Jackson Labs PHASE STATION 53100A, Руководство пользователя

Страница: 69 / 225

PhaseStation 53100A User’s Manual V1.04 25-Apr-2022 69
Understanding instrument spurs
While TimeLab can sometimes overlook true spurs and report nonexistent ones, a third
type of measurement error may need to be considered as well: the presence of spurs
which originate within the instrument itself.
The spur-free dynamic range (SFDR) of the ADCs used in the 53100A is approximately
100 dB, which determines the corresponding specification limit of the instrument as a
whole. In principle, your noise plots may contain fictitious spurs as high as -100 dBc, but
the 53100A’s typical performance is much better than this. Any spurs with amplitudes
above -120 dBc that remain visible as the measurement converges are likely to be “real”
in the sense that they originate from the DUT, from the reference, or from
environmental RFI.
Regardless of the equipment used, spur-free observations of AM noise and phase noise
become challenging below -130 dBc. There are simply too many potential sources of
coherent artifacts to rule out in most cases. Still, spurs that originate within the 53100A
can sometimes be identified manually. An informal technique that may help determine
whether a given spur or spur cluster originates within the 53100A hardware itself is
illustrated in the side-by-side plots below.
At left, the phase noise of a crystal oscillator has been measured over the course of
several hours. Most of the visible spurs are below the 53100A’s specified limits, so it
isn’t clear which ones the instrument may be responsible for.
However, because complex signals are used throughout the 53100A’s DSP pipeline, we
can take advantage of a mathematical observation.
12
While the cross spectrum of a
physically-realizable signal should fall on the real axis of the Argand plane, at least some
instrument spurs are caused by data-conversion errors that can show up elsewhere in
phase space. Like random ADC noise, the numerical distribution of these errors is not
12
See Rubiola and Vernotte, The cross spectrum experimental method , and Grove
et al.,
Direct-digital phase noise measurement
Figure 30: Imaginary and real components of the cross-correlated noise spectrum