[time-nuts] Low noise voltage regulators

[Bruce Griffiths]

Garry Thorp wrote:
> With the 723, you can make the reference noise as low as you want, by
> heavy RC filtering. This applies whether you use its own reference or a
> better external reference.
> The 723 also seems to work quite happily with a feedback capacitor from
> the output to the inverting input, reducing the AC gain to unity. The
> output noise will then just be the buffer amplifier's input noise
> voltage, ~5-6nV/rtHz. This is about 20dB better than most modern
> 'low-noise' LDOs will do. (I can't remember what the amplifier's noise
> was like at very low frequencies  -  it was many years ago at a
> different company, and I don't have the results any more.)
>
> Garry
>    
Using a capacitor from the output to the inverting input of the error 
amplifier to reduce the high frequency noise gain of the regulator is 
often used in high performance series regulators.

Another option/enhancement, when using a 723 with external pass 
transistors is to preregulate the 723 input.
Even a simple resistor + zener + emitter follower can be effective in 
improving the line rejection.
The extra supply voltage required is easily generated using a simple 
diode voltage doubler driven from the transformer secondary.

Another possibility is to use a tracking preregulator such as an LM317 
or similar regulator.
In this case the tracking pre-regulator maintains a constant voltage 
drop between the LM723 supply and output pins.

With some elaboration (perhaps using a higher current regulator) this 
can be extended to the case where an external pass transistor is used.
However this increases the minimum regulator input to output voltage.

Bruce