[time-nuts] a better LO for an HP4352B.

[ct1dmk]

Hello,

After playing a bit with phase-noise measurement on a 4532B I realized
that the contribution of the external LO it requires to function 
dominates the setup noise floor (...no surprise here).

There are quite a few generators out there (reasonable prices, surplus 
etc.) up to 1 or 2GHz to choose from and I have already tested a few 
different models I have here.

Seems the 4352B noise floor on itself is not too bad but it ends up
showing the PN of the external LO, that in reality does not need to be a 
full blown HP or R&S etc. bench-top instrument.

Knowing that the 4352B it only requires the LO in 100kHz steps, I'm 
tempted to build something myself (say up to 500MHz or 1GHz) aiming for 
the very best possible crazy lowest PN a one can get.

... and now I can't get out of my head the search for the architecture 
for such synthesizer (kind of 250-500MHz for a start, divided down for 
lower freqs. I don't fear building myself the best VCO's possible and I 
don't mind to have several to cover an octave.)

But the very best synthesizer architecture to lock them aiming solely 
for low PN is what I would like to hear comments about.

My initial idea is direct sampling the VCO with a 100KHz reference.
(a micro-controller would get the VCO close enough to the lock point and 
then gets out of the way and let the loop grab it). Can't think of 
anything else simpler or with less noise contributions... but...

(needless to mention using the best OCXO reference can have)

I would be happy to hear all sort of comments.
Including suggestions of existing commercial models that can't be beaten 
by any of my attempts.


Luis Cupido
ct1dmk.