[time-nuts] re low noise regulators

[Jeffrey Pawlan]

I checked both datasheets, the 723 and the LT1762.  It was specsmanship that
rated the 723 as having lower noise. Here's how:

The internal comparison opamp in the 723 was OK for its day back in Fairchild's
history but by today's standards, it is noisy and has poor gain and BW. So in
order to make the regulator low noise, a 5uF capacitor had to be placed across
the ref comparison input.

The low noise, and even the standard linear regulators today are much quieter
and few people need 5uV of power supply noise, especially given that the
circuitry we place on the power rails will almost certainly spoil that.
So no one specs putting a 5uF cap on the ref line. The LTC LT1762 does suggest a
Cbyp of .01uF will provide a noise of 20uV broad bandwidth as you correctly
stated. I am certain that putting a higher value cap and perhaps even a RC
filter on the reference input will lower this substantially. But again, I
suggest that it will become meaningless to all but Super-Audio Snake Oil
people because any circuitry powered from the produced Vcc line will add more
noise than the regulator. Even with the now ancient history LM317 style
adjustables, I always put a tantalum cap across the reference to ground. It
does wonders for getting rid of traces of hum.

Just for your collection I will send John two App notes in PDF form to post.
They cover the subject of reducing spikes and switching regulator noise.

regards and 73,


Jeffrey Pawlan  WA6KBL
Pawlan Communications