wondering if the performance of these for audio have been significantly
exceeded by another ic lately. i'm too lazy to build walt jungs design.
cost don't matter much. application is mic pres.

Dufus

dufus wrote:
wondering if the performance of these for audio have been significantly
exceeded by another ic lately. i'm too lazy to build walt jungs design.
cost don't matter much. application is mic pres.

Some of the Linear Technoology designs have much lower output noise and
slightly better regulation.

Personally, I think a 317 with a hell of a lot of bypass capacitance is
just as good, but my friend Sheldon disagrees completely and is a huge fan
of the LT parts.

Unless you care about dropout voltage, in which case you want a totally
different family of low-dropout regulators. In this battery-powered age,
there is a huge push for LDO regulators and most of the latest high tech
designed have been designed for LDO at the expense of other parameters,
so watch out.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

"Scott Dorsey" wrote in message
...
dufus wrote:
wondering if the performance of these for audio have been significantly
exceeded by another ic lately. i'm too lazy to build walt jungs design.
cost don't matter much. application is mic pres.

Some of the Linear Technoology designs have much lower output noise and
slightly better regulation.

Personally, I think a 317 with a hell of a lot of bypass capacitance is
just as good, but my friend Sheldon disagrees completely and is a huge fan
of the LT parts.

The LT1085 positive regulator tests significantly better than the LM317 in
line rejection, noise and output impedance. The negative equivalent, the
LT1033, offers no significant improvement over the LM337, other than higher
current capacity, probably not an issue for you. (I'm quoting Walt Jung's
tests, in the article series describing his super-regulator design.)

Peace,
Paul

"Scott Dorsey" wrote in message
...
dufus wrote:
wondering if the performance of these for audio have been significantly
exceeded by another ic lately. i'm too lazy to build walt jungs design.
cost don't matter much. application is mic pres.

Some of the Linear Technoology designs have much lower output noise and
slightly better regulation.

Personally, I think a 317 with a hell of a lot of bypass capacitance is
just as good, but my friend Sheldon disagrees completely and is a huge fan
of the LT parts.

The LT1085 positive regulator tests significantly better than the LM317 in
line rejection, noise and output impedance. The negative equivalent, the
LT1033, offers no significant improvement over the LM337, other than higher
current capacity, probably not an issue for you. (I'm quoting Walt Jung's
tests, in the article series describing his super-regulator design.)

Peace,
Paul

dufus wrote:
wondering if the performance of these for audio have been significantly
exceeded by another ic lately. i'm too lazy to build walt jungs design.
cost don't matter much. application is mic pres.

Some of the Linear Technoology designs have much lower output noise and
slightly better regulation.

Personally, I think a 317 with a hell of a lot of bypass capacitance is
just as good, but my friend Sheldon disagrees completely and is a huge fan
of the LT parts.

Unless you care about dropout voltage, in which case you want a totally
different family of low-dropout regulators. In this battery-powered age,
there is a huge push for LDO regulators and most of the latest high tech
designed have been designed for LDO at the expense of other parameters,
so watch out.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

dufus wrote:

wondering if the performance of these for audio have been significantly
exceeded by another ic lately. i'm too lazy to build walt jungs design.
cost don't matter much. application is mic pres.

Personally I've never found the couple of hundred microvolts output noise (
audio band ) to be a problem. If you're getting more than that, your problem
is something else.

If to make it really quiet, considered an emitter follower output buffer
after the 317/337 with an R to the base and C to ground ? Really cheap
although you'll lose about a volt.


Graham

"Pooh Bear"

** Hello Pooh !

Personally I've never found the couple of hundred microvolts output noise
(
audio band ) to be a problem. If you're getting more than that, your
problem
is something else.


** Supply rail noise from IC regulators easily finds its way into a mic
pre-amp or other audio circuits - whether they have inherently good SVRRs
or not.

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are enough
to guarantee a noisy ground. Only a microvolt or so of such noise bleeding
into the input or feedback grounds is enough to destroy the noise figure of
a mic-pre.


