Optical noise

Noise sources at/near DC aren't statistically random, they're dominated by (1) filament temperature variations (due to AC drive, or to gas convection eddies inside the bulb) (2) op amp offset/drift and 'popcorn' noise sources (3) there might be some leakage through that bypass capacitor, as well. (4) Unlikely sources (is your photodiode microphonic? Are the connecting wires subject to triboelectricity? Does the input of the software radio have ripple on its reference voltage? ...)

yes Phil, I meant KLA-Tencor. We used silicon diodes to detect backscatter and secondary electrons in e-beam microscopes.

Thanks, Jure Z.

Good questions. I need to get more systematic about my noise characterization. The first incarnation I used a two transistor amp of my own (lousy) design. Lots of RFI. I could pickup 810Khz AM station real well. Second go round I used a opamp and was much more careful on the shielding. I use a hp function generator to put a small (10 mV out of 1V DC) sine wave modulation on an LED. Low and behold while I was checking out my gain curve that pesky AM station showed up at about the 3uV level. This time it turned out to be leaking into the LED drive voltage and coming in on the light. Most of my experimenting has been done with unfiltered sunlight. I get about 20 mV noise in the audio fluctuations from the (I think) the atmosphere. Above 10 kHz the radio noise is about at the 0.7 to 1 uV level. Turning the diode amp on (no light) about doubles this. Direct sunlight (300 uA ish) doubles to triples this again above 10Khz.

I need to add a DC out to the box so I can make more meaningful measurements of the AC noise relative to the DC level. I don't claim to be an expert on anything, just having fun.

Thanks Paul C.

Dang. Used a scope to look at the noise on an incandescent versus a red LED. The incandescent is a tiny high pressure light that was used in a machine vision application. LED on/off showed almost no noise increase ~20 uV. The incandescent upped the noise to good 500+ uV. Noisy little bugger.

Paul C.

The classical fluctuations of thermal light are small compared to the shot noise unless you're far below the peak wavelength of the Planck curve. Hanbury Brown was working with the bluest stars he could find, and PMTs with the longest available red tails, and he never got classical fluctuations above ~1% of the shot noise, iirc. (I was the one that posted the H-B pdf, with some formatting help from others on the NG.)

Tungsten bulbs are pretty quiet except that they have some response to thermal forcing up into the kilohertz.

Cheers

Phil Hobbs

Trying and "failing" is a really good way (for me) to learn some things. I can read till I'm blue in the face but until I get some hands on feel for a subject I don't absorb much.

Thanks!! This is a really good read. Still struggling to understand some of the arguments. A little blue in the face.

What's Thermal Forcing? BTW, I powered both lights from the same DC lab supply. HB says the LED should have more wave noise because it has less bandwidth. I understand this is still way way below my detection level (and very likely always will be).

Oooh, I've got an anecdote to relate:

I built this LED string,

The LEDs are four pin "high efficiency red" types (i.e., actually a deep orange), 70mA rated. As you can imagine, they make excellent point sources. They are good enough that, beyond a few feet distance from an emitter, I clearly see a halo of interference fringes and speckle in my eyes. At a distance of 8' or so, the speckles are several line widths apart ("lines" meaning, the finest alternating black and white lines which can be discerned at a given distance). I doubt I could photograph the effect, but some diffraction phenomena in a dark room might be possible.

Point being, I think the light is at least coherent or isotropic enough to enable short range diffraction phenomena. It's my understanding that speckle is effectively the space-domain equivalent of time-domain "wave noise" (interference is interference, right..?). It seems to suggest these LEDs are at least bright enough to illustrate these lower-order phenomena, if not actually exhibit wave noise by bandwidth or what have you.

Tim

No kidding. They were one of our customers back in the 80's. They used to buy our Multibus I tape controller boards.

I've used the shot noise from a lightbulb or LED (onto pin photodiode) to measure the charge of the electron... to a few percent, depending on how hard you work on measuring the bandwidth. The incandescent bulbs are more subject to vibrations, (mostly low frequencies) and you need to either filter that out or reduce the vibration coupling.

George H.

_LEDs.html

The last time I went looking for this effect(HBT) it was with a diode laser run below threshold. The diode laser has a tiny area, and when I ran the numbers it looked like it might work. But below threshold there was still all this excess noise... kinda 1/2 lasing of the laser. Which might be interesting but not what I wanted to see. Here's a graph of current noise density (i^2/Hz) vs DC current. At 10uA it's a laser and at 100nA it was just an led.

So I was (recently) thinking that if I could 'crap up' the output facet of the diode laser. I might be able to stop it from lasing. If I could keep it as an LED upto a few micor amps of DC photocurrent I think I could see the effect.

So how does one 'crap up' the output facet? Could I scratch it up with some fine sand paper? Crocus cloth? other ideas?

George H.

_LEDs.html

The last time I went looking for this effect (HBT) it was with a diode laser run below threshold, shinning on a small photodiode. The DL has small area, and running the numbers it looked like it might work. Unfortunately when it gets near threshold the DL gets all this excess noise, its kinda 1/2 lasing. Here?s a graph of noise density (i^2/ Hz) vs DC photo current.

The line is drawn by hand but it?s the expected shot noise level.

At 10uA it?s lasing.

So I was thinking about this again. And I wondered if I could (somehow) ?crap-up? the output facet of the DL and stop it from lasing. I could then get LED type light out of it at a higher drive level. So any ideas how to ?crap-up? a diode laser? Could I scratch up the output face with some fine sand paper? Crocus cloth?

George H.

ec_LEDs.html

an

ich

to

Oops... well I posted twice. The picture in the second post is better.

George H.

AFAIK, most small incandescents are evacuated.. I would guess that a pressurized halogen bulb would be a lot noiser because of turbulence in the gas flows around the filament.

Best regards, Spehro Pefhany

string,

You can just gouge the back facet with a diamond scriber. That's the most common way of making superluminescent diodes (SLDs) in the lab, AFAICT. You'll probably have some residual ripples in the spectrum due to feedback, and of course you lose the nice hermetic seal.

Cheers

Phil Hobbs

lec_LEDs.html

ep

an

hs

hich

he

e.

h to

y (i^2/

l.

it from

t -

Ahh, Right, there's some reflective coating on the back... unfortunatley there's not a lot of space between the back facet and the monitor photodiode. (I can barely see the DL to begin with.) Maybe it would be easier to buy a super luminescent diode.

George H.

ated

ell.

Well I've never looked at a halogen bulb.... (I'm not sure I even have one.)

The easy thing about (vacuum) incandescent's is they come in small packages and I can drive them with a low voltage DC source.

George H.

"The Journey is the reward"

eff.com- Hide quoted text -

string,

Or get a bigger one. You'll need to dump a lot of current into it anyway.

Cheers

Phil Hobbs

dominated

well.

Either that was before my time or i missed it in some other way. Could you either repost or PM me?

?-)

Cheers

Phil Hobbs