PMT preamp

[Dajgoro]

Hi.

I have two soviet PMT tubes feu31 and feu68.
I need to build a preamp for those tubes so that I am able to see single photons when using a regular CRO scope.
I already built the power supply, and I recycled an old uhf antenna preamp, but even with that I can't see single photons.
For now it can only detect very faint light flashes, but that is it.
I found some schematics on the web, like this one:
http://www.amptek.com/products/dp5g-oem-digital-pulse-processor-for-scintillators-and-pmts/

I tried to build something like that, but it didn't work. I never had luck building hi gain stuff with opamps.

I also found this blog posts:
http://doggie2.blog.so-net.ne.jp/2012-05-04
http://doggie2.blog.so-net.ne.jp/2012-06-10
Which describe what I am trying to do, but there is no schematic that I can find, and google translate gives out a lot of garbage when translating that.

[ejeffrey]

Do you know what the requirements are?  You need to know the PMT gain, which will tell you the total charge you need to detect and the bandwidth to know how fast the pulse is.  You may also need to deliberately slow it down, the PMT may be faster than your scope.

[plesa]

I will stay away such an old PMTs, their dark counts can be quite big. And rather pickup some on e*ay. For the photon counting PMT you can use eg OPA657 with 1,6 GHz GBP.
I reccomends to desigh some cooling for PMT to reduce the dark counts to minimum.
Also check the datasheet for the voltage distribution on dynodes for photon counting, it is little bit different to normal analog mode distribution ratio.

[David Hess]

Isn't this a place where a high bandwidth low current noise transimpedance amplifier like those used with photo diodes would be appropriate?

If I was digitizing the output and did not care about frequency response, I would try using an integrator as a charge amplifier with a servo loop to remove low frequency and DC drift.

[Dajgoro]

When reading the Russian datasheet, I can see in the text that those two PMT-s are meant to be used for gamma spectroscopy. I have those two PMT-s and I can't get anything else.
I do have some high speed LT opamps that could do the job, but I need a decent schematic for that.

This project is meant to be a project for a school/science fair that where I am assisting, so it is more of a demonstration thing, than a serious device.

[David Hess]

There are lots of application notes and tutorials discussing photodiode transimpedance amplifiers online.  I would start with a simple design using a low input current (FET input) high bandwidth operational amplifier and I would include input protection using low leakage current diodes.

[MK]

Horowitz and Hill the EE textbox, Hill even posts on here sometimes. has a simple circuit to enlarge and stretch each pulse from single photons, looks like a simple circuit to build too

[mikeselectricstuff]

See if there are any appnotes on the Hamamatsu site - they make lots of PMTs

[ejeffrey]

The circuit is trivial, especially if you use a high speed opamp.  It is a load resistor, gain resistors chosen from the data sheet, and and and output resistor.  But you need to know the gain and bandwidth you want.  At the very least, you need to specify the bandwidth of your scope!

Are you sure you are not saturating the PMT?  That is super easy to do, even if you think it is in complete darkness, and would show up as you describe.  Most PMTs can produce signals big enough to see on a scope without further amplification.

[plesa]

See if there are any appnotes on the Hamamatsu site - they make lots of PMTs

https://www.hamamatsu.com/resources/pdf/etd/PMT_handbook_v3aE.pdf

and lot of info about transimpedance amplifier can be found in
http://www.amazon.com/Photodiode-Amplifiers-OP-AMP-Solutions/dp/007024247X

or useful for beginners
http://www.jensign.com/transimpedance/index.html

As others mentioned the photocathode of PMT is extremely sensitive and exposure to ambient light when HV applied will damage PMT.
Send me PM and I can send you some PMT newer than Russian types you mentioned.

[Dajgoro]

The circuit is trivial, especially if you use a high speed opamp.  It is a load resistor, gain resistors chosen from the data sheet, and and and output resistor.  But you need to know the gain and bandwidth you want.  At the very least, you need to specify the bandwidth of your scope!

I have an 1970' Hameg HM512 scope, which is able to display 60MHz. The final goal was to try to sample those signals with a digital system. If possible to use them for some crude gamma spectroscopy, if that won't work, then just as a photon counter. I also have a plastic scintillator.

Are you sure you are not saturating the PMT?  That is super easy to do, even if you think it is in complete darkness, and would show up as you describe.  Most PMTs can produce signals big enough to see on a scope without further amplification.

I kinda exposed it to some light. Oops. Luckily I have two pmt-s. I tested it by leaving it in the original cardboard box. When I open it just a little bit and flash with a led i can see the signal on the scope clearly with my jerry-rigged uhf-modded preamp. The thing that puzzles me is that when I leave it in the cardboard-dark I can see soma garbage coming out, and i thought that it might be light leaking, but then it goes away, and comes back. I tried staying still, making it darker, but seems that that wasn't the source. The signal looked completely random, and when I connected my headphones to the amp the sound was the same as when turning a crusty volume pot on the stereo.

Next step I need to figure out an exact schematic for that opamp preamp.

[Artraze]

I have a similar project in the eventual works, so while I unfortunately haven't designed the preamp yet, I can maybe offer some advice:

Plastic scintillators are extremely fast, with ~1ns rise time and <4ns decay time.  Your PMT was almost certainly developed for use with crystal scintillators which are two orders of magnitude slower and so may not be able to accurately capture the light pulse.

It's also worth mentioning that if you're planning on doing spectroscopy, a 'photo counter' isn't technically what you want...  Normally those refer to an amp with a comparator to detect pulses beyond a threshold.  Spectroscopy requires a pulse size measurement, which means integrating the detector current for the duration of the pulse, measuring that, and resetting the integrator.

[Wim_L]

Dark counts, for new and good photomultipliers, can be in the thousands per second. Even those specifically selected for low dark count still can have a few hundred per second, unless cooled.

Other than that, I can't say I ever needed a preamplifier to see the pulses from any PMT I've ever used on a scope, as long as the resistive divider and power supply were providing adequate amplification inside the PMT. That's the whole point of the PMT, if properly set up, it is the amplifier. Very low bandwidth scopes might have some trouble with it, as the pulses tend to be quite short (this depends on the particular model of tube). A low bandwidth scope might flatten the peak into something that's harder to see.

[Dajgoro]

When looking at this blog post it is clearly doable:
http://doggie2.blog.so-net.ne.jp/2012-05-04
The datasheet says 3M max, so I used 220K.

And how can scintillators be fast or slow. If it is single photon detection it either works or it doesn't?

I also have a xray cartridge for illuminating xray photos. I also tried cutting out a piece of that and sticking it to the pmt. No result (illuminated with various weak sources). When placing that smoke detector Americium sample on that cartridge I can see the faint green glow, so it must be quite a bit of photons.

[Artraze]

And how can scintillators be fast or slow. If it is single photon detection it either works or it doesn't?

Scintillators work a little like glow in the dark powder.  The gamma ray 'charges' it up with energy and then the crystal releases that energy by glowing.  How fast the glow starts and how long it lasts are the crystal's rise times and decay times respectively.  You can find them on the datasheet.  For comparison, here is the datasheet of a common inorganic crystal.

Perhaps a point of confusion is that when you are using a scintillator you aren't detecting single photons (unlike using a Geiger tube, for example).  When a gamma ray hits a scintillator in deposits a lot of energy which is then released as multiple photons, usually about 10-40 per keV depending on the scintillator.  So while you may be detecting a single gamma photon with your scintillator, your PMT is actually detecting multiple photons emitted by the crystal.  This is also how spectroscopy works: by determining the total light detected in a pulse you can estimate the energy of the gamma ray that hit the crystal.

If all you are concerned about is detecting if a gamma struck and not it's energy, the ultra fast plastic scintillator shouldn't be a problem.  The PMT should be able to detect the light burst no problem.  However, the output may suffer from some non-linearity and the tube's own rise time limitations making accurate spectroscopy difficult and the SNR poor.   

[ejeffrey]

I have an 1970' Hameg HM512 scope, which is able to display 60MHz. The final goal was to try to sample those signals with a digital system. If possible to use them for some crude gamma spectroscopy, if that won't work, then just as a photon counter. I also have a plastic scintillator.

That is a pretty slow scope compared to the normal PMT output, so it is likely that the scope is filtering a good bit of your signal.  If you increase the load resistor you will make the signal bigger and slower, and you should have no problem seeing it on your scope.

I kinda exposed it to some light. Oops. Luckily I have two pmt-s. I tested it by leaving it in the original cardboard box. When I open it just a little bit and flash with a led i can see the signal on the scope clearly with my jerry-rigged uhf-modded preamp.

Cracking open a cardboard box with any kind of room lights on and flashing an LED is enough to completely saturate your device.  Assuming you don't blow out the PMT you will still see a signal, but gain will be dramatically reduced and you will be measuring a continuous light level, not single pulses.  PMTs are even sensitive to light when off -- exposure to light will cause their dark counts to temporarily go way up, but will recover over time if you run them with a bias voltage in the dark for a while.  When you shine a suitable pulse of light on a PMT you should not see a single signal pulse corresponding to your illumination level as you would expect from a photodiode.  You should see a burst of single-photon pulses of roughly the same height, with the density of pulses proportional to the intensity of the light.

The thing that puzzles me is that when I leave it in the cardboard-dark I can see soma garbage coming out, and i thought that it might be light leaking, but then it goes away, and comes back. I tried staying still, making it darker, but seems that that wasn't the source. The signal looked completely random, and when I connected my headphones to the amp the sound was the same as when turning a crusty volume pot on the stereo.

PMTs have dark counts which are strongly dependent on temperature and the cathode material, and also vary with time.  You could have been seeing this.  To be clear, a working PMT biased correctly and exposed to no light should show random pulses of similar shape, but the pulse amplitudes will vary due to gain noise.  The amount of gain noise depends on the device, but you might get a variation in pulse height of ~50%.

When bringing up a PMT for photon counting you want to first make sure it is in absolute darkness.  Then look for dark counts.  Try setting your scope to 1 mV/div and trigger threshold to -5 mV.  Once you see dark counts, you can try *very* carefully increasing the light level and watching the pulse density go up.  It won't look like much on an oscilloscope: the probability of getting two pulses in a single sweep is still rather low, but you should be able to see something.  If it ever stops pulsing, lower the light levels: you are saturating it!  At this point you want to implement a discriminator and hook it up to a counter.  Then you will clearly be able to see the increase pulse density as you increase the illumination level.

[Dajgoro]

However, the output may suffer from some non-linearity and the tube's own rise time limitations making accurate spectroscopy difficult and the SNR poor.

It is only for a demonstration, so precision isn't critical.

How do the values in the resistor ladder affect the performance. I noticed resistors from 100k up to 1M.

[plesa]

Dark counts, for new and good photomultipliers, can be in the thousands per second. Even those specifically selected for low dark count still can have a few hundred per second, unless cooled.

Other than that, I can't say I ever needed a preamplifier to see the pulses from any PMT I've ever used on a scope, as long as the resistive divider and power supply were providing adequate amplification inside the PMT. That's the whole point of the PMT, if properly set up, it is the amplifier. Very low bandwidth scopes might have some trouble with it, as the pulses tend to be quite short (this depends on the particular model of tube). A low bandwidth scope might flatten the peak into something that's harder to see.

It is almost impossible to detect photons due to the scope input capacitance. So preamplifier is a must.
Depends on the enviroment magnetic shielding including the photocathode insulation (to prevent leakage) can be handy.

[vindoline]

Hi.

I have two soviet PMT tubes feu31 and feu68.
I need to build a preamp for those tubes so that I am able to see single photons when using a regular CRO scope.
I already built the power supply, and I recycled an old uhf antenna preamp, but even with that I can't see single photons.
For now it can only detect very faint light flashes, but that is it.
I found some schematics on the web, like this one:
http://www.amptek.com/products/dp5g-oem-digital-pulse-processor-for-scintillators-and-pmts/

I tried to build something like that, but it didn't work. I never had luck building hi gain stuff with opamps.

I also found this blog posts:
http://doggie2.blog.so-net.ne.jp/2012-05-04
http://doggie2.blog.so-net.ne.jp/2012-06-10
Which describe what I am trying to do, but there is no schematic that I can find, and google translate gives out a lot of garbage when translating that.

You kind of glossed over the power supply and dynode resistor chain, but they are probably the most fundamental aspect to getting a PMT tube operating. The gain of a PMT is highly dependent on the operating voltage and the correct resistive divider chain among the dynodes. Typical operating voltages are 1000V and up. For example, take a look at: http://www.diyphysics.com/wp-content/uploads/2013/01/Prutchi-Paint-Can-Scintillator.pdf for a nice DIY project.

[Wim_L]

It is almost impossible to detect photons due to the scope input capacitance. So preamplifier is a must.
Depends on the enviroment magnetic shielding including the photocathode insulation (to prevent leakage) can be handy.

Not