Table of Contents
Advertisement
purity. Noise sidebands are specified in terms of dB down
and Hz away from a carrier in a specific bandwidth. Long
term stability is characterized by the frequency drift of the
analyzers LOs. Frequency drift is a measure of how much
the frequency changes during a specified time (i.e., Hz/
min. or Hz/hr).
Resolution
Before the frequency of a signal can be measured on a
spectrum analyzer it must first be resolved. Resolving a
signal means distinguishing it from its nearest neighbours.
The resolution of a spectrum analyzer is determined by its
IF bandwidth.
The
IF bandwidth
is usually the 3dB
bandwidth of the IF filter. The ratio of the 60dB bandwidth
{in Hz) to the 3dB bandwidth (in Hz) is known as the shape
factor of the filter. The smaller the shape factor, the
greater is the analyzers capability to resolve closely spaced
signals of unequal amplitude. If the shape factor of a filter
is 15:1, then two signals whose amplitudes differ by 60dB
must differ in frequency by 7.5 times the IF bandwidth
before they can be distinguished separately. Otherwise,
they will appear as one signal on the spectrum analyzer
display.
The ability of a spectrum
analyzer to resolve closely
spaced signals of unequal amplitude is not a function of
the IF fifter shape factor only. Noise sidebands can also
reduce the resolution. They appear above the skirt of the
IF filter and reduce the offband rejection of the filter. This
limits the resolution when measuring signals of unequal
amplitude.
The resolution of the spectrum analyzer is limited by its
narrowest IF bandwidth. For example, if the narrowest
bandwidth is 10kHz then the nearest any two signais can
be and still be resolved is 10kHz. This is because the
analyzer traces out its own
IF band-pass shape as it
sweeps through a CW signal. Since the resolution of the
analyzer is limited by bandwidth, it seems that by reducing
the IF bandwdith
infinitely, infinite resolution will be
achieved. The fallacy here is that the usable IF bandwidth
is limited by the stability (residual FM) of the analyzer. If
the internal frequency deviation of the analyzer is 10kHz,
then the narrowest
bandwidth
that can be used to
distinguish a single input signal is 10kHz. Any narrower IF-
filter willresult in more than one response or an intermittent
response for a single input frequency. Apractical limitation
exists on the IF bandwidth as well, since narrow filters
have long time constants and would require execessive
scan time.
Sensitivity
Sensitivity is a measure of the analyzers' ability to detect
small signals. The maximum sensitivity of an analyzer is
limited by its internally generated noise. This noise is
basically of two types: thermal (or Johnson) and nonthermal
noise. Thermal noise power can be expressed as:
Py = kxTxB
where:
P, = Noise power in watts
k = Boltzmanns Constant (1.38x10-5 Joule/K}
T = absolute temperature, K
B = bandwidth of system in Hertz
As seen from this equation, the noise level is directly
proportional to bandwidth. Therefore, a decade decrease
in bandwidth results in a 10dB decrease in noise level and
consequently 10dB better sensitivity. Nonthermal noise
accounts for ali noise produced within the analyzer that is
not temperature dependent. Spurious emissions due to
nonlinearities of active elements, impedance mismatch,
etc. are sources of nonthermal noise. A figure of merit, or
noise figure, is usually assigned to this nonthermal noise
which when added to the thermal noise gives the total
noise of the analyzer system. This system noise which is
measured on the CRT, determines the maximum sensitivity
of the spectrum analyzer. Because noise level changes
with bandwith,
it is important,
when
comparing
the
sensitivity of two analyzers,
to compare
sensitivity
specifications for equal bandwidths.
A spectrum
analyzer sweeps
over a wide frequency
range, but is really a narrow band instrument. All of the
signals that appear in the frequency range of the analyzer
are converted to a single IF frequency which must pass
through an IF filter; the detector sees only this noise at any
time. Therefore, the noise displayed on the analyzer is
only that which is contained in the IF passband. When
measuring
discrete
signals,
maximum
sensitivity is
obtained by using the narrowest IF bandwidth.
Video Filtering
Measuring small signals can be difficult when they are
approximately the same amplitude as the average internal
noise level of the analyzer. To facilitate the measurement,
it is best to use video filtering. A video filter is a post-
detection low pass filter which averages the internal noise
of the analyzer. When the noise is averaged, the input
signal may be seen.
If the resolution bandwidth is very narrow for the span, the
video filter should not be selected, as this will not allow
the amplitude of the analyzed signals to reach full amplitude
due to its video bandwidth limiting property.
Spectrum Analyzer Sensitivity
Specifying sensitivity on a spectrum analyzer is somewhat
arbitrary. One way of specifying sensitivity is to define it as
the signal level when signal power = average noise power.
The analyzer always measures signal plus noise. Therefore,
when the input signal is equal to the internal noise level,
the signal will appear 3dB above the noise. When the
signal power is added to the average noise power, the
power level on the CRT is doubled (increased by 3qB)
because the signal power=average noise power.
Subject to change without notice
M8 5005/5006
Advertisement
Table of Contents
