William Donzelli wrote:
Still,
at the rates they were sending data in those days, FSK
could have been as simple as a solenoid on a capacitor, tied to
the frequency reference.
That was the normal way of doing things at that time. Some of us
can still do it that way...
Modem technology of the late 1930s and World War 2 era was far more
advanced that the simple trick described above. FSK and tone
modulation were both characterized by fairly complex equipments for
the time with active, tube circuitry. Many sets could also cram a few
telegraph or teleprinter channels onto one voice channel or pair of
wires.
There are some technical details about the modems on a page that Chuck
provided the link to in an earlier message:
http://www.codesandciphers.org.uk/cevent.htm
under the heading "The radio transmissions".
Extract:
"These will be exactly as transmitted in WW II. Multi-tone modulation on the
transmitted RF carrier. Two sets of tones were used, one representing a mark
or 1 in the Baudot code, the other representing a space or 0 in the Baudot
code. Three tones were used in each one and these tones were chosen to be not
harmonically related. Their frequencies were 540 cps + n * 180 cps. Actual
frequencies, Mark 900,1620,2340 Space 540,1260,1980.
So what is required is a set of audio tone generators at the transmitting end
and a set of selective tone filters at the receiving end."
It doesn't say why there are 3 frequencies each for mark and space. I was
wondering if it was manually selectable for reception conditions or something,
but Will's mention of multiplexing mutiple channels seems more likely.
The actual tube circuitry wouldn't be too complex.