[time-nuts] Low noise power supplies for time nuts circuitry

lists at lazygranch.com lists at lazygranch.com
Tue Nov 22 14:59:49 EST 2011


If you want low noise voltage regulation, use a shunt regulator. If the load is low current, feed it with a floating current source such as a depletion mode jfet (or several in parallel). These are also sold as current regulating diodes. 

I've used this scheme in IC designs. You also find it in high end audio design. 

The shunt design is push pull. If the shunt is designed well, the ultimate high frequency feed through from the power supply, assuming ideal caps, is simply a capacitor divider based on the bypass cap and the capacitance across the current source. 

The bozos at Broadcom have actually patented using shunt regulation in their chips. Good luck enforcing that patent. 

-----Original Message-----
From: SAIDJACK at aol.com
Sender: time-nuts-bounces at febo.com
Date: Tue, 22 Nov 2011 14:18:20 
To: <time-nuts at febo.com>
Reply-To: Discussion of precise time and frequency measurement
	<time-nuts at febo.com>
Subject: Re: [time-nuts] Low noise power supplies for time nuts circuitry

As Bob Pease used to say: Spice plots are good for padding bird cages, not  
much else :)
 
Hittite has a fantastic new ultra low noise LDO for VCO's. Haven't had a  
chance to check that out, but it looks very promising. Has anyone tried that  
part here?
 
Also, remember that those caps are microphonic most of the time, and could  
actually worsen supply noise in the presence of vibration.
 
I get noise floors of below -170dBc with just simple RC or LC filtering on  
the power supplies, say 10 Ohms into 100uF Tantalum in parallel with some 
10nF  to 100nF ceramics (or better Polyester caps), all following the typical 
Linear  Technologies low noise LDO's. That will cut off at 160Hz already, 
and go down at  40dB per decade if two filters are cascaded. Use two 100uF 
Tantalums or  22Ohms resistors for a <80Hz cut-off.
 
At close-in frequencies, the crystal will likely be the worst noise source  
and overpower the supply noise by far. It's hard to get better than -108dBc 
at  1Hz, and -138dBc at 10Hz at 10MHz anyhow.
 
For high frequency switcher noise, use shielded (TDK etc) 33uH inductors in 
 series to a 10 Ohm resistor, into 100uF Tantalums with 10nF and 220pF  in 
paralell. Cuts off <160Hz and has very good isolation at high  frequencies 
without radiating.
 
For the best performance against supply noise, simply use differential  
techniques.
 
bye,
Said
 
 
In a message dated 11/22/2011 10:52:29 Pacific Standard Time,  
phk at phk.freebsd.dk writes:

In  message <6F9E458F-B701-49C1-8D83-EBDA35784764 at ulrich-bangert.de>, 
Ulrich  Ban
gert writes:

>It seems to turn out as if the well known  Wenzel suggestions for voltage
>regulator finesse were not state of the  art [...]


I've played a bit with the Wenzel circuits and they can  provide truly
outstanding damping, my best was 80dB using a HP VHF  transister I can't
remember the number of.

And yes, they are very  sensitive to just about everything, in particular
temperature.

But  getting at good solid 30+dB damping is not _that_ hard, in particular
for  very low-current constant loads, such as X-tal oscillators.

--  
Poul-Henning Kamp       | UNIX since Zilog Zeus  3.20
phk at FreeBSD.ORG         | TCP/IP since RFC  956
FreeBSD committer       | BSD since 4.3-tahoe   
Never attribute to malice what can adequately be explained by  incompetence.

_______________________________________________
time-nuts  mailing list -- time-nuts at febo.com
To unsubscribe, go to  
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the  instructions there.

_______________________________________________
time-nuts mailing list -- time-nuts at febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


More information about the time-nuts mailing list