I've been in Sweden (again!) last week, so have only followed this from
afar.......
Frequency dividers using 'linear' parts are quite old. The simplest of them uses
an amplifier at the divided frequency feeding a mixer: the input to the mixer
is the fequency to be divided, and the output feeds the amplifier. Take a
divider from 1MHz to 100kHz as an example: noise at 100kHz mixes with the 1MHz
to give 900kHz: this mixes with the 1MHz to give 100kHz, which is amplified and
fed to the mixer, where it mixes with the 1MHz to give 900kHz and so on. Used
to be a technique used in frequency synthesisers of the add-multiply-divide
type, before PLLs with programmable dividers came along. P147 of Pappenfus has
an example of a very similar type.
My first thoughts were that there was an injection locked oscillation, but it
seems that it is a system problem, rather than an equipment one. Parametric
dividers can be made with a varactor: see Manassewitsch, Frequency Synthesisers,
Theory and Design, second edn., p376 et seq.: one would expect a power
dependency in the effect. Carrier storage devices can also be used to make a
divider - see Manassewitsch again. A diode with high charge storage could appear
from 'rusty bolt effect', the answer for which is 'Anti-Intermodulation Grease'.
'Sub harmonics' is a term widely used, generally inappropriately when applied to
RF....
73
Peter G3RZP
|