"Tom" wrote in message
...
Paul Burridge wrote:
Feed the square output to a coil+capacitor in parallel and you'll get
a sine wave. Tune the this 'tank' circuit for harmonics of the
funamental and you can double, triple, quadruple and so on the
original square wave's frequency.

Thank you Paul! This is a good hint. Can you tell me in which book did you
learn it? I would really like to get into this stuff.

IMO you are being led astray. I don't see why the emphasis on frequency
multiplication. It is fairly simple to make an LC oscillator cover a 2 to 1
frequency range, or even 3 to 1. That could mean 4 to 12 MHz in one tuning
range and 1.3 to 4 in another. The problem is lower frequencies. These days
it is hard to find a large-enough variable capacitor for operation down in
the few-hundreds of kilohertz range, let alone lower.

I know of only two ways that will cover the whole zero to 12 MHz range
without bandswitching or switched filters.

One is the direct digital synthesizer DDS. Single chip DDS units are
available [Analog Devices Inc] fairly cheaply, but they are tiny and require
a computer or equivalent to control them. You get "perfect" frequency
accuracy and stability and a sine-wave output.

The other is a beat-frequency method with two oscillators operating
considerably above 12 MHz. One crystal controlled, the other knob
controlled tuning from the crystal frequency to 12 MHz higher. Feed them
into a double-balanced mixer (MiniCircuits) followed by a low-pass filter
that passes 0 to 12 MHz, and strongly rejects the crystal osc frequency and
all higher. One knob, no switching, no computer needed for control - but
also much poorer frequency accuracy and stability.