BTW, I just built the suggested circuit, on one of those old white
plug-boards which is NOT a great idea at RF. It worked fine. Easy to
tune just with a scope. Performance was as I expected: full -0.6V to
+5.6V or so swing at 18MHz at the input of the "18MHz" gate, and an
18MHz rectangular wave at its output. 74HC04 hex inverter; 3 stages
as a simple RC oscillator to generate the 3.7MHz input. Duty cycle of
that 3.7MHz was fairly close to 50%, but definitely not right on. The
performance tells me that the 'HC04 really does have a fairly high
input impedance at 18MHz, and requires under a milliwatt to
drive--perhaps well under. There's AMPLE fifth harmonic power in that
3.7MHz square wave out of an 'HC04, and you just need to filter and do
whatever impedance transformation is appropriate to get the desired
result.

Incomplete filtering of the non-fifth harmonics causes variation in
the output pulse width (in a cycle that repeats every cycle of the
3.7MHz input), and frankly if I wanted a pretty pure 15th harmonic at
the final output, I'd use a bit better filter to get a cleaner fifth.
But try the simple one first, to convince yourself that it's possible.
It's a pretty "hack" circuit, in that I just made educated guesses at
the 'HC04 input and output impedances at 18MHz, and at a reasonable
load resistance for an 'HC04. It could probably be optimized a bit,
but it DOES work as described.

Inductor Qu: The loaded Q of the circuit I described depends on the
output impedance of an 'HC04 inverter, and the input impedance of the
same with the bias resistors included. It should be running somewhere
around 20-25, I believe. That determines how well the third and
seventh are rejected. The unloaded Q of the inductor determines how
much fifth-harmonic power is lost in the filter. You can stand some
loss, but you don't want to waste too much. If the loaded Q is 25 and
the inductor has an unloaded Q of 50, then you will waste half the
power in the inductor. That probably would still work, but it would
be better if the Q was more like 100, as a minimum. Based on what
I've been hearing so far, I'm not very confident that you will be able
to tell the Qu of your junque-box inductors, especially since you need
to know it at the operating frequency. Qu of a lot of manufactured
inductors is quite low, especially at non-optimal frequencies. If you
picked a ~20uH inductor designed for use in a switching power supply
at 100kHz, it could well be awful at 18MHz. It only takes a couple
minutes to wind up an air-core one, so that's what I'd suggest. The
one I used was 49 turns on a T-94-2 powdered iron core.

Cheers,
Tom


Paul Burridge wrote in message . ..
On 18 Mar 2004 10:16:53 -0800, (Tom Bruhns) wrote:

Paul Burridge wrote in message . ..

Exactly how much power do you need?

Only enough to feed another inverter gate.

Egad, Paul! You've been wasting this much net bandwidth just to drive
another HC gate?? All you need is a filter/matching circuit that
steps up the voltage. This is DOG SIMPLE! See below.

Thanks, Tom. I'll give your suggestion a try. Just one question: is
the unloaded Q of 200 you specify critical? I probably have a 20uH
factory-made inductor laying about in my parts stash but won't know
it's Q without measuring it. Need I bother?