How fast can an LED blink? How dim can it get? (revised title: Photomultipli
ElectricAye
Posts: 4,561
Greetings,
I'm trying to calibrate a homemade scintillation counter, which employs a surplus photomultiplier tube of unknown characteristics operating in a photon counting mode. One idea is to use an LED to help get some ballpark information by generating photons of a particular color (a particular energy level that can be correlated with the output pulse height of the photomultiplier). Ideally, I'd like to get the LED to spit out one photon at a time - ha ha - but I don't think that will happen. As a compromise, I thought about getting the LED to provide a pulse of only about 20-30 ns wide and as dim as possible, then maybe I could run the photons through a gauntlet of optical filters. I have a pulse generator that could probably give me an electronic signal of that shortness, but I can't find anything on LED data sheets that says what are the rise times, frequency responses, etc. of LEDs.
Anybody have any ideas? Clues about how/where to search for dim-witted "high speed" LEDs? Or some other, clever way to approach this problem?
my Planet of Ignorance is eternally grateful,
Mark
Post Edited (ElectricAye) : 1/27/2009 1:58:27 PM GMT
I'm trying to calibrate a homemade scintillation counter, which employs a surplus photomultiplier tube of unknown characteristics operating in a photon counting mode. One idea is to use an LED to help get some ballpark information by generating photons of a particular color (a particular energy level that can be correlated with the output pulse height of the photomultiplier). Ideally, I'd like to get the LED to spit out one photon at a time - ha ha - but I don't think that will happen. As a compromise, I thought about getting the LED to provide a pulse of only about 20-30 ns wide and as dim as possible, then maybe I could run the photons through a gauntlet of optical filters. I have a pulse generator that could probably give me an electronic signal of that shortness, but I can't find anything on LED data sheets that says what are the rise times, frequency responses, etc. of LEDs.
Anybody have any ideas? Clues about how/where to search for dim-witted "high speed" LEDs? Or some other, clever way to approach this problem?
my Planet of Ignorance is eternally grateful,
Mark
Post Edited (ElectricAye) : 1/27/2009 1:58:27 PM GMT
Comments
Depends on the LED.· You will need to take a look at the datasheet for a particular LED.· I have seen rise and fall times a low as 5nS, but this can vary greatly depending on the manufacturer and color of the LED.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
Beau,
That's encouraging news. Can you remember what brand that might have been or application? My problem is that, so far, I haven't found any data sheets at all that hint at frequency response. Ideally, I'd like something in blue, green, and red. Maybe even infrared.
thanks,
Mark
I can't remember the actual LED at the moment... but I do remember the 5ns rise and fall times.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 1/20/2009 9:34:21 PM GMT
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Phildapill,
I can't find it off hand, but there is some kind of quantum-oriented method of creating single photons, which I don't understand at all. It seems to me it has to do with bombarding a target with electrons and via some kind of interference effect(?), out pops a single photon. I'm probably totally wrong about this, but my gut memory tells me it's like generating a single beat with some kind of intricate beat-frequency generator. But I can't find the articles I saw recently on this. It seems to me it was a fairly recent development. What blew my mind is that creating single photons would be impossible because to measure it (and know it is only one and not two or three) would mean intercepting the photon and thus destroying it. But this technique avoided all of that.
Mark
My hope is to attentuate an LED of known wavelength (roughly speaking). I can observe a "rain" of photons, but I would like them to come at me in pulses so I know they are from the LED and not from, say, scintillations caused by cosmic rays shooting through my photomultiplier, etc.
That is why I'm looking for an inexpensive LED that can live up to the high expectations that my pulse generator will have for it. I want a nice little burst that I can distinguish from background, especially when it comes to the fringes of the photomultiplier's response curve - dark red and violet.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Single photon detectors are available from id Quantique:
www.idquantique.com
They use a single photon source in their RNG.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Just a thought.... As Leon suggests, filtering may ultimately be a better solution rather than trying to modulate an LED or a Laser. The reason is that temperature variations may (<- most likely will) produce different turn on and turn off times. It would be best to leave the source (LED or Laser) on all of the time during the test and filter the light to attenuate it. Not that temperature won't affect any of the filtering, but I would imagine that the effect from the filters due to temperature will be orders of magnitude lower than a modulation approach.
Depending on what equipment you have at your disposal, you may also be able to very precisely filter or attenuate your laser output with a phase delay recombination technique.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
To Kwinn, thanks for the info. Often just knowing that somebody somewhere at some time has done something similar to what you're trying to do provides enough encouragement to try that route.
thanks to all,
Mark
www.luxeonstar.com sells them in small quantity.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Think Inside the box first and if that doesn't work..
Re-arrange what's inside the box then...
Think outside the BOX!
You're going to have to get quantitative. Figure that with 1 mW of 633 nm red light, there are going to be about 10^16 photons per second. It will take many orders of magnitude of attenuation to get it down to a countable rate, much less one photon at a time. Not that it can't be done, as suggested.
