If you need to measure the presence of photons down to a very small number of them, you are looking at the use of a photomultiplier, as explained in a recent video by [Huygens Optics] on YouTube. The only way to realistically measure at such a sensitivity level is to amplify them with a photomultiplier tube (PMT). Although solid-state alternatives exist, this is still a field where vacuum tube-based technology is highly relevant.
Despite being called ‘photomultipliers’, these PMTs actually amplify an incoming current (electron) in a series of dynode stages, to create an output current that is actually easy to quantify for measurement equipment. They find uses in everything from Raman spectroscopy to medical diagnostics and night vision sensors.
The specific PMT that [Huygens Optics] uses in the video is the Hamamatsu R928. This has a spectral response from 185 nm to 900 nm. The electrode mesh is where photons enter the tube, triggering the photo cathode which then ejects electrons. These initial electrons are then captured and amplified by each dynode stage, until the anode grid captures most of the electrons. The R928 has a gain of 1.0 x 107 (10 million) at -1 kV supply voltage, so each dynode multiplies the amount of electrons by six, with a response time of 22 ns.
PMTs are unsurprisingly not cheap, but [Huygens Optics] was lucky to find surplus R928s on Marktplaats (Dutch online marketplace) for €100 including a cover, optics and a PCB with the socket, high-voltage supply (Hamamatsu C4900) and so on. Without documentation the trick was to reverse-engineer the PCB’s connections to be able to use it. In the video the components and their function are all briefly covered, as well as the use of opamps like the AD817 to handle the output signal of the R928. Afterwards the operation of the PMT is demonstrated, which makes clear just how sensitive the PMT is as it requires an extremely dark space to not get swamped with photons.
An interesting part about the demonstration is that it also shows the presence of thermionic emissions: anode dark current in the datasheet. This phenomenon is countered by cooling the PMT to prevent these emissions if it is an issue. In an upcoming video the R928 will be used for more in-depth experiments, to show much more of what these devices are capable of.
Thanks to [cliff claven] for the tip.
sure would be a shame if somebody used semiconductor tech in photonics to build a monolithic photomultiplier that was only a few mm tall…
APDs still are not as good as PMTs… Noisy, poor stability, small apertures, smaller dynamic range…
Semiconductor single-photon detectors have nowhere near the gain and noise characteristics of these avalanche-style tubes.
“Silicon photomultipliers” (which are really just a huge array of APDs operating in Geiger mode) functionally have similar gain to PMTs (~1E6): it’s mainly just dark current that kills them as pure replacements – few thousand times higher. There are other things that can be killers, too, but it’s usually dark current.
It’s more complicated than saying “they’re noisier,” though: with PMTs, the gain is statistical and staged, meaning that for a gain of 1E6 you don’t get 1E6 electrons every time for a single photoelectron – you get some distribution. The later stages don’t matter much because there are lots of electrons being multiplied, so things average out – but the early stages add a lot of fluctuation due to statistics.
This gets lumped into a term called the “excess noise factor” – for PMTs it’s typically around 1.2 or so (depending mainly on the first dynode gain). Because APDs in Geiger mode functionally just are 1 or 0 (‘hit’ or ‘not hit’) the ENF can be very low (close to 1). When you add everything up SiPMs can be competitive or better in a lot of use cases.
The microchannel plate is a very good photomultiplier/ imager only a few mm thick. https://en.wikipedia.org/wiki/Microchannel_plate_detector
The focusing grid being “ahead” of the photocathode is just because this is a side-on PMT, which is cheaper to make (but worse performance all around, I think) than a “head-on” PMT. The focusing grid has nothing to do with light – it’s just there to shape the electric field to guide the photoelectrons to the first dynode.
Head-on PMTs (which are what all the larger PMTs are) deposit the photocathode on the inside of the glass, and then the focusing electrodes are just ahead of the first dynode.
“Sensitive” is also a tough thing to define compared to your instincts: PMTs aren’t actually that good at capturing individual photons, with a quantum efficiency usually around 20% or so – compared to a CCD where QEs can get to ~90%+, that’s pretty bad. Where they excel at is their noise level – the dark current – which is exceptionally low compared to alternatives. That’s why PMTs are great for photon counting – lots of detectors can respond to single photons, but their noise levels are high enough that the peaks due to individual photoelectrons are washed out.
I have a PMT in a box in the garage. It and its socket are sitting in its original box.
I picked it out of a discard pile at a former workplace. I haven’t tried monkeying around with it yet.
I’m in a similar situation, with about a dozen 931-A tubes, and a couple of other oddball photomultiplier tubes.
Once in a while, I start delving into what I’d need to build something with them, e.g. to measure ionizing radiation. But I think I last stopped when I compared the 931-A sensitivity range (400 ±500 nm) to the output spectrum of several commonly available (read: affordable) scintillators, and found them to be quite different. I could be wrong. Heh, there’s my rabbit-hole for the rest of the weekend!
For anyone interested, the Hammatsu R928 is only $1000, and is available through WalMart (online ordering.)
Ha!
I remember working with these, it was really fun.
And quite simple once you manage to give it stable supply and quiet environment.
One of my buddies picked up a gigantic pmt for another one of my buddies at a hamfest a few years ago. This has to be 12″ across if not larger. We are a few hours away from howe caverns and they have places in there that are not normally accessed. I have been thinking it would be fun when my pal retires to see if they would help us place it in one of those spaces. This would be interesting as there is also no power and no wifi or cell service. So it would need to be battery powered and log it’s data (semi) locally.
When they fired one of the first lasers at the moon they didn’t expect anything but a PMT on the telescope got a single photon or few not more and the timing was right on proving thanks to that low noise floor.