If to make it really quiet, considered an emitter follower output buffer
after the 317/337 with an R to the base and C to ground ? Really cheap
although you'll lose about a volt.


** A much better solution is a simple RC filter - a 10 to 22 ohm
resistor in series with each rail followed by a 1000 uf cap grounded to the
REGULATOR ground reference point will produce a virtually silent supply
rail.


BTW The published data for the LM317/337 regs suggests they are noisier
than the LM78xx / 79xx series while experience shows very much the opposite.



.............. Phil

"Phil Allison" wrote in message ...
"Pooh Bear"

** Hello Pooh !

Personally I've never found the couple of hundred microvolts output noise
(
audio band ) to be a problem. If you're getting more than that, your
problem
is something else.


** Supply rail noise from IC regulators easily finds its way into a mic
pre-amp or other audio circuits - whether they have inherently good SVRRs
or not.

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are enough
to guarantee a noisy ground. Only a microvolt or so of such noise bleeding
into the input or feedback grounds is enough to destroy the noise figure of
a mic-pre.


If to make it really quiet, considered an emitter follower output buffer
after the 317/337 with an R to the base and C to ground ? Really cheap
although you'll lose about a volt.


** A much better solution is a simple RC filter - a 10 to 22 ohm
resistor in series with each rail followed by a 1000 uf cap grounded to the
REGULATOR ground reference point will produce a virtually silent supply
rail.


BTW The published data for the LM317/337 regs suggests they are noisier
than the LM78xx / 79xx series while experience shows very much the opposite.



............. Phil

Keep in mind that 22 ohm series resistance will raise the regulator's
impedance to the level of the esr of the following bypass cap. Very
low impedance caps like the Rubycon Z series will yield lower post
resistor impedance.
G.L's suggestion of a pass transistor offers a lower impedance than
the less elegant series resistance solution. If you raise the
regulator's output a volt and a half you will compensate for the
voltage drop across the transistor.

Jim Williams
Audio Upgrades

"Jim Williams"

Keep in mind that 22 ohm series resistance will raise the regulator's
impedance to the level of the esr of the following bypass cap.

** You forgot about the reactance.

Very low impedance caps like the Rubycon Z series will yield lower post
resistor impedance.

** Even the lowest ESR caps still have reactance - the best 1000uF cap
in the world has 8 ohms reactance at 20 Hz.

G.L's suggestion of a pass transistor offers a lower impedance than
the less elegant series resistance solution.


** At the very low end of the audio band - maybe so - across the rest of
the audio band the 1000uF electro wins hands down over an emitter follower.


If you raise the regulator's output a volt and a half you will compensate
for the
voltage drop across the transistor.

** But watch out for even the briefest short on that DC supply - bye bye
pass transistor.




............... Phil

Phil Allison wrote:

"Jim Williams" wrote

snip

Very low impedance caps like the Rubycon Z series will yield lower post
resistor impedance.

** Even the lowest ESR caps still have reactance - the best 1000uF cap
in the world has 8 ohms reactance at 20 Hz.

Perfectly trrue Phil. A good reason NOT to use RC decoupling as a solution to
poor regulator performance.

G.L's suggestion of a pass transistor offers a lower impedance than
the less elegant series resistance solution.

** At the very low end of the audio band - maybe so - across the rest of
the audio band the 1000uF electro wins hands down over an emitter follower.

You care to elaborate ? Actually - supply impedance isn't the real issue - noise
was what was being discussed !

If you raise the regulator's output a volt and a half you will compensate
for the
voltage drop across the transistor.

** But watch out for even the briefest short on that DC supply - bye bye
pass transistor.

Nope - the LM 317/337 - 7815/7915 will protect it. Zero problem Phil. Anyway -
where's the short going to come from ? Bench techs ?


Graham

p.s. - I just got a pre-prod sample of a PSU using exactly this technology back
from my friends at Kwan Asia.. Lovely performance.

