Winfield Hill wrote:

Phil Hobbs wrote...

Tom Bruhns wrote:

So the right way to do this is to lower the _effective_ minimum
capacitance. You can do that by adding an inductor, to cancel out
capacitance. You can end up making the tuning range as wide as you
want, but at the expense of the crystal (ceramic resonator in your
case) being less of the overall frequency determination. In other
words, there comes a point where you'd be as well off to just do an
LC oscillator. But to double, say, the range, it's a good way to go.

I guess I re-discovered what was already well known, but a few years
ago I designed such a VCXO, and was amazed how linear the
freq-vs-controlvoltage curve was (a good thing for use in a PLL).
Don't know what range you're trying to achieve, but I had no trouble
getting a bit more than 0.1% (~20kHz at 14MHz) that way, with a
crystal.


An inductor in series with the varactors, then another one in
parallel with the series combo can get you a very wide range of
impedance from
a decent varactor.



Sounds good. How about a specific example?

Thanks,
- Win

whill_at_picovolt-dot-com

Last time I used this was with an MV104 common-cathode dual hyperabrupt,
to make a 110-MHz phase shifter. It used a Mini-Circuits quadrature
hybrid in the usual way, coming in the 0 degree port, coming out the 180
degree port, and hanging matched reactances on the 90 degree ports.

Each section had its own inductors, and the cathodes were bypassed
heavily (1000 pF) to ground so that the two sides didn't interact too
much. The component values were 45 nH in series and 43 nH in parallel.
It was linear to within +-4 degrees, and the one section gave phase
shifts from 12 to 164 degrees, both dramatically better than I could get
with a bare varactor.

The idea is to have the varactor resonate with the series inductor just
off the low-voltage end of the range, and have the series combination
resonate with the parallel L just off the high-voltage end of the range.
Since the series-resonance doesn't even notice the parallel L, the
design equations decouple nicely, too. You adjust the placement of the
resonances to get the range and linearity desired.

Cheers,

Phil Hobbs