Paul Burridge wrote in message . ..

What's a sensible target for the multiplication factor using diodes
from say a 4Mhz source? I'd be nice to go straight to say 20X on
account of the large physical size of the filters at lower HF., not to
mention the savings on intermediate stages.

Well, somewhat counter to what Harold wrote, the old HP Journal
article I've dredged out to copy says a step recovery diode can do 20X
at 30% efficiency. But that's guys who really understood what they
were doing! (Just coincidentally, the one table they have is for
20X!)

He's right...the filters don't have to be huge. But I'd think of them
as large compared with a SOT-23. The tradeoff is that if you do it
all in one step, you have to have a filter sharp enough to take the
19x and 21x down to your desired level, whatever that is. But with a
5x and a 4x, you can use a couple filters with wider percentage
bandwidths, since for the 5x you'll be using a square-wave input which
has practically no 4th and 6th anyway, and the 3rd and 7th are a big
percentage removed from the 5th...and the 4x can similarly avoid the
3rd and 5th if you do it right.

But again, a PLL can be very small indeed. And I never did see an
answer to the question about does it really need to be locked to your
low freq, or can it be a separate xtal at 70MHz or whatever?

Cheers,
Tom