USER’S MANUAL__________________________________________________________________
80 __________________________________________________________________ M211322EN-D
There are many reasons for staying clear of the Nyquist frequency
multiples. Most of these considerations apply to all types of digital
processors, and are not specific to the RVP900.
As an example of what can go wrong at the Nyquist frequencies, suppose
that an intermediate frequency of 35 MHz was used. This is only 1 MHz
away from the (approximately) 36 MHz sampling rate. The external anti-
alias filter must now be designed more carefully, since a spurious input
signal at 37 MHz is aliased into the valid 35 MHz band. If the valid signal
bandwidth were 2 MHz, then the anti-alias filter has the difficult task of
passing 34 MHz to 36 MHz free of distortion, while rejecting everything
above 36 MHz. The filter's transition zone has to be very sharp, and this is
difficult to achieve.
Another problem that arises with a 35 MHz IF on a magnetron system, is
the RVP900 computation of AFC. If the processor can not distinguish
37 MHz from 35 MHz, then it can not tell the difference between the
STALO being correctly on frequency, versus being 2 MHz too high. The
symmetric AFC tracking range is reduced to the very small interval
34 MHz to 36 MHz.
For similar reasons (that is, transition band width), the digital FIR filter
also becomes difficult to design when its passband is near a Nyquist
multiple. But there is an additional constraint that the digital filter should
have a very large attenuation at DC. This is so that fixed offsets in the A/D
converter do not propagate into the synthesized "I" and "Q" data. Since
36 MHz is aliased into DC, we are left with the contradictory requirements
of a zero very close to the edge of the filter's passband.
3.3 RVP902 Main Chassis
3.3.1 RVP902 Main Chassis Overview
0916-023
Figure 21 RVP902 Main Chassis
To Next Page
Loading...