[time-nuts] Defective Z3801 and strange PS board

[ed breya]

I agree that the extra converter is to provide isolation and input 
voltage flexibility. Mine runs on 24V, and I assume that the first 
converter is the only difference from the more common 48V units. 
Also, each stage of conversion should provide additional power supply 
rejection performance. It is apparently very complicated for what it 
does, and makes plenty of heat and switching noise. I have often 
wondered just how much switching and digital noise is actually 
present in that box environment and in the output signal (common-mode 
too). It would mostly be 20 kHz and up, so far out from the carrier, 
and maybe doesn't matter in the intended applications. Has anyone 
ever looked at or quantified how much extra stuff is there?

This jogs my memory of some theories I have about a  design weakness 
in the Z3801A, and how to greatly improve the close-in short-term 
(1-100 sec) performance. I'll post it after some time to recall and ponder.