Don't know.
The thing to do would be to put the detector in a dark box with an LED, and pulse the LED. The optical path would have to be such that you'd expect, say, 0.1 photons to hit the detector per pulse. Then analyze the statistics.
Just had this weird thought: connect two leds in parallel. Expose one to light. The other one should light up. The efficiency might be low.
Thanks, but I really don't have time right now.
-- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
I would suggest you do a rough calculation to see how many 'room light' photons are likely to intersect the LED chip per second. I suspect it might be a few more than 5000.
-- Regards, Adrian Jansen adrianjansen at internode dot on dot net Note reply address is invalid, convert address above to machine form.
"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...
Setup a candle. But from the scope traces, it sure looks like something.
Cheers
ges
out
ust
h a
ry
ythat
ed some
way?
Well I do have a source of single photons, but they are random. Is that OK. (light bulb at low voltage, through green interference filter.) But linear is enough for me... If it takes two photons, then it should go as intensity squared.
The dark count is low... but there also appears to be some hysteresis in the avalanche voltage... after it photo-conducts it keeps dribbling on, if you set it near the edge.
I think I saw that the GaP leds have poor effeiciency.
That I understand.
How 'bout this,
A follower to drive a cable to 50 ohm counter.
+V bias--+---+ | | LED--> ^ | - | | |/ +--| |\> +------cable---+ | | 1k Ohm 50 Ohm | | +--------------+ GND GNDWell, I need something else between the 1k ohm and cable. Maybe use a darlington?
George H.
ed text -
t
th a
ly
at
ey?
tde quoted text -
Hi Adrian, you most likely weren't following the thread on SEB. The 'gain' is a strong function of the bias voltage. At 24 Volts I get zero counts in room light, at 24.5 Volts I get ~400 counts, at 24.9 I get 4k counts, at 25 volts it starts to avalanche with no light.
I'm not sure how to think about the very low efficiency. I've heard talk about single 'channels' in the avalanche process. So the 'right' area, might be a channel area? (whatever that means)
George H.
I am trying to see how you plan to have that work?
The ASCII art is confusing in terms of the LED orientation. One way it looks like your operating it as a PV mode, which that will only yield up to what ever the forward break down of the LED is, in which case, you have to over come the one diode drop in the Base-Emitter.
If your intention is avalanche mode, then, you still need enough to get over the one diode drop at the base, which you can do this however, by the time you have enough photons in that LED, it then looks like a high band of noise or, if any noise at all other than a slight ripple of DC?
To keep things at a simple level, you could use a JFET with a drain bias R that has a trimmer in it to adjust the knee threshold to pinch it in the off state. The LED would then be able to trigger it at the gate. JFETS have low charge at their gates so this may work well.
I would suggest a 2N7000 type MOSFET however, I don't think the base capacitance (charge) is going to allow for the 1 uA average current of the LED to remove the charge fast enough to give you a response that would be usable?
You must consider the time constants of charge verses what the diode will be able to deliver to over come that. Going by memory, I don't think there is enough area to produce the current required as a real PV cell would generate.
Then of course, we could always select a JFET front end amp. A BiMos I think they call them. TL081 or something of that order as a comparator.
Jamie
The best I've heard of is bursts of 30 or so photons in the blue.
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
Avalanche is dominated by impurities. A transition metal atom acts to 'short circuit' the bandgap, generating leakage current (and seeding avalanche events), trapping charge carriers (acting as recombination centers*), and putting a bump in the electric field (acting to worsen avalanche in the local area).
*Hence the use of gold doping in some high-speed devices, including the 2N2369 and 7400 TTL logic series. It reduces both the breakdown voltage and storage time. (I don't know for sure what TTL would've operated at without gold doping, but contemporary processes were doing 30V in NPN bipolar (LM741, etc.), and 15-18V (CD4000 series) in metal-gate CMOS. TTL breaks down at 7V.)Possible the effect the OP describes is due to a small photosensitive impurity in the crystal. LEDs may generally have low leakage, but it's an uncontrolled amount (they only spec them to 5V!), so it shouldn't be surprising there may be poor behavior, like strong avalanche amplification regions, or photosensitivity (or both), etc.
Tim
-- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
What you report is rather interesting. So, assuming they are indeed responding to individual photons, is there an estimate to "efficiency" like one of 10^20 photons are detected?
On a sunny day (Mon, 30 Apr 2012 12:41:33 -0700 (PDT)) it happened George Herold wrote in :
Very nice paper, the voltages seem really high, like 130V + From the spectral curve it seems they are only sensitive to 'light'. That would not stop me from testing them with some radioactive material however. You can order Thorium containing welding rods on ebay that are a lot safer to handle than those gass light mantles, the welding rods contain about 2% Th IIRC, and give several times background on a GM counter, But as you see no background at all.... n x 0 = 0.
