Features Of Spectrum Analyzers; Frequency Measurement; Stability; Resolution - Hameg Instruments HM5510 Manuel

F e a t u r e s o f S p e c t r u m A n a l y z e r s
Zero span operation
If the sweep is switched off the LO will stay at a frequency which
is 1369,3 MHz above the input frequency, it functions like a radio
and displays only this one frequency and such neighbouring
frequencies which fall into the bandwidth of the if fi lter.
Normal operation
In normal operation the sweep sawtooth sweeps the LO
through the selected span range. If a span of e.g. 1000 MHz
was chosen and the center frequency was 500 MHz, the dis-
play would start on the left hand side of the display at 0 Hz
and sweep up to 1000 MHz at the right hand side. The center
would correspond to 500 MHz.
As the response time of a fi lter depends on its bandwidth and
shape the sweep must not be too fast, otherwise too low am-
plitudes and distorted spectral lines may result. If unsuitable
combinations of span, resolution bandwidth are chosen and
UNCAL will be displayed.

Features of Spectrum Analyzers

The main applications of spectrum analyzers start where
the limited analysis performance of scopes end. As mentio-
ned spectrum analyzers excel especially by their enormous
dynamic range which, together with logarithmic amplitude
display allow to show several orders of magnitude on the
same display.

Frequency measurement

As the frequency scale of modern spectrum analyzers is
derived from a highly accurate and stable crystal oscillator
very precise frequency measurements are possible. First a
coarse display with large span will show the frequency to
be measured, this can then be shifted to the display center
while the span is reduced and the smallest RBW selected at
the same time, increasing the accuracy. It is also possible
to select zero span and minimum RBW and then turn the
center frequency control knob until the maximum amplitude
is reached: the frequency can then be read from the center
frequency display.

Stability

The frequency stability of a spectrum analyzer should be much
better than that of the input signal. The 1
termine the quality. Most important is the short term stability
including noise, residual FM and spectral purity.

Resolution

The smallest bandwidth and the fi lter slopes of the if bandpass
fi lter determine the available resolution of a spectrum analyzer.
The defi nition of bandwidth is the frequency span between the
– 3 dB points. The relationship between the – 60 dB bandwidth
and the – 3 dB bandwidth is called form factor.
32
Subject to change without notice
In addition to the form factor residual FM and spectral purity
of all oscillators will also affect the capability of a spectrum
analyzer to separate neighbouring frequencies. The noise side
bands created by residual FM and insuffi cient spectral purity
will deteriorate the stop band attenuation of the fi lters.
With the smallest RBW of 20 kHz 2 frequencies must be more
than 20 kHz apart if they should be recognized as separate. The
spectrum analyzer displays its own IF fi lter curve if there is any
signal. It appears that infi nite resolution should be possible
with an infi nitely small RBW. In practice this does not happen.
The stability of the oscillators sets one limit, if the signal
moves too much with frequency it will move back and forth
with a very narrow bandwidth fi lter, no usable display would
result, only jitter. Residual FM of the oscillators would cause
the display of several spectral lines instead of one. The second
practical limit is given by the relationship of fi lter bandwidth
and response time, the narrower the fi lter the slower must
the frequency be swept across, otherwise the fi lter will yield
a decreased amplitude and a distorted display.

Noise

The maximum sensitivity of a spectrum analyzer is determined
by the noise level, to be differentiated between thermal noise
and non-thermal noise.
Thermal noise is given by: P
K = Boltzmann's constant
T = absolute temperature
B = bandwidth
Noise is hence directly proportional to bandwidth, thus if the
fi lter bandwidth is reduced by a factor of ten the noise will de-
crease by 10 dB. The sensitivity increases by the same factor.
All other noise sources in a spectrum analyzer are regarded
as non-thermal. Sources of such non-thermal noise are e.g.:
distortions caused by nonlinear behaviour, mismatches, hf
leakage. The quality = noise fi gure of a system is given by
the noise fi gure of the non-thermal sources plus the thermal
noise. This visible noise limits the sensitivity of the instru-
ment. When comparing spectrum analyzers it is important
to compare identical instrument settings, i.e. the bandwidths
must be identical. Although a spectrum analyzer covers a
very broad frequency range the noise depends mainly on the
IF fi lter bandwidth, the detector following the IF sees only the
noise passed by it.
Video fi lter
st
LO' s properties de- The measurement of small signals close to the noise level
becomes diffi cult. In order to separate the signal more from
the noise a video fi lter may be inserted following the detector.
This fi lter typically has a bandwidth of a few kHz and averages
the noise. Here it also applies that small bandwidth fi lters
respond slowly, hence it is advisable to switch this fi lter off
if the IF bandwidth becomes small compared to the scan
selected which means that the sweep speed becomes too
high, otherwise the amplitudes will be displayed too low. An
UNCAL light will indicate any unfavourable combinations of
settings.
The smaller the form factor the better can adja-
cent frequencies be separated. E.g.: if the form
factor is 15:1 2 frequencies which differ in amplitu-
de by 60 dB must differ in frequency by at least the
factor of 7.5, if they should still be discernible as
separate, otherwise they will melt into one signal.
HINT
= K x T x B
noise