[time-nuts] Generating a stable 26MHz and 19.2MHz from 10MHz

[ed breya]

If you don't need the 26 and 19.2 to be exactly phase locked to the 
10, and you can find crystals at those frequencies, I would suggest 
that you go straight digital. There are a number of simple divide, 
mix, multiply, and filter combinations that would make those 
frequencies directly. The main thing is to not have to multiply up 
too high to get the desired mixer products, versus how well your 
filters have to work.

For example, if you divide the 10 MHz by 5 to make 2, and XOR mix 
them, you'll have 8 and 12 available, both standard crystal 
frequencies. The 8 can be filtered out with a crystal filter (but the 
phase info may be lost) and then XOR doubled to 16, another standard 
crystal frequency, filtered out and mixed with the 10 to get 26. The 
16 can be divided by 5 to make 3.2, also standard, filtered out and 
mixed with the 16 to make 19.2. It may even be possible to skip the 
intermediate filtering steps if you have  very good output filters - 
it depends on how clean they need to be further out from the carrier. 
You would of course have to make sure that the components have 
sufficient phase noise performance, and you would have to design the 
crystal filters - possibly multi-staged, and protected from thermal 
and vibration effects enough for the stability time range needed.

Ed