High Quality White Noise Gen

rote:

he signal

other from

tage

the

ds a

spaced

pacing

ine

quist

The

t

ayout

I

ou

the

a

dth

)

ss

ot

nd a

.

t -

Oh...(Silly me) Well so how 'bout a photo-transistor with three terminals (one where the base has a lead connected to it.) I'll short the base and collector together and then reverse bias the EB junction. With the bias voltage set just below where the thing wants to zener. Then a photon could set the whole thing off?

I thought I had a photo transistor around here someplace, but I couldn't find it. (The original prototype of optical puming used a phototransistor, until I convinced every one that a photodiode was, more sensitive (larger area), less noisy and faster.)

Ahh the other silly thought that this generated was using an avalanche zener diode as a photo detector. I know the Zener's I use are sensitive to light, so again if I was to park them just below the knee voltage and then hit them with light could I get some gain out of them? Kinda a poor man's APD.

George H.

signal

from

Shot noise is really, really Gaussian--I've measured it out to 7.1 sigma, where (false alarm rate)/(bandwidth) = 10**-11 (about 1 count per day in a 1 MHz bandwidth). A decent comparator driving a one-shot can make nice Poissonian pulses (as someone already suggested).

You can make okay phototransistors by connecting a real photodiode between the base and collector of a real transistor. Still slow, but much more sensitive. The next big advance is to replace the transistor with an op amp. ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net

the signal

e other from

oltage

of the

e l

eeds a

y spaced

spacing

sine

I

Nyquist

er

=A0 The

my

hat

layout

d I

you

p the

in a

width

f

o.)

less

shot

t

So

find a

em.

xt -

You might want to search on single photon avalanche photo-detection.

I haven't looked at the full paper but the list of references includes some useful stuff.

-- Bill Sloman, Nijmegen

signal

other from

Geiger-mode APDs are mostly a crock unless you need timing accuracy better than, say, 10 ns. Their dark count rate is a good six orders of magnitude worse than a PMT of the same area, and their dead time is 10 times longer.

On the other hand, they don't die if you put them in a helium atmosphere, and they last longer than 5 years.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net

signal

other from

They don't do this, either...

Hamamatsu makes these tubes. They look like basketballs.

John

d the signal

he other from

voltage

of the

ge

al

feeds a

ly spaced

y spacing

e sine

1 I

Nyquist

ver

=A0 The

my

that

e layout

nd I

e you

ip the

in a

dwidth

ff

oo.)

less

shot

it

=A0So

find a

h y t

them.

ext -

rt

s

Horses for courses. They can be a lot more compact and robust than PMTs - on which I'm rather more expert.

-- Bill Sloman, Nijmegen

the signal

other from

voltage

the

spaced

spacing

Nyquist

layout

the

bandwidth

Vain fathead.

John

d the signal

he other from

voltage

of the

ge

al

feeds a

ly spaced

y spacing

e sine

1 I

Nyquist

ver

=A0 The

my

that

e layout

nd I

e you

ip the

in a

dwidth

ff

oo.)

less

shot

it

=A0So

find a

h y t

them.

ext -

rt

s

t -

PMT's and APD's are too expensive, for a simulation, PD shot noise should be easy, or a zener. (how much does an APD cost, in hundreds?)

George H.

te:

add the signal

d the other from

he voltage

rid of the

et

ridge

gital

at feeds a

ually spaced

ency spacing

the sine

7541 I

han Nyquist

never

f. =A0 The

and my

Is that

the layout

ny

y and I

aybe you

clip the

e

all in a

bandwidth

stuff

l too.)

or less

edo shot

the

mm it

=A0So

ith

h I find a

with

ally

that

of them.

dtext -

short

ants

d a

che

nee

es

of

I've worked with PMT's - which is more than I can claim about SPAD's - and I've persuaded the IEEE that I do know a little about PMTs - see the IEEE Transactions on Electronic Devices volume 38 pages 679-680, published in March 1991.

You do need to base your abuse on something more than your right-wing intuition. Going off half-cocked like this does rather expose your enthusiasm for believeing what you want to believe despite the absence of any evidence to support your point of view.

-- Bill Sloman, Nijmegen

Sl>

tors. Then

ter

er than

urrent

t low

l random

y best.

a few mA

signal

er from

m a

r C

to

em

And the peer was peer of you brain-damaged right-wing friends? Such as James Arthur?

