equal
to the
averaging factor
, for
a fully
averaged trace
.
Each
point
on
the
trace
is
the vector
sum of
the current
trace data
and the
data from
the
previous
sweep
.
A
high
averaging
factor
gives
the
best
signal-to-noise ratio
, but
slows the
trace
update
time
.
Doubling
the
averaging
factor
reduces
the
noise
by 3
dB.
The algorithm
used for
averaging
is:
A
(
n )
=
S
( n
)
n
+(1
0
1
n
) 2
A
(
n 0
1)
where:
A(n)
=
current
average
(1

n

F)
S(n)
=
current
measurement
(1

n

F)
F
=
average
factor
IF
Band
Reduction
IF
bandwidth reduction
lowers
the
noise
oor
by
reducing
the
receiver
input
bandwidth.
It
has
an
advantage
over
averaging
in
reliably
ltering
out
unwanted
responses
such
as
spurs
,
odd
harmonics
,
higher
frequency
spectral
noise
,
and
line-related
noise
.
Sweep-to-sweep
averaging,
however
,
is
better
at
ltering
out
very
low
frequency
noise
.
A
tenfold
reduction
in
IF
bandwidth
(from
200
Hz
to
20
Hz,
for
example)
lowers
the
measurement
noise
oor
by
about
10
dB
.
Another
dierence
between
sweep-to-sweep
averaging
and
variable
IF
bandwidth
is
the
sweep
time
.
A
veraging
displays
the
rst
complete
trace
faster
but
takes several
sweeps
to
reach
a
fully
averaged
trace
.
IF
bandwidth
reduction
lowers
the
noise
oor
in
one sweep
,
but
the
sweep
time
may
be
slower
.
Group
Delay
F
or
many
networks
,
the
linearity
of
the
phase shift
over a
range of
frequencies is
an
important
parameter
.
Group
delay
is
the measurement
of signal
transmission time
through a
test
device
.
It
is
dened
as
the
derivative
of the
phase characteristic
with respect
to frequency
.
Because
the
derivative
is
the instantaneous
slope (or
rate of
change of
phase with
frequency), a
perfectly linear
phase
shift
results in
a
constant
slope
,
and
therefore
a
constant
group
delay
(Figure
A
-8
).
A-16 Basic Measurement Theory