Peter Able wrote:
Incidentally, you attack power supply noise at the power supply. I
wonder if the last components in your proposed PSU are - electrolytic
and / or other capacitors across the outputs.

Nitpick: you attack noise at the source of the noise. Noise generated by
the power supply should be attacked at the power supply, as you say. Noise
generated on the power supply rails by other components should be attacked
close to where those components are.

The reason is that the power supply wiring has a certain impedance at the
frequency of consideration (due to 'parasitic' R/L/C). The further you go
from the source of the noise, the higher the impedance to reach it. And any
mitigation you put in is less able to work effectively because of that
impedance.

This is the reason for the folk wisdom of always fitting a 0.1uF capacitor
next to every IC - it has the minimum trace length and so the minimum
impedance to being able to damp current spikes from that IC switching.

(PDN analysis is a thing you have to do when the design is a bit more
critical, can avoid instability, and can also save you money on unnecessary
components. But folk wisdom gets you a long way if you're not manufacturing
in volume)

Theo