[time-nuts] Power Supply Noise Affects Thunderbolt 1 PPS

[David]

On Tue, 28 Feb 2012 00:13:18 -0500, "Charles P. Steinmetz"
<charles_steinmetz at lavabit.com> wrote:

>Ed wrote:
>
>>So, for best HF noise performance where the input noise may be 
>>large, it's best to use a follower or shunt regulator topology, 
>>despite the lower efficiency - unless efficiency is more important.
>
>To put it more bluntly, the last time I looked (it has been a while, 
>so there may be something new I haven't seen) LDOs were all 
>substantially inferior to standard regulators in terms of noise 
>feedthrough and HF load regulation.  The rule of thumb has always 
>been to avoid LDOs unless low dropout was the paramount concern.  Has 
>something changed with current LDO designs to make them competitive 
>with standard regulators in terms of noise feedthrough and HF load regulation?

As Ed points out, the topology makes a significant difference.

The common emitter/source low dropout designs also have problems with
frequency compensation because of the voltage gain in the power stage.
Negative regulators with NPN pass elements have the same problem
leading to the output capacitor requirement.  Usually they have worse
transient response.

The highest performance low dropout regulators I have seen use
emitter/source followers with a low current bias supply.