-- Bill Sloman, Nijmegen

Sl>>

transistors. Then

than

current

random

best.

few mA

signal

from

When's the last time you did any real electronics? Or any real work?

You resent me having James as a friend? I sure don't. He can cook, sing, and think. And ski, albeit sort of slowly. I mean, there's G, the slope, and the coefficient of friction of p-tex on ice. Why mess with that?

John

the signal

the other from

voltage

of the

feeds a

spaced

spacing

sine

Nyquist

The

my

layout

you

clip the

in a

bandwidth

too.)

less

shot

a

them.

net

-

All you do here is claim how "expert" you are, or maybe were, without ever making actual contributions.

You never *do* anything.

And when I don't have convincing evidence, I experiment and collect some. A mouse isn't a soldering iron.

John

te:

on.

ludes

ers of

10

Nothing that you'd be willing to acknowledge, particularly since most of my contributions are references to the publshed literature, a source that you seem ill-equipped to exploit

Not at the moment, and I find it frustrating.

You burble about 140dB of ripple rejection and then post the results of your experiment that shows 67dB. You've got a lab full of expensive equipment - if we are to take your boasting seriously - and boast like a lion about your "insanely good" electronic designs, but then post a result that might qualify you as a mouse in an undemanding environment.

Somebody with your experience ought to be aware that if you want 140dB of attenuation you are going to need more than a single stage of filtering - stray impedances usually make it difficult to get more than 60dB in a single stage.

-- Bill Sloman, Nijmegen

Hobbs

Burble? I questioned whether the Spice models of the c-multiplier were accurate at mid-frequencies. None came close to 140, or even 80, dB at frequencies where Early slope matters. So I tried some experiments.

and then post the results

This is the sort of thing you could do. It doesn't take expensive equipment to measure, say, 80 or 100 dB of ripple rejection. Just a little thought and patience.

So try it - for real - and we'll compare numbers.

If not for money, just to keep your mind from rotting.

My experience was entirely inadequate to make such a pronouncement.

I did some tests. Now I know more than I did before. You don't approve?

John

te:

wrote:

ction.

2

includes

acy

orders of

is 10

- e

Not to mention the transistors, the capacitors and the resistors - I don't keep a stock of components, and while I could get them, there is a mininum order charge which menas that I'd better order all of what I'd need in one hit

I'm still trying to use the gEDA circuit design software to prepare a schematic - and the concommitant parts list - for my oscillator project, which is more interesting, if still not interesting enough to motivate me to spend enough time rooting around in the gEDA hierachical design system.

My mind is fine, but my motivation isn't what it could be.

It probably isn't, but you clearly didn't think hard about the earlier experience you were getting while you were getting it.

It is certainly a step in the right direction - but one that you might have taken a few decades earlier. In your case, it obviously needs the right sort of customer asking the right sort of questions to get you interested. People with academic curiousity tend to get more out of the experience they have had because they put more time into making sense of their results. They often find that some academic had written up the whole issue a few decades earlier. It's hard to avoid re- inventing the wheel if you did't know that the wheel existed in the first place.

-- Bill Sloman, Nijmegen

e:

add the signal

the other from

e voltage

id of the

t

idge

ital

t feeds a

ally spaced

ncy spacing

the sine

541 I

an Nyquist

never

. =A0 The

nd my

s that

the layout

y

and I

ybe you

clip the

ll in a

andwidth

tuff

too.)

or less

do shot

he

m it

=A0So

th

I find a

ith

lly

hat

f them.

text -

hort

nts

a

he

ee

s

of

xt -

I've got a pair in my drawer that cost about $100 each, some ten or twelve years ago. I don't know if they work ...

-- Bill Sloman, Nijmegen

te:

wrote:

ction.

2

includes

acy

orders of

is 10

- e

Maybe he say you coming through the turgey(sp) wood.

The Early effect goes away at at high frequencies?

George H.

Hobbs

photo-detection.

includes

of

10

At high frequencies other parasitics become more important? Early gets swamped by Cce?

te:

ld

:

etection.

2232

es includes

curacy

orders of

ime is 10

han

's -

ee

0,

ing

ence

ut

Hmm OK... LTspice didn't seem to show that, (90dB of attenuation at

100kHz), but maybe I have to do some real measurements. I'm afraid I don't really understand the Early effect/ voltage.

George H.