"Jim Williams"

Keep in mind that 22 ohm series resistance will raise the regulator's
impedance to the level of the esr of the following bypass cap.

** You forgot about the reactance.

Very low impedance caps like the Rubycon Z series will yield lower post
resistor impedance.

** Even the lowest ESR caps still have reactance - the best 1000uF cap
in the world has 8 ohms reactance at 20 Hz.

G.L's suggestion of a pass transistor offers a lower impedance than
the less elegant series resistance solution.


** At the very low end of the audio band - maybe so - across the rest of
the audio band the 1000uF electro wins hands down over an emitter follower.


If you raise the regulator's output a volt and a half you will compensate
for the
voltage drop across the transistor.

** But watch out for even the briefest short on that DC supply - bye bye
pass transistor.




............... Phil

"Phil Allison" wrote in message ...
"Pooh Bear"

** Hello Pooh !

Personally I've never found the couple of hundred microvolts output noise
(
audio band ) to be a problem. If you're getting more than that, your
problem
is something else.


** Supply rail noise from IC regulators easily finds its way into a mic
pre-amp or other audio circuits - whether they have inherently good SVRRs
or not.

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are enough
to guarantee a noisy ground. Only a microvolt or so of such noise bleeding
into the input or feedback grounds is enough to destroy the noise figure of
a mic-pre.


If to make it really quiet, considered an emitter follower output buffer
after the 317/337 with an R to the base and C to ground ? Really cheap
although you'll lose about a volt.


** A much better solution is a simple RC filter - a 10 to 22 ohm
resistor in series with each rail followed by a 1000 uf cap grounded to the
REGULATOR ground reference point will produce a virtually silent supply
rail.


BTW The published data for the LM317/337 regs suggests they are noisier
than the LM78xx / 79xx series while experience shows very much the opposite.



............. Phil

Keep in mind that 22 ohm series resistance will raise the regulator's
impedance to the level of the esr of the following bypass cap. Very
low impedance caps like the Rubycon Z series will yield lower post
resistor impedance.
G.L's suggestion of a pass transistor offers a lower impedance than
the less elegant series resistance solution. If you raise the
regulator's output a volt and a half you will compensate for the
voltage drop across the transistor.

Jim Williams
Audio Upgrades

Phil Allison wrote:

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are enough
to guarantee a noisy ground. Only a microvolt or so of such noise bleeding
into the input or feedback grounds is enough to destroy the noise figure of
a mic-pre.

So what do you suggest as a solution to this problem of noisy grounds?
I'm sure you're not suggesting we omit the bypass caps. Should they
each have their own ground trace all the way back to the power supply?

Is that practical? Or do you have a better idea?

ulysses

"Justin Ulysses Morse"
Phil Allison

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are
enough
to guarantee a noisy ground. Only a microvolt or so of such noise
bleeding
into the input or feedback grounds is enough to destroy the noise figure
of
a mic-pre.

So what do you suggest as a solution to this problem of noisy grounds?
I'm sure you're not suggesting we omit the bypass caps. Should they
each have their own ground trace all the way back to the power supply?

Is that practical? Or do you have a better idea?



** How about you read the thread.





.......... Phil

"Justin Ulysses Morse"
Phil Allison

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are
enough
to guarantee a noisy ground. Only a microvolt or so of such noise
bleeding
into the input or feedback grounds is enough to destroy the noise figure
of
a mic-pre.

So what do you suggest as a solution to this problem of noisy grounds?
I'm sure you're not suggesting we omit the bypass caps. Should they
each have their own ground trace all the way back to the power supply?

Is that practical? Or do you have a better idea?


Phil Allison wrote:

** How about you read the thread.

I read the thread. It didn't answer or even discuss the question I
asked. The thread discussed using a pass transistor vs. a series
resistor for local decoupling. I'm asking about the ground paths.
Same issue, different question.

