Does anyone know of an analog circuit that will generate white noise from a 6V rail?
This appears to be too low for the conventional back-biased junction approach.
Please correct me if I am wrong.
What are the alternatives?
Leo Kemp
Digital pseudo-random is nice.
Photocoupler shot noise.
A 5 volt zener plus amplifiers should be pretty noisy.
Amplify Johnson noise, plus an opamp's input current noise.
Just use a noisy opamp.
A lot depends on the bandwidth you want, and the spectral purity, and the acceptable statistics. What's it for?
John
Let me throw another one into the hat: A small hot filament, as in light bulb.
Leo, if you already have a good noise source you are familiar with and it requires more than 6V, why not step up the voltage? A little boost regulator maybe? Or a gate driver with Schmitt input running as an oscillator, followed by a Cockroft-Walton style doubler which should net you around 10VDC.
[...]-- Regards, Joerg http://www.analogconsultants.com/
Use an RF transistor such as MPSH-10 the breakdown voltage is 3v (typ) as opposed to the GP transistor 5v (typ)
-- We have failed to address the fundamental truth that endless growth is impossible in a finite world.
How do you process that?
Yeah, a little charge-pump IC could give you 12 volts, to drive a 9 volt zener at a few mA. Two of them, differential, if symmetry matters.
John
Dang, I'll have to try a light bulb some day. You might have to vibration isolate it, which is inconvenient.
To add to JL's list, you can also do the shot noise from a forward biased diode into a TIA. (You need a cap across the bias resistor... Or from the resistor-diode node to ground.) I found this has some excess noise at higher bias currents.. but that may not bother you.
George H.
Forward biased diodes don't actually obey the diode equation, though, because they get into high-level injection so easily. The current has full shot noise, but the junction impedance isn't what you'd calculate, so the noise PSD is wrong. A diode-connected transistor is better, but (as I had cause to remember early last year) it's actually the world's simplest feedback amp, so the bandwidth isn't as wide as one might expect.
Photocurrents are pretty good.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
ht
tet
I've got data somewhere... (I'll see if I can find it.) The noise followed the shot noise 'prediction' up to something like 10uA.. (depending on the feed back resistor) and then things got out of hand. At higher bias currents the low impedance of the diode causes a big gain in the voltage noise from the opamp... but rolled of by the GBW product of the opamp. OK maybe not the best noise source....
What's high level injection?
George H.
George H.
=A0A diode-connected transistor is better, but
.t -
It's when there are enough injected carriers (i.e. forward current) to change the equilibrium carrier density. That leads to shorter carrier lifetime, because the recombination rate is more or less proportional to the product of the electron and hole densities. Ebers-Moll and its brothers assume that the recombination rate is constant.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
This seems to be the tidiest solution, keeping in mind my OP stating "analog".
We would need to order these in. Can anyone confirm this will work from experience?
Leo
I'm going to 'vote' for the digital pseudo-random sequence, but increase its run length by putting its output through polynomial scrambler.
Why use digital pseudo-random? Because, you have TOTAL control over the spectral density function, it's repeatable in Production, and as sure as we all exist, just when you finish you'll be asked to 'tweak' the distribution to some other distribution.
se
on
ight
eat
nd
tnet
as
, at -
Thanks Phil,
Here's some PN data. (A 1N4148 I think.) Sorry for the 'weird' scale on the x-axis.
George H.
p)
Hi Leo, you should tell us what frequency range you are interested in.
(I have no knowledge of the RF transistor.)
George H.
p)
It's going to work just as well a 3V zener - and, since both breakdown by the Zener mechanism - it won't produce as much noise as higher voltage Zeners (which break down by avalanching) do at low currents.
Why buy a more expensive RF transistor rather than a cheap low voltage zener? The parallel capacitance will certainly be lower, but this doesn't seem to be an advantage worth paying money for, unless you want very high frequency noise.
-- Bill Sloman, Nijmegen
That's equivalent to increasing the register length, at least for non-crypto use. XORing a number of different-length registers has about the same effect.
National used to sell a PR-based noise generator chip, but it's gone now. Probably someone else has one. For audio bandwidths, some little ARM or PIC could generate nice noise with a repetition period past the age of the universe. With a trick or two, it could be genuinely random, based on physics and not just math. If it doesn't exist already, somebody here could do it and sell it. May as well do bandwidth control while you're in there.
John
Nice. You were running the diode straight into the summing junction, I gather, which is why the background started rising once the noise gain got too big.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
)
oise
tion
light
the
se at
and
ost
ld net
,has
te,
s a
r oxt -
edid you try using the soundcard for 'auto' sweep and data collection?
you can actually plot the density function with variable BW's, too.
Digital pseudo-random noise generators can be deceptively crappy. Certainly if you were going to do one with a LFSR you'd want it to be long and clocked very fast compared to your bandwidth.
That having been said -- just about any random noise generator can be deceptively crappy. There are a lot of pitfalls in trying to realize statistical ideals in the real world.
-- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
If you mean an LFSR, running it through a polynomial scrambler won't increase its run length -- it'll just change the apparent code.
But it's really easy to make a loooooong LFSR, so why not go that route?
Good points.
-- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Shot noise is really really Gaussian. I've measured it to be within 0.1 dB out to 7.1 sigma, which corresponds to a threshold crossing rate of
10**-11 times the measurement bandwidth. (And that was on the output of a laser noise canceller, with the beam scanning around inside a vacuum chamber.)Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.