beta at low currents

I worked on one military system where the spares buy was one plus the square root of the number of units in the system.

John

That spares buy rule apparently didn't work well enough. I get requests all the time for MC1530 and MC1648, chips I designed in the mid '60's.

...Jim Thompson

ot

Is this just the effect of the larger band gap? I can fit a lot of kT's into 1 Volt. Can you cool down an Si photodiodes and reduce the dark current?

Ahh I think I remember some pictures made by your instrument. You had low resolution images of footprints several minutes after they were laid down. Could you turn this into an imager that could see where the heat is leaking out of buildings/ houses? Or 'cool' leaks into houses?

Sorry to hear about your employment situation. I infered that you were 'out on your own' from previous posts. If you ever have a killer idea for a physics advanced lab experiment and want to make minimal money selling, it drop us a line. ( snipped-for-privacy@teachspin.com) We toyed with the idea of trying to make the Hanburry-Brown and Twiss intensity correlator, but placed it on the way back burner. 50 years later this still looks like a 'hard' experiment.

George Herold

Yes, and yes. The Vf of your average LED is about a volt higher than that of a Si diode, which as you point out is a lot of kTs. (The zero-bias resistance of a diode is ideally (kT/e)/Is, and the Is of a LED is theoretically something like exp(-40) times that of a Si diode of the same size.)

It isn't quite that sensitive--at a guess, you'd need ~1-10 microkelvin sensitivity to see actual footprints. 'Footprints' just seemed like a good name for the system. Looking at heat leaks would certainly be possible, and would be a good application. Probably a handheld device would need a small camera so you could overlay video on top of the thermal image--otherwise it would be harder to see what you were looking at, and anyway the video effectively enhances the resolution of the thermal image.

Thanks. It was a bit of a wrench at first, but it's actually much more fun than what I was doing at IBM, and I'm actually in positive cash flow territory just now, doing an expert witness gig (helping defend the innocent) and some other consulting and product design. Never would have had the guts to bail out myself--I have 5 years' college tuition and 7 years' mortgage payments to go--but it seems to be working out pretty well so far. (Otherwise I wouldn't be planning to spend all that dough on fancy cabinets, as discussed on another thread.) Besides, now I can deduct all these toys on my income tax.

Robert Hanbury Brown is one of my technical heroes. I suspect that you could do significantly better than he could, using a hot tungsten lamp source and big ol' solar cells with, say, 900 nm longpass filters in front of them. Doing the correlation in analog would still be the right answer, I think, and nowadays would be much easier than in his version, which used discrete transistors! Using a common-base input stage, you can get 20 kHz or so of bandwidth from a solar cell, which is probably enough if you have lots of photocurrent. Kilohertzy speed simplifies the correlator design a lot, too--something like a linearized Gilbert cell multiplier in a chopping loop, where one of the inputs and the output get commutated at, say, 10 Hz, might be a good place to start thinking.

I have a scanned PDF of RHB's book "The Intensity Interferometer", which is long out of print. It's a big file, so if anyone wants it, I'll stick it on my web page for a week or two. It'll scroll half of ABSE otherwise.

Cheers

Phil Hobbs

Hi, George,

Thanks. Googling "Landauer shot noise" turns up a lot of good stuff.

I just got an email from one of the resistor manufacturers. They say that high-value thickfilms have shot noise due to non-ideal materials properties, roughly an order of magnitude more shot noise than thinfilm resistors. I have no idea how that compares to full shot noise, so I suppose I'll have to measure some. THAT is known to be a nuisance.

I did once try an assortment of megohm-range metal film, carbon film, and carbon comp resistors and saw no shot noise in a setup that should have easily resolved full-shot noise.

John

ew

de

Glad to hear you've landed on your feet ...running.

I've got copies of chapters 2,4 and 6 of RHB's book so I don't need the PDF. (I might down load it anyway.) Thanks for the correllator layout ideas. I've printed 'em out and stuck them in my HB&T folder.

George

mp)

ry

g you

. If

bly

se. I

e but

ving a

ublic

ll

or, like

get

tween

hem,

al

w,

ell

The reference I have is "Solid-state shot noise" Phys. Rev. B. Vol. 47 num 24, page 16 427 (that seems like a big page number.) June 15,

I wonder if they really mean shot noise or is it 1/f noise? You'd have to look at the spectrum to decide. Hmm is this really a 1 Tohm resistor? Given a fraction of a pF of stray C the noise will roll off at 1 Hz or so... I don't know of a way to measure the spectrum at such low frequencies. Perhaps the question of 1/f vs shot noise is moot.

George Herold

The photodiode/photocell sources sound good, remember to bias the output diode. An optoisolator even puts the whole assembly together for you, mass-production and all.

Alternately, you can feed a small voltage (30 mV) into a 1Mohm/1 uF integrator-connected op amp to make a ramp output at 30 mV/second. Put a 33 pF capacitor from the op-amp output, it sources I =3D C dV/dt =3D 1 pA into a load with impedance significantly lower than 33 pF...

The ramp will saturate, eventually, but at 30 mV/second, you get a good ten minutes of operation between resets (at +/- 10V output). Temperature/age stability might be better than LED lamps.

Phil Hobbs wrote: [snip]

Phil, I'd be interested in a pdf of the book. your site still is electrooptical.net ?

Rene

I don't think that using superbeta transistors is going to help. I would try high voltage small signal types or EHF small signal types instead.

teeny-weenie

university.

shot

not

make

drew

cents--far

magnitude

magnitude

I have a back burner wish list item of a cheap fiber optic communications kit suitable for high school kids or younger. I was figuring bare plastic fiber of about 300 feet (100 meters) to be far enough to get the idea across well. .

Plastic won't make it nearly that far; attenuation is way too high.

There's surely some cheap glass fiber on ebay, maybe some of the older big-core stuff.

John

which=20

Me too. Next i need the time to read it. .

teeny-weenie

university.

current

full shot

not

make

the

by

drew

cents--far

positive

magnitude

weird.

by

drew

positive

magnitude

weird.

Perhaps, but the safety considerations for kids handling glass are scary. Barely to mention some ruggedness issues. =20

nie

ty.

shot

ot

ke

ew

ar

de

ew

de

Many years back, I bought such a kit. It had a modestly powerful Laser, a long thick plastic fiber and a photo-detector. The PCB for driving the LED could take audio in and AM modulate the Laser. The PCB for the photo-detector had a high gain audio amplifier. It had quite a bit of distortion and not a great bandwidth so it wasn't hifi but it sounded ok. IIRC the fiber was about 100 feet.

A fiber-optic cable is about as dangerous as a hairbrush.

And what's this nonsense about never allowing kids to be exposed to the tiniest hint of danger? I'm sure glad I let The Brat take some risks, and occasionally get hurt. She's no wuss.

John

Completed cables are safe to handle, but bare glass fiber is a considerably different case. Part of the exercise is seeing how some of it works. Thus plastic. Perhaps for the longish run (100 yd)i will have to use a completed cable. @