As an aside, Phil, you'd get a lot more respect and credibility around
here if you didn't go out of your way to be insulting to people like
Paul and Graham who are both tremendously knowledgeable and extremely
graceful and generous with their time.

ulysses

On Wed, 09 Jun 2004 22:32:34 -0500, Justin Ulysses Morse
wrote:

So what do you suggest as a solution to this problem of noisy grounds?

Nakamichi had an interesting solution back when they made fancy
electronics. Their floating supply technique was used in the CR7
era tape decks and some contemporary home preamps and amps.

Pass transistors naturally float, but voltage references must
reference ground. Nakamichi took advantage of bipolar supplies
by supplying current for the reference zener from the opposite
polarity supply and buffering the zener with a high Gm JFET.

Only the JFET's gate was connected to ground. IIRC they used
Toshiba 2SK147 and 2SJ72 Blue or violets. The JFETs' source
connected to one end of the reference zener and its drain to
the other polarity supply through a resistor.

IOW, the JFETs provided a buffered ground.

Chris Hornbeck

"Justin Ulysses Morse" wrote in message
m...
Phil Allison wrote:

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are
enough
to guarantee a noisy ground. Only a microvolt or so of such noise
bleeding
into the input or feedback grounds is enough to destroy the noise figure
of
a mic-pre.

So what do you suggest as a solution to this problem of noisy grounds?
I'm sure you're not suggesting we omit the bypass caps. Should they
each have their own ground trace all the way back to the power supply?

I've done that a few times, and gotten nice quiet circuits as a result. Pain
in the ass for circuit card design, but it works.

Peace,
Paul

"Phil Allison"

BTW The published data for the LM317/337 regs suggests they are noisier
than the LM78xx / 79xx series while experience shows very much the
opposite.



** Mystery solved - the spec sheets are misleading.

The data sheets published by Natsemi for the LM317, LM337 and LM338
adjustable regulators all give the noise output as typically 0.003% of the
output voltage, measured from 10 Hz to 10 kHz. Data for similar regulator
ICs from LT gives the *identical* numbers to Natsemi.

However, what they *both * fail to tell you is that noise figure applies
when the adjust pin is *unbypassed*. Once the usual bypass cap is fitted (
10uF will do) the output noise no longer depends on the output voltage and
is fixed at 0.003% of the lowest possible voltage of 1.25 volts. This
equates to about 37 uV and not 450 uV as predicted by the data.

An LM338 I tested ( my 1.25 - 12 V, 6 A bench PSU ) showed only 26uV output
of white noise measured in a 20 kHz bandwidth.




............ Phil

"Pooh Bear"

** Hello Pooh !

Personally I've never found the couple of hundred microvolts output noise
(
audio band ) to be a problem. If you're getting more than that, your
problem
is something else.


** Supply rail noise from IC regulators easily finds its way into a mic
pre-amp or other audio circuits - whether they have inherently good SVRRs
or not.

The very low output impedance of IC regulators and the existence of
bypass capacitors of various values strewn around the PCB pattern are enough
to guarantee a noisy ground. Only a microvolt or so of such noise bleeding
into the input or feedback grounds is enough to destroy the noise figure of
a mic-pre.


If to make it really quiet, considered an emitter follower output buffer
after the 317/337 with an R to the base and C to ground ? Really cheap
although you'll lose about a volt.


** A much better solution is a simple RC filter - a 10 to 22 ohm
resistor in series with each rail followed by a 1000 uf cap grounded to the
REGULATOR ground reference point will produce a virtually silent supply
rail.


BTW The published data for the LM317/337 regs suggests they are noisier
than the LM78xx / 79xx series while experience shows very much the opposite.



.............. Phil

dufus wrote:

wondering if the performance of these for audio have been significantly
exceeded by another ic lately. i'm too lazy to build walt jungs design.
cost don't matter much. application is mic pres.

Personally I've never found the couple of hundred microvolts output noise (
audio band ) to be a problem. If you're getting more than that, your problem
is something else.

If to make it really quiet, considered an emitter follower output buffer
after the 317/337 with an R to the base and C to ground ? Really cheap
although you'll lose about a volt.


Graham