SV: Re: [sdiy] Linear PSU options .. 7815/7915 vs LM317/LM337vs 723 etc

[harrybissell]

If you have traces of audiophile in your jeans... I'd get a 
new wardrobe...

Some 'audiophiles' may choose a circuit topology that is very
sensitive to power supply noise... for some other esoteric reasons
(unknown and unknowable to mere mortals :^)

Engineering is a total package...  sometimes sloppy engineers
(now raise your hands if you're one - my hand is up) will specify
a component without looking at the total 'error budget'.  I once
spec'd an LM4040BIZ for a reference. When the manufacturer could not
get them... I studied carefully and found that even the BARGAIN
grade 4040DIZ would be way less error than anything else in the
total system.

These self-same 'audiophiles' might eschew DECOUPLING caps because of
their
inferior sonic performance... or the 'coloration' of the sound. Some of
them
might even LIKE the sound of power supply related coupling. (like a
no-feedback
design that really DOES have feedback, through the supply :^)

Hey... Audiophiles like TUBES.  I have heard many people rave about tube
amp designs that have enough residual white noise to make Rykhaard
D.A.M.I.A.N
envious. I agree they sound good... but I just ignore the hiss, its easy
to
do and probably most listeners have trained themselves to do so since
childhood !

That said, I use the 723 designs all the time... in module form from
PowerOne
(and such ilk).  If I was doing my own supply... the 78xx and 79xx would
be
the first weapon of choice for a fixed regulator, follwed by higher
grade
fixed regulators (error budget again) followed by the LM317 337... IF I
needed
and adjustable regulator.

GACK. NOW I REMEMBER YOU :^P
You're using a 4069 as a distortion element and worried about the 78xx
noise ???
Dude, kick back and take a (insert relaxant of your choice here).  Try
the
generic power supply design... then some decent layout techniques...

... if that does not work, then add the premium power supply... and the
Mu Metal
shield to keep the magnetic flux from your own, and everyone else's
transformers
out of the mix.  That will be a bigger hit than the 78xx.

If you did the board and decoupling right, the power supply noise is
most often not an issue. I agree with Paul Perry's post on the subject
of local references
when needed.

H^) harry

Seb Francis wrote:
> 
> This is a very good point, but nevertheless many hi-end audio circuits
> do seem to use better than 7xxx PSUs.  Maybe this is kind of over the
> top audiophile engineering, but I'm not so sure.
> 
> In my application there isn't anything like a VCO that needs a really
> precise voltage, but there are however hi-end audio components (THAT
> VCAs in the compressor) that need to pass/process the audio with as
> little added noise as possible.  Also, the distortion processor elements
> have very high gain which will probably amplify any power supply noise.
> 
> So while I don't want to get over-complicated, if a 317/337 PSU can give
> lower noise than a 7815/7915 PSU with only an extra couple of resistors
> then it seems worth it (maybe I have some traces of audiophile in my
> genes! ;)
> 
> Seb
> 
> P.S. The only digital bit (a PIC micro for the MIDI interface) has its
> own 78L05 driven power rail.
> 
> Paul Perry wrote:
> 
> >There's another side to this regulator debate, and that is:
> >just how good does the regulated supply need to be?
> >When one looks at precision analog circuitry, you often see
> >power rails no better than you would get from a LM78xx,
> >but wherever a *really* precise voltage source is required, there is a
> >precision source dedicated to just that. And it is right there on the
> >board where it is wanted.
> >Which at a stroke solves the problem of noise generated on the
> >power rails by switching transients etc.
> >And, speaking of transients...... anyone serious (including some
> >members here, I notice) have separate power rails for the digital stuff,
> >and for the analog that they try to keep noise free.
> >I personally think, the difference between the 78xx and other regs,
> >is swamped by these other techniques..... and if you want to go the
> >whole distance, you are going to need to have precision voltage
> >diodes & separate analog/digital lines anyway!
> >
> >paul perry Melbourne Australia
> >
> >
> >
> >