[time-nuts] Power supply noise

[Attila Kinali]

On Thu, 28 Apr 2011 17:04:25 -0600
Joseph Gray <jgray at zianet.com> wrote:

> Power supply noise and ripple has been mentioned before, in relation
> to OCXO's and rubidiums. So, what is considered acceptable in these
> applications?

This highly depends on your system and what you want to achieve.
Just like anything else in engineering ;-)

For OCXOs you have usually a "frequency variation on power supply
voltage change" or something similar. I guess Rb's have something
similar (dont have a data sheet at hand). From this you can guestimate
how much modulation you get on PSU noise.

If you have this, then you can calculate how much noise you get from
the other components in the path of your signal, with regard to the
PSU noise.

After you have that value, you can cross check with the stabilty you
wanted go achieve.

Although this looks quite simple, there is a slight "problem" with this
approach: PSU noise often induces non-linear behavior in circuits.
And often, the behavior varies a lot with the frequency of the noise.
Ie you'd have to model a PSU noise transfer function for each device,
but there no data for this (unless you measure it yourself).

So, usually the approach is to build a system that has very little
PSU noise. Eg use an LDO after a switched power supply to get rid
of the switching noise. If this isnt enough, use additional filters
or noise reduction LDOs (special LDOs made to filter out noise).
If this still isnt enough, add more filters... until you are satisfied.


			Attila Kinali


-- 
The trouble with you, Shev, is you don't say anything until you've saved
up a whole truckload of damned heavy brick arguments and then you dump
them all out and never look at the bleeding body mangled beneath the heap
		-- Tirin, The Dispossessed, U. Le Guin