I think the pulse height produced by a single event in a PM tube may not be a very good indication of the energy level of individual photons. The tube will have a certain quantum efficiency, i.e., maybe 10% of incident photons will be converted into an electron shaken loose from the photocathode, and then there will be a very stochastic conversion of that one photoelectron into a cascade of secondary electrons. It is not as if a single photon of blue light is going to cause a bigger pulse amplitude than a photon of red light.
Never expose a PM to daylight or any "bright" source while it is powered up. Or even when it is just sitting there. If it survives at all, it may take days for it to quiet back down.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
As I understand it, a laser functions by having photons (plural) hammering up and down a cavity, knocking excited electrons down a shell, releasing more photons in lockstep with the photon that did the knocking.
How is this going to release single photons?
Steve
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
To Tracy Allen,
Your point about the randomness of the amplification process is a very good one. I can see how a single blue photon would not necessarily give more pulse than a single red. But actually I'm not really aiming at getting down to a single photon. I had hoped to get down to a rain of photons that I think my photon counter can handle (which is not yet determined but I'm guessing 1 MHz). As for measuring that rain of photons: my system repeatedly scans through energy outputs of the photomultiplier (PMT) by periodically changing the threshold on a comparator. All the results are written to an SD card whose data later gets crunched through an Excel sheet. Since I'm using a statistical approach at this point, is it possible that, say, a measurement of red photons will show up as a lump in one place on the energy graph and a pile of, say, blue photons will show up at a higher energy level??? I was hoping this would be the case since blue light (I think) has roughly 50% more energy than red. But I suppose for my entire system, any such "bumps" in the graph could get so spread out as to be undetectable.
I guess I should state that, at this point, I'm not trying to get some sort of absolute calibration of my system out of this. I'm just trying to get some very ballpark results that can provide a sanity check on all that is going on. At the very least, I want to ascertain that the pulses I see on my oscilloscope are indeed caused by single photon events (and not photon pile ups) and that the "noise" pulses I'm seeing are not actually single photon events. Since I'm using surplus PMTs, I'm feeling very insecure about their performance. Thus my quest for some sort of ballpark testing method.
thanks for providing some insights on this,
Mark
I was wondering if, instead of using a laser, I use some kind of laser line filter or interference filter? Since I'm trying to get approximate results, I don't think I need the precision of a laser at this point. Would that approach make sense?
a thousand thanks,
Mark
And to metron9,
thanks for the link. Interesting decorative ideas my wife my like.
cheers to all,
Mark
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 1/22/2009 4:01:17 PM GMT
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Think Inside the box first and if that doesn't work..
Re-arrange what's inside the box then...
Think outside the BOX!
For a given energy, a single 350nm photon carries twice the energy of a 700nm photon, so a large number of UV photons will carry about twice the energy as the same number of near IR photons. PM tubes are more sensitive to blue than to red, that is, the quantum efficiency is greater for blue than for red, but if a photon succeeds in shaking loose a photoelectron, the resulting pulse height will not convey information about the color of the photon that caused it. Don't quote me on that, though. There are so many clever inventions!
Even lasers are multimode, meaning there is a spread of wave energies into several distinct or overlapping peaks. Single mode lasers are available that emit most of their energy in exceedingly narrow bandwidth. For example, some VCSELs (vertical cavity surface emitting laser) have that characteristic and furthermore, they can be modulated or switched on and off at gigahertz frequencies. The narrow bandwidth comes from the Bragg mirror structure, an mirror built up of precisely spaced layers of different refractive index built parallel to the surface, and a very short and narrow cavity that allows one mode of oscillation and fast response.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com
Post Edited (Tracy Allen) : 1/22/2009 5:52:26 PM GMT
Tracy Allen,
Sorry for quoting you on this but your point is of extreme interest to me. What you are saying is that the energy of a photon only affects the quantum efficiency (the probability of whether or not a given photon will start the cascade of electrons that finally results in an output pulse). Okay, I can see how that might be. But when I watch the output of my photomultiplier on an oscilloscope, I see pulses that vary in height by at least an order of magnitude. Some of them are probably scintillations caused by cosmic rays hitting the tube or from radioactive breakdown of things like potassium 40 inside the tube itself. Such a particle reaction can release a bunch of photons all at once and thus give a big tall pulse. But when I shine a dim light at the hookup wires (which act as a crude light guide, I think), I'll see a wild mess of pulses, all of different heights. I suppose it's possible that these pulses are actually "pile-ups" (superpositions) of many smaller pulses and not the energy differences of differently colored individual photons. If that is the case, then I guess my present statistical approach is not going to work the way I thought it would. I would need to add some kind of color filter wheel or something. Hmmm..... I might have to do more homework on this. I'm suddenly feeling really stupid.
I greatly appreciate your thoughts on this,
Mark
Photomultiplier tubes produce quite a bit of random thermionic noise at room temperature and that is probably what you are seeing. In order to separate a usable signal from this noise a liquid scintillation counter (counts Beta particles in a liquid fluorescing medium) uses 2 PMT's and only counts events simultaneously appearing on both PMT's.
The coating on the PMT is sensitive to various wavelengths, but there is a minimum energy required to free an electron. Any photon below that level can not be detected by that PMT, and any photons that have multiples of that energy can free more than one electron.
