Regulator Noise

[jean-charles maillet]

>A short while back, there was a discussion on the noise output of three in
>regulators.  I did some looking in a databook and found some interesting
>numbers:
>
>Fixed Regulators:
>
>    Regulator      Voltage          Current    Noise Level

>    LM78xx         +5, +8, +12, +15   1.5A       75uV
>    LM79xx         -5, -8, -12, -15   1.5A       125uV
>
>Variable Regulators:
>
>    Regulator      Voltage          Current    Noise Level
>    LM317          +1.2 - +37         1.5A       .003%

>If my math is correct, .001% on a 15V output from a variable regulator
>translates to 150uV, while .003% on a 15V output translates to 450uV!

>I had been using LM317 & LM337 units since they allow 15dB tighter
>regulation on the supply line!  After seeing this, I'm wondering if having
>3x the noise level should make me reconsider!

just some thoughts ...

The 317 seems to be widely used for it's very low Zout ... I would think
it's a good idea to be thinking about good regulation, but how much matters
?! ... if your synth is running on op-amps and sophisticated IC's that have
internally referenced supply circuits (with regulation or high PSRR) then
your machine can probably live on a wall-wart which has very little
regulation ... on the other hand if the design has some single-ended stuff
in it then yeah you need those suckers ... if you've got a spikey circuit
like Paul's LFO then you should power the synth from a regulator which has
strong drive ... but, again, this depends on how the circuits are grounded
... if the power and grounding lines start from a strong regulator and then
proceeds to deliver current to all the circuits (in series) and then
finally we arrive at the noisy LFO circuit and THEN the audio output
circuit we'd have a serious ground loop involving the last two circuit
blocks... in which case any change in the regulator type would be of little
help there ...

The noise levels you mention are very informative ... I would think that
overall circuit noise levels would be in the few millivolt range at least
and so the worse case regulator noise levels would be fairly less than this
and would then become secondary ... besides if we're talking about supply
rails which have noise levels of mVolts (and should be periodically shunted
by caps to short that stuff out) then the question becomes: what is the
PSRR of the modules to mV supply noise ?

Even in single ended designs I think these figures are way low enough not
to bother with too much ... I hear from this old tube audio consultant that
the 8-pin 317's (I think they're 317T's) are considered overall quietest
(combined dynamic regulation and self-noise) for single-ended circuit rail
supplies and are good enough for powering low-voltage Nuvistor Phono
preamps.  I suspect any of them wouldn't make much difference on a properly
wired synth circuit.

jc

/////////////////////////////   vivaAnalog    \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\

/////////////////////   http://www.lynx.bc.ca/~jc/    \\\\\\\\\\\\\\\\\\\\\\

     Vacuum tube modeling - Great guitar players - Traynor amp schematics

//////////////////////////////////    \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\