It is interesting that in landline SSB, IMD was measured by loading the
channel with a noise signal: this had a 'slot' in the spectrum with no
noise in it, and the IMD was measured as the amount of noise appearing in
that slot. This is (provided the noise has the correct characteristics - a
fair amount of LF energy modulating it) a good simulation of the voice
signal. For similar reasons, some tx specs requiring 2 tone testing get
nasty with requirements for tone spacings of around 50Hz or less: this
really shows up power supply dynamic characteristic faults. There was some
stuff published on noise testing in the early 1970's and I know the
Russian adminstration is keen on it as a method of testing
commercial/professional SSB transmitters today
The concept of two tone testing originally developed in point to point HF
ISB work, where the degree of crosstalk into the other sideband needed
quantifying. This was very much 1930's stuff, and the triodes used then in
PA stages were generally fairly benign in the way the higher order IMD
products tailed off. I suppose the advantage of two tone testing is that it
is fairly simple and easy to do, but like several of these 'tried and
tested' methods, doesn't really give the answers we need for today's
equipment. I wonder how repeatable voice testing is unless you have a
standard voice - certainly, the words spoken can give interesting
variations in the energy per word and thus how hard the tx is driven.
73
Peter G3RZP
|