I was looking for a low noise regulator to power a log amp/detector earlier
this year and was rather surprised to find the 78xx regulators were
considerably better than many of the "low noise" devices.

But!!

I've also had odd experiences with some brands of 78xx devices (and way
before the 'net was anything more than SLIP dial up to a shell so I doubt
they were Chinese fakes) ,  one was bad enough that it gave some very
random voltage measurements on a digital meter,  turned out of the was
creating all sorts of RF hash in the low VHF range up to and possible
beyond the FM broadcast band.

On 7 Dec 2016 20:10, "Joe Leikhim" jleikhim@leikhim.com wrote:

I don't think so.  I first ran into a batch of LM340-5's
that were excellent oscillators back in the 1970's... long
before counterfeiting was even remotely possible.

The symptom is the regulator puts out only 4.5 out of 5V.

LM309's were, however, totally immune.

Usually, I had to be really bad to make it happen, things
like using clip leads between the power supply and load
with the LM340-5 dangling in between.

The answer is as simple as a couple of 0.1uf ceramic caps
soldered right at the input and ground, and the output
and ground pins.

LDO (low dropout) regulators are very susceptible to
oscillation.  They need to have a couple of hundred uf
of good quality capacitance right on the input and output
leads.  Where people usually get in trouble, is in not
knowing that electrolytic capacitors lose most of their
capacitance as the temperature starts hovering around 0C.

The circuit works great on the bench, but fails when out
hanging on a light pole...

-Chuck Harris

Joe Leikhim wrote:

Hi

There are a bunch of people making stuff like 78xx regulators. Even
20 years ago, there was a lot of difference between brand M, brand T,
and brand F on these devices. Today the spread is even larger. Toss in
outfits that sub contract the work to who knows where this week ….
What you get today may not be what you got yesterday or what you
get tomorrow.  At least 20 years ago you got the same thing when you
re-ordered.

Crazy stuff

Bob

On 12/7/2016 12:16 PM, Clint Jay wrote:

Are you kidding me?  Check out the Linear Technology LT3042 and
LT3045 with 2 (yes, TWO) NANOVOLTS/root-Hz spot noise.  Orders of
magnitude better than the 78XX introduced 45 years ago.

Rick N6RK

I've no doubt there are many excellent low noise regulators out there that
are orders of magnitude better than the 78xx series,  but there are also
many that claim low noise as a headline feature and are actually worse
when you dig into the specification.

On 7 Dec 2016 23:42, "Richard (Rick) Karlquist" richard@karlquist.com
wrote:

On 12/7/16 3:32 PM, Richard (Rick) Karlquist wrote:

I'd second the recommendation for the LT3042 - I'm using them in a
science instrument which is very noise sensitive.

Hi

You can indeed find regulators with noise densities that work out to be uV/sqrt(Hz) advertised as
“low noise”. Who knows how marketing justified making the claim other than “not as totally rotten
as our older part”.

Bob

Am 08.12.2016 um 00:50 schrieb jimlux:

I can only point again to my own measurements under comparable conditions:

<
https://www.flickr.com/photos/137684711@N07/24070698809/in/album-72157662535945536/

The LT3042 outclasses the old crowd, whether they are called low noise
or not.

Don't overdo it with the output capacitor, the 4.7uF from the data sheet
is ideal.

Too much, and the response gets a peak. The pic was taken from a dead
bug implementation

involving a Micro-SO with thermal pad on the belly, too horrible to show.

There is also a pic with an external D44VH10G power transistor to supply
more current.

And remember, 0dB == 1nV/rtHz is the INPUT voltage noise of an LT1028 or
AD797.

(give or take 10%...)

regards, Gerhard

p.s.

Ulrich, I seem to remember that you were in Ulm in a previous life.

Then you might recognize the Blau valley on the leftmost image. :-)

Yes, I hit one of those. I forget the frequency other than around 1 MHz, but it was pretty warm when it shouldn't have been, and it was several volts amplitude.

This thing I'm chasing is much more subtle.

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of Chuck Harris
Sent: Wednesday, December 7, 2016 1:57 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Totally unrelated, but..

I don't think so.  I first ran into a batch of LM340-5's that were excellent oscillators back in the 1970's... long before counterfeiting was even remotely possible.

The symptom is the regulator puts out only 4.5 out of 5V.

LM309's were, however, totally immune.

Usually, I had to be really bad to make it happen, things like using clip leads between the power supply and load with the LM340-5 dangling in between.

The answer is as simple as a couple of 0.1uf ceramic caps soldered right at the input and ground, and the output and ground pins.

LDO (low dropout) regulators are very susceptible to oscillation.  They need to have a couple of hundred uf of good quality capacitance right on the input and output leads.  Where people usually get in trouble, is in not knowing that electrolytic capacitors lose most of their capacitance as the temperature starts hovering around 0C.

The circuit works great on the bench, but fails when out hanging on a light pole...

-Chuck Harris

Joe Leikhim wrote:

time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

One has to be very careful with LDOs.  As we say in applications,
RTFDS.  :-)  The output capacitor often needs a finite ESR (effective
series resistance).  A too-large ceramic cap WILL make them oscillate.
I have noticed that they are used in some GPSDOs.

David

On 12/8/16 11:39 AM, Van Horn, David wrote:

In message <CY1PR04MB2347AE1EF637565B5B4D0C2FE0840@CY1PR04MB2347.namprd04.prod.
outlook.com>, "Van Horn, David" writes:

How long are the wires supplying the regulator ?

The LC of the wires inductance and the shunt capacitor over the
regulators input has a resonance frequency...

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

HI

Ok, so to fully evaluate the problem, you will need to de-encapsulate the parts. That is the next step
in this adventure. First pass would be to pull the plastic off of a couple of bad ones and get some pictures
of the die. After that, trace them back and see if they are Brand A or Brand B. Next, do the same for a good
one (or two) and see if the die trace to the same manufacturer.

Next after the die are traced down would be probing with something like a Cascade probe system. You can
get them that are alleged to be good into the > 20 GHz range. Team that up with some low noise test gear
and you will be able to rule in / rule out the regulator it’s self. Yes this is a bit expensive. If you want to troubleshoot
the issue, it is the next step.

It’s a whole lot easier to swap out the regulator for a different one and just move on.

Bob

I am pretty sure that I ran across this once as well with the
7800/LM340 series in the early 1980s with parts from a major
manufacturer like Motorola, National, or Texas Instruments; the
regulators ran hot and the output voltages were slightly low no matter
what decoupling arrangements were made at the input and output.  The
stability was poor as if the parts were oscillating but no oscillation
was apparent.

There is some discussion here about different designs for these
regulators although the photographs are broken:

http://www.righto.com/2014/09/reverse-engineering-counterfeit-7805.html

On Wed, 7 Dec 2016 15:56:55 -0500, you wrote: