White Noise Source

[Baron]

Jasen Betts Inscribed thus:

if linux, this is easy:

cat /dev/urandom > /dev/audio

(dev/audio is (IIRC) mono 8 bit Ulaw and runs at a fixed sample rate
of about 8000 samples per second so it's not propper white noise.

for something better try sox or audacity.

Well, I'll be... ! One to put in the vade mecum.

 

[Chiron]

On Fri, 10 Feb 2012 19:22:39 +0000, Jasen Betts wrote:

if linux, this is easy:

cat /dev/urandom > /dev/audio

(dev/audio is (IIRC) mono 8 bit Ulaw and runs at a fixed sample rate of
about 8000 samples per second so it's not propper white noise.

for something better try sox or audacity.

That's very close to what I did - cat /dev/urandom > randomfile. Then I
imported that as a raw file into audacity, converted it to ogg format.
When I looked at its spectrum, it was suspiciously flat all the way from
zero to 20kHz, where it fell off at > 30 dB per octave. Just a little
*too* perfect. I go by the maxim that, if everything goes according to
plan, you've obviously overlooked something.

Anyway, it's a nice, noisy signal, but I'm still going to do the other
one just for fun (amplified diode noise). As someone pointed out, having
a quick noise source around could be handy for other reasons besides
talking to ghosts.

 

[Jamie]

With the a 20V zener you don't need any amplifictaion, just a
buffer.
I wrapped this together on the top of a power supply.


+15
|
1 Meg
|
+--0.1uF--+---->scope (use x10 probe,
| | not a crappy clip lead.)
Z 100k
^ |
| GND
-15V

I saw a random RC step response with a time constant of ~2us.
With something like a few hundred mV step size (on average)
(Driving home I realized that the ~80pf in the clip lead I used to
connect to the scope was my RC.. so I can't pull any 'real' numbers
out.)
You can try it yourself with a x10 probe and report results.
You should do better than two micro seconds.

George H.

You know that is strange, I have some 39V zeners and when applying 50
50 volts with a test like above, I don't see what I would say is noise
being generated from the node of the zener and R ?

I tried this with my Tek 350mhz scope and the 100 DSO one.

Maybe I have different zeners ? Of course, these are 2 watt zeners
I tried, they were left overs from a job .

Jamie

 

[Bob Masta]

On Fri, 10 Feb 2012 19:22:39 +0000, Jasen Betts wrote:




That's very close to what I did - cat /dev/urandom > randomfile. Then I
imported that as a raw file into audacity, converted it to ogg format.
When I looked at its spectrum, it was suspiciously flat all the way from
zero to 20kHz, where it fell off at > 30 dB per octave. Just a little
*too* perfect. I go by the maxim that, if everything goes according to
plan, you've obviously overlooked something.

Actually, the spectrum of random-amplitude samples should be
flat. So I'd be more suspicious of the roll-off above 20
kHz. I haven't delved into ogg format, but my guess is
that it may be rolling things off above the audio range.

Note that when an FFT generates a spectrum, it doesn't
really know anything about frequency... it just operates on
a set of numbers. If the numbers are random, the resulting
spectrum is flat. The frequency interpretation comes by
knowing the sample rate of the raw data values.

If audacity can't look at the spectrum of the raw data, can
you convert to WAV instead of ogg? WAV is just a simple
header in front of the raw data... it doesn't actually
convert anything. The header holds the sample rate info and
related stuff.

Best regards,


Bob Masta

DAQARTA v6.02
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Scope, Spectrum, Spectrogram, Sound Level Meter
Frequency Counter, FREE Signal Generator
Pitch Track, Pitch-to-MIDI
Science with your sound card!

 

[Ben / SM0KBW]

George Herold skrev 2012-02-11 00:16:

With the a 20V zener you don't need any amplifictaion, just a
buffer.
I wrapped this together on the top of a power supply.


+15
|
1 Meg
|
+--0.1uF--+---->scope (use x10 probe,
| | not a crappy clip lead.)
Z 100k
^ |
| GND
-15V

I saw a random RC step response with a time constant of ~2us.
With something like a few hundred mV step size (on average)
(Driving home I realized that the ~80pf in the clip lead I used to
connect to the scope was my RC.. so I can't pull any 'real' numbers
out.)
You can try it yourself with a x10 probe and report results.
You should do better than two micro seconds.

George H.

One trick, at least at RF is to use the Zener breakthrough in a regular
transistor, the base - emitter diode backwards can give a very good
noise source ranging upp to 1GHz and even above.

/Ben - SM0KBW

 

[Baron]

Ben / SM0KBW Inscribed thus:

George Herold skrev 2012-02-11 00:16:

One trick, at least at RF is to use the Zener breakthrough in a
regular transistor, the base - emitter diode backwards can give a very
good noise source ranging upp to 1GHz and even above.

/Ben - SM0KBW

Has anybody tried LEDs as noise sources ?

 

[[email protected]]

Ben / SM0KBW Inscribed thus:


Has anybody tried LEDs as noise sources ?

Wouldn't there be correlation problems with any light sources around? Seems
like it's needlessly asking for trouble.

 

[Michael Black]

Wouldn't there be correlation problems with any light sources around? Seems
like it's needlessly asking for trouble.

Yes, and someone's already warned about that, but the issue isn't because
of it's an LED but because of the seethrough casing.

I remember one article about a VCO where the author suggested trying
different types of diodes for the varactor. And he mentioned trying some
small signal diodes and having AC hum on the output, until he realized the
hum was coming from the desk lamp that was shining on the clear-cased
diode.

I've seen other mentions of problems because of light. I recall one
article warning about it, but not because of hum, but because the external
light was affecting diodes, and they were biased wrong or something, or
maybe turning on when they shouldn't.

Since LEDs have been used as low voltage zeners, perhaps they do have some
ability to better generate noise at a lower voltage or something. With a
breadboard, it costs very little to try. If it works, simply paint the
LED, get rid of any external problem.

Michael

 

[Baron]

Phil Hobbs Inscribed thus:

LEDs make great current switches--the garden variety ones I used in
Footprints had leakage is down in the femtoamps from ~-5V to +0.5V.
The photosensitivity can be a feature as well as a bug, when you use
it to supply DC bias for a capacitive sensor such as a pyroelectric.
(See http://electrooptical.net/#footprints, especially the paper with
the gory technical details.)

Cheers

Phil Hobbs

Seems most of the photos are missing !

 

[Ben / SM0KBW]

George Herold skrev 2012-02-19 17:04:

Thanks Ben, I've heard that, but never tried it.

George H.

I've only used that kind of noise source when i'm aligning VHF and UHF
preamps. The zener breakthrough is at around 6V for ordinary
transistors as the 2N2222. You have to test a handful to find the best
noise source

/Ben - SM0KBW

 

[Baron]

Phil Hobbs Inscribed thus:

Baron wrote:

Right! Thanks. Fixed now.

Cheers

Phil Hobbs

Looks fine now. Thanks.

 

[JW]

What are you going to paint it with?

I discovered over the weekend, that black electrical tape is not as
good as I thought it was. (I'll have to do some testing, when I have
time.)

Try using some heat-shrink tubing, then pinch the ends closed while still
hot.

 

[Chiron]

Opps, "Mystery Managed". Operater error... I've got a two sets of
diode on the pcb. I had the wrong set covered with tape. The Zeners
were still 'wide open'.

Moving the tape to cover the 'noise diodes' fixes the light sensitivity
problem...
Duh!

George H.

I hate when I do that... and it's not a rare occurrence.

 

[Chiron]

Opps, "Mystery Managed". Operater error... I've got a two sets of
diode on the pcb. I had the wrong set covered with tape. The Zeners
were still 'wide open'.

Moving the tape to cover the 'noise diodes' fixes the light sensitivity
problem...
Duh!

George H.

Darn... I was thinking that it might be fun to investigate stuff in the
IR range using your technique of masking the diodes with visually opaque
tape that passed IR. A whole new area to check out...

Oh, well. Better luck next time.

BTW, after all this, I opted to just use software-generated noise for my
brief experiment. I'll put together a hardware version that I'll likely
use for testing circuits (amps and such).

Thanks again for all the great ideas. You guys sure taught me a lot
about all this stuff - which I guess is the idea.

 

[Chiron]

Well it might not be a complete waste. I 'someday' want to make a
variable noise source that shows how shot noise depends on the size of
the electron. Current noise density i sub n = sqrt(2*e*I), (amps/sqrt
Hz). Where I is the average current. So if you could turn a knob and
make electrons four times bigger, (while keeping the average current the
same.) Then the noise density would go up by a factor of 2. In my model
this is exactly what's going in in the zener. With no light I get big
pulses but not very often. With the light on you get less noise.
Because an absorbed photon makes an e-h pair and this can cause the
zener to discharge sooner than it would have without the photon. The
average current is kept roughly constant by the zener bias resistor.

Sounds interesting... never would have thought of it.

Personally, I don't feel most of my mistakes (in electronics) are a
waste. If nothing else, I wind up learning something.

For example: I learned to make sure there's nothing between me and the
door when I power up a circuit I've built. Also, try not to be the high-
value resistor that discharges high voltage capacitors. There are
endless opportunities for learning, if you don't incinerate yourself.