Phil knows all about those diodes.
s
out
just
th a
ery
ly
that
some
way?
oDC?
nse
IHi Jamie, (Say I thought for sure you'd like a few of these LED's.)
The circuit was just a late night thing. The LED is reverse biased. I origianlly drew up a follower. Maybe like this,
+V bias--+---+ | | LED--> ^ | - | | |/ +--| | |\> 100k-> R +------cable---+ | | | bias-> V 1k Ohm 50 Ohm diode - | | +--- +--------------+ GND GNDBut then I thought it was a waste to use most of the electrons in the 100 k resistor.... so I cut it out. (I'll give it a try) (and hang my head in shame when it doesn't work.)
George H.
'
e and
hout
ks
an
nThanks Tim, What's different about the above LED (and also low power ~20V zeners) is that it looks like the whole device discharges at once. You get a big spike of charge, then the whole thing recharges for another blast.
"(they only spec them to 5V!), " Yeah that's just strange. Most of the LED's I tried I couldn't see any leakage (at the ~10nA level) for voltages up to 50V.
George H.
ages
t t aat
ied some
y?
- Hide quoted text -
What's the area of an LED? I don't have a good idea for this number. I also have to take the lens into account. Then there's the unknown spectral response. But on Friday I did some comparisons between a photodiode and LED. I could do some handwavy upper bound from those numbers. So with a yellow LED shinning on a PD and then the reversed biased LED I generated 25uA of current in the PD and 2500 counts/ second in the LED. (that's 4x10-16 'equivalent amps'... so yeah an effeciency of at least 10^-10. :^)
The photodiode has an area of 16mm^2.
What is more fun is the 'gain' of the LED. If the total charge delivered is C*V (~10pF, 100mV) then a gain of 10^7. It's like a PMT, but with no 'messy' high voltage.
George H.
rge
.gge=3D
owever.
r% Th IIRC,
I was reading in Sze (2nd edition page 101) about the breakdown voltage in GaP versus the impurity concentration. (They have to dope GaP to make it an LED.) at 10^17/cm^3 the break down is near 25 volts and climbs to over 100 at 10^16. So I'm guessing different doping levels.
That paper also talks about hysteresis effects.
George H.
Hide quoted text -
You could use the LED itself as a photodiode, at a lower non-avalanching voltage. That would roughly calibrate the photon rate hitting the junction.
I'm guessing you aren't detecting single photons to any meaningful extent.
-- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
Rods apparently respond to single photons, but a cluster of rods must be hit to be perceptible.
Cool stuff.
I wonder why we don't respond to fewer or single photons. It would seem to be useful.
As I've noted, I was able to detect the light from a green LED at about 700 pA DC current. The LED was emitting, very roughly, something like a million photons per second. My retina was probably getting 1% of them, poorly focussed into a fuzzy blob.
-- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
--- The fourth sentence from your link reads: " If we could consciously see single photons we would experience too much visual "noise" in very low light, so this filter is a necessary adaptation, not a weakness."
Also, if instead of a single photon triggering a reactive response from a single rod, several photons from the same source impinging on a group of rods did, then directional information could be gleaned from the beam/flash.
Why else, if you see something out of the corner of your eye, do you swing around to see where it's coming from?
-- JF
I disagree with that. My night-vision intensifier lets me see stuff that I can't see with my eyes, useful stuff. If I could resolve single photons, I'm sure I could figure out where walls and doors and trees and tigers were. Some nocturnal animals are reported to have single-photon sensitivity.
Why wouldn't you get directional information from a single photon? What you would see would be a point-source flash of light, same as if you saw a multi-photon flash.
The only drawback to single photon detection might be a high background rate in total or near-total darkness, like when you were asleep. I have that already, a high continuous phosphene rate, and it doesn't disturb my sleep.
-- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
Not unusual for some regimes of avalanche -- 2N3904s reliably go from reasonable open circuits ( the LED's I tried I couldn't see any leakage (at the ~10nA level) for
How big are the actual spikes, and what fall time? You said you measured through an amplifier, so perhaps the amplitude isn't much to begin with?
The risetime on the scope shots wasn't very impressive, likely due to slew rate limiting. The actual performance of these phenomena can be very fast indeed! (The next fastest phenomena are step recovery in specially doped diodes, ranging from 300 down to 20ps or so, and all-out monolithic "shock line" generators, usually implemented in long InP schottky junctions I think ($$$!). Anything less than that is optical domain -- femtosecond lasers and such.)
Tim
-- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
oltages
out
RC
just
th a
ery
ly
that
tried some
way?
is
ed?- Hide quoted text -
Hey, that's a good idea!
Hmm, not sure what you mean by 'meaningful'. But I've got a few uses for a random pulse generator. I've got count rates up above 10kHz, but the pulse width is still 1-2uS. If I can get that down to 100ns and 100kHz pulse rates. That would be wonderful.
A big pile of work today, and no time to play.
George H.
ed text -
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.