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

Attila Kinali attila at kinali.ch
Mon Feb 27 10:23:22 UTC 2012


On Mon, 27 Feb 2012 01:29:11 -0800
gary <lists at lazygranch.com> wrote:

> On 2/27/2012 12:48 AM, Attila Kinali wrote:
> >
> > If you are really time-nutty, you can let the DC/DC converters produce
> > a voltage about 1V above what you need and use low noise LDOs (ie not
> > the 78xx or LM317&  Co) to produce the voltages for the thunderbolt.
> > This should give you a 60-80dB damping of the noice produced by the
> > DC/DC converters.
> 
> Having designed LDO chips, people expect them to perform miracles well 
> beyond reality. If you have a PNP pass and you are sitting near dropout, 
> you get control loops that are an ugly combination of a path to keep the 
> PNP from getting saturated plus one to control the voltage. With P-fet 
> pass devices, the control is better (no sat killer needed with a fet), 
> but still you are trying to regulate with what amount to be a variable 
> resistor.

Hmm.. i had the impression if you stayed beyond two times the drop out
voltage that the regulation loop could take most of everything?

And afaik all modern LDOs are FET types? Aren't they?

 
> You really don't get that much filtering at switcher frequencies with 
> LDOs, plus some regulators can't handle too low of an ESR.

Modern LDOs all handle zero ESR. And they have to.
But yes, switcher frequency filtering is something you should do with
lots of Cs and maybe some Ls. That's another reason why i said to use
DC/DC converters with MHz switching frequency. It makes it much easier
to filter them out.

> If you care 
> about noise, screw efficiency and go with a shunt regulator. There are 
> hack circuits on the net to take 431s plus external components to roll a 
> decent shunt. Note the shunt regulator makes the DC/DC happy by 
> presenting a uniform load.


Yes, that would be also a possibility. But it only makes sense if
you have a more or less constant power consumption. Otherwise you'll
just waste the peak current (plus some headroom) permanently.

Eg, with a OCXO you'd have to design the regulator for the heating
current. And when the OCXO goes into steady state, you waste ~80%
of the energy to heat the enclosure.

> If you do a shunt right, the bypass capacitor will do all the work. This 
> is somewhat true with a P-fet pass regulator, where Cds is forming a cap 
> divider with your bypass.

I don't exactly see how the Cds helps with regulation as a capacitive
divider. Could you elaborate this a little bit.

> I never really warmed up to PNP pass devices, 
> but they are best for high voltage applications.

While the notion of "high voltage" is constantly moving up :-)

			Attila Kinali

-- 
Why does it take years to find the answers to
the questions one should have asked long ago?



More information about the time-nuts mailing list