Stability; Resolution; Noise - Hameg HM5530 Handbuch

the instrument may be swept and displayed in „full span" (SF
= 3000 MHz), in this mode the frequency of a spectral compo-
nent may be determined roughly. Subsequently, this frequency
can be shifted to the screen center by changing the CENTER
frequency, then the SPAN is decreased, thus the frequency
resolution increased.
The smaller the SPAN, the narrower the fi lter bandwidth (RBW),
the better the accuracy of frequency measurements, because
the display and the MARKER accuracies are increased.
In the „Zero Span" mode and selecting the smallest bandwidth,
it is suffi cient to tune the (unmodulated) signal, displayed as a
horizontal baseline, with the CENTER adjustment for maximum
amplitude and read the frequency from the readout. The analyzer
operates as a selective voltmeter with selectable bandwidth.

Stability

It is important that the frequency stability of the analyzer sur-
passes that of the signal. The frequency stability depends upon
the stability of the fi rst local oscillator (1st LO). One must discri-
minate between short-term and long-term stability. Residual fm
is a measure of the short-term stability. Noise side bands are
a measure of the spectral purity of the 1st LO and contribute
to the short-term (in)stability; they are characterized by their
attenuation in dB and their distance in Hz from the signal to be
analyzed with respect to a specifi ed fi lter bandwidth.
The long-term stability of a spectrum analyzer is mainly deter-
mined by the frequency drift of the 1st LO; it is a measure of
how much the frequency may change within a predetermined
time period.

Resolution.

Prior to measuring the frequency of a signal with a spectrum
analyzer, the signal must be detected and resolved. Resolution
means the signal resp. the spectral line must be separated from
neighbouring signals within the spectrum being analyzed. This
ability of resolution is a decisive criterion in many spectrum
analyzer applications. The resolution is determined by:
– sweep time
– span (dispersion)
– 6 dB bandwidth of the narrowest amplifi er stage resp. fi lter.
The 6 dB bandwidth of the narrowest amplifi er resp. fi lter, Gauss
behaviour assumed, is called the resolution bandwidth. This
is the smallest bandwidth which can be displayed if the other
parameters (sweep time, span) are varied.
The bandwidth and the slope of the if fi lters are thus the important
characteristics which determine whether two adjacent spectral
lines of widely different amplitude can be resolved. In general,
the bandwidth is defi ned as the –3 dB bandwidth, for spectrum
analyzers it is customary to specify the –6 dB bandwidth which
also applies to the HM5530. The different bandwidth defi nitions
are to be borne in mind when comparing instruments. The ratio of
the bandwidth at –60 dB to the bandwidth at –3 dB is defi ned as the
form factor; the smaller the form factor, the better the capability
of the analyzer to separate two adjacent spectral lines.
If e.g. the form factor of a fi lter in the analyzer is 15 :1, two spec-
tral lines differing in amplitude by 60 dB, must be at least 7.5
times the fi lter bandwidth apart in frequency if they are still to be
recognized as two signals, otherwise they will merge and appear
as a single signal.
However, the form factor is but one parameter infl uencing
the separation of spectral lines of different amplitude and
S p e c t r u m a n a l y z e r s p e c i f i c a t i o n s
frequency; the residual FM and the spectral purity of the inter-
nal oscillators are as important, because they generate noise
sidebands, thereby reducing the achievable resolution. Noise
sidebands will show up at the base of the if fi lter display and
deteriorate the stopband behaviour of the fi lters.
If the narrowest if bandwidth is 9 kHz, the smallest frequency
distance possible between two spectral lines is also 9 kHz if
they are still to be recognized as separate. The reason is that,
when detecting a signal, the spectrum analyzer displays its own
if fi lter shape while sweeping the frequency. As the resolution
is mainly dictated by the if fi lter bandwidth, one might assume
that infi nite resolution will be obtained with an infi nitely nar-
row fi lter bandwidth. As mentioned above, the residual fm of
the oscillators also limits the resolution and determines the
narrowest practical if bandwidth. If the residual fm is 9 kHz,
e.g., the narrowest practically useful if bandwidth will be also
9 kHz if a single signal is to be measured. An if fi lter with still
lower bandwidth would show more than one spectral line or a
jittery display, depending upon the sweep speed, also a partly
complete display is possible.
There is another important limitation to the narrowest practical
if fi lter bandwidth: the frequency sweep speed relative to the if
fi lter bandwidth selected. The narrower the fi lter, the slower the
sweep speed; if the sweep speed is too high, the fi lter can not
respond fast enough, and the amplitudes of the spectral lines
will be incorrectly displayed, in general too low.
A socalled optimum resolution is defi ned by:
optimum resolution =
A socalled optimum resolution bandwidth is defi ned by:
optimum resolution bandwidth =
For very long sweep times both become identical.
The optimum resolution bandwidth for pulsed signals is:
Optimum (–3 dB) bandwidth for pulsed signals ≤0.1 pulse
width.
If the bandwidth is narrower, the amplitudes of the side lobes
will be displayed too low. With the optimum bandwidth, there
are sharp nulls and a correct spectrum display. If the bandwidth
is too large, the side lobes will become averaged, thus less
pronounced, the nulls will be hardly discernible, the spectrum
distorted.

Noise

The sensitivity is a measure of the ability of a spectrum analyzer
to detect small signals. The maximum sensitivity is limited by
its internal noise. There are two kinds of noise: thermal and
non-thermal noise. Thermal noise is given by:
PN = K x T x B
PN: Noise power in watts
K: Boltzmann's constant (1.38 x exp - 23 Joule/K)
T: absolute temperature
B: Bandwidth
The equation shows that the noise power is directly proportional
to bandwidth. Hence reducing the fi lter bandwidth by a decade
SQRT Span in Hz
—————————
Sweeptime in s
0,66 x SQRT Span
—————————
Sweeptime
Subject to change without notice
37