You can use a led with an entry slit, an optical grating or prism to split the light from the led into a spectrum of its constituent wavelengths and use a narrow slit or small hole in front of the PMT to pick out a very narrow range of wavelengths. Doing so will also greatly reduce the intensity of light.
Ah, yes, thank you. I see the thermionic noise as a phosphorescent spiky shag carpet in the 5mV to 10 mV range. And your comment has cleared up another mystery for me: the reason that I'm seeing single photon pulses produce pulses of various heights (and not just one common pulse height for each photon no matter what color) is because the process of electron multiplication from one dynode to the next is also something of a quantum probabilistic phenomenon. So not every electron jumping from the first dynode to the second dynode always generates more electrons, thus the resulting pulse heights end up being different in the end.
In other words, photon A and photon B might have identical wavelengths and they might both knock loose an electron from the photocathode as they enter the photomultiplier. But what happens to those electrons afterwards, as they jump from dynode to dynode, will probably be somewhat different, so the resulting output pulses will not be identical in height. And those are the pulses I see when I shine a flashlight against my hookup wires.
thanks to you guys, I'm beginning to see the light!
Mark
First, the sodium iodide crystal and PMT measured gamma rays with energies down to below 15Kev and the spectrum is well above the background noise and consistent enough to separate isotopes based on pulse heights.
Second, the liquid scintillation counters detect even lower energy events that are buried in the tube noise and require 2 tubes with coincident detection to separate from that noise. In spite of that a pulse height analyzer can still be used to separate one isotope from another.
I am beginning to wonder if it is possible to measure single photon events without cooling the PMT to a very low (liquid nitrogen or lower) temperature, and using a very sensitive low noise PMT.
When an electron leaves the surface of the photocathode, it is attracted to the next more positive electrode.· In "falling" to that electrode, it acquires enough energy to knock off more than one electron when it hits the surface.· Some of these will fall back, but we hope that more than one will fall to the next-more-positive electrode, knocking loose a still greater number of electrons to continue the process to the next electrode.· That's where the gain comes from.
The number of new·electrons knocked loose by each falling electron is random, and so is the percentage of these that will fall to the next electrode instead of merely going back whence they came.· By the time the cascade goes through several stages (several electrodes) and showers the final anode with electrons, the number of electrons that land will be very large, but will vary greatly.· For each single·photon, we then get a big shower of electrons at the end of the process.· How many?· It depends on the geometry of the electrodes, on the voltages applied, on the condition of the photomultiplier tube, and on several million throws·of the dice.
Each individual photon, then, ultimately results in a very variable number of electrons hitting the most-positive anode.· Since every electron has the same charge regardless of its energy, your pulse height (which is a measure of·instantaneous current)·is simply a measure of the number of electrons ultimately reaching the most positive anode.· That's a random number, totally unrelated to the energy of the original photon.· Its maximum may, however, be a measure of the gain of the photomultiplier and thus the condition of the secondary-emitting electrodes in the tube.
So, no, your pulse height tells precisely nothing about the energy (color) of the original photon.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net
Post Edited (Carl Hayes) : 1/23/2009 4:27:31 AM GMT
(1) The intermediate electrodes are usually called dynodes, which adds precisely nothing to the explanation.
(2) If you've got a PM system you can play with, you can prove all of this by admitting a very attenuated monochromatic light.· It has to be attenuated enough so that you can see individual pulses resulting from single photons.
(3)· If you try this, be careful.· PM tubes usually have magnetic shields around them.· These are metal, and typically at several thousand volts negative with respect to ground.· They bite.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net
As I thought I'd understood this, when a gamma ray hits a scintillator (like a NaI crystal, etc), it does so with enough energy to blast a "large" number of photons from the crystal. This crowd of photons then hits the photocathode of the PMT almost all at once, knocks loose a crowd of electrons almost all at once, whose numbers then get multiplied as they cascade from dynode to dynode inside the PMT. So the final output pulse from a gamma ray vs. NaI is actually the descendent of many nearly-simultaneous photons. Different isotopes produce gamma rays of characteristically different energy levels, which then knock loose from the NaI crystal characteristically different numbers of photons. By measuring the pulse height(s) caused by a particular lump of something, we can generate its gamma ray spectrum and, from that, identify its isotope.
I've been under the impression that liquid scintillation counters worked much like the gamma ray/NaI process mentioned about except that the efficiencies for the liquid scintillation counters are much lower, so they have to count pulses for a longer period of time. Consequently, liquid scintillation counters are more prone to picking up flashes of light caused by cosmic rays, etc. shooting through the PMTs themselves. To cut down on those false flashes, they use two PMTs and rely on coincidence counting. But I didn't think that the pulses per se were buried inside the thermionic noise of the PMTs. In other words, I could see how you want to discern a bright flash inside the liquid vs. a bright flash caused by a the radioactive decay of potassium 40 in the glass of the PMT, but I sorely hope that my photon pulses can be seen above the thermionic hiss even without coincidence circuitry. If not, then I'm really confused about what I'm seeing on my oscilloscope screen.
thanks for your help,
Mark