[time-nuts] Parallel voltage regulators

[SAIDJACK at aol.com]

In a message dated 10/26/2007 14:59:29 Pacific Daylight Time,  
bruce.griffiths at xtra.co.nz writes:

>>  That is a recipe for disaster if one wants a really low noise  oscillator.
>> This is particularly true if one is a  beginner.
>> Reducing the output noise of a switching regulator to  100uV rms or less
>> is neither easy nor simple.
>>   
>


I Agree, it get's even worse when several independent switchers are on the  
PCB, with similar switching frequencies. This can generate beat frequencies  
that vary over time, temperature and from board to board.
 
NEC Tokin makes a ferrous sheet material that is very good at shielding  
switching regulator noise - both electro, and magnetic components.
 
When trying to get noise floors approaching -160dBc/Hz the RS-232  
transmitter on the Fury GPSDO can be used as an example of a potential  interference:
 
The RS-232 transceiver chip generates the RS-232 level signals from a  3.3V 
DC input rail using switched-capacitor voltage doublers. There are no  
inductors or flyback diodes involved. The switching currents involved are very  small 
(orders of magnitude smaller than a switchmode power supply), 0.1uF caps  are 
used as charge reservoirs.
 
Even so, the current spikes generated in the RS-232 transceiver are  enough 
to create a couple of low measurable spurs at around -120 to  -130dBc/Hz on the 
output spectrum of the 10MHz sine wave. This happens even  though we have 
allocated a separate 3.3V regulator for the RS-232 chip, and have  extensive 
power supply filtering in front of the regulator and before the  transceiver chip.
 
There are no switching regulator power supplies at all on the Fury PCB  for 
this reason.
 
To completely remove the spurs we added a mode to disable the transceiver's  
DC-DC switching regulator ("Quiet Mode"). That takes care of the  spurs.
 
bye,
Said



************************************** See what's new at http://www.aol.com