[time-nuts] Thunderbolt Supply

[Charles P. Steinmetz]

Bob wrote:

>A three terminal regulator, properly chosen, will do a great job down at
>audio.

Depends on what you mean by "great job."  All of the 3T regulators 
I'm familiar with have at least two orders of magnitude more noise 
than a well-designed LN analog regulator -- many have much more 
(especially LDO parts).  At the low impedance levels involved, it's 
challenging to filter it out.

>Finding inductors that will work from a few tens of KHz to a MHz while
>moving over an amp - much easier than at 30 Hz.   *     *     *
>
>Next stop above that - something designed for RF.   *     *     *
>
>The net result is a gizmo that's quiet, small, and low dropout.

In both of these ranges, the high current switching transients of the 
switcher tend to get into everything, including "grounds."  There are 
resistive drops ("ground loops") plus inductive and capacitive 
coupling (every wire/trace is an antenna).  It takes great care with 
layout and design to reduce these even to a dull roar, and 
extraordinary measures to clean up a switcher for precision analog 
work between DC and several MHz.

I'm not saying it can't be done, but it is a lot of work and you have 
to know exactly what you are doing in an environment where it is very 
hard to measure what you are trying to eliminate.  It almost always 
involves shielding each module separately, using feedthrough 
capacitors from module to module, etc.  Most of this can be avoided 
by using only analog regulators.

Best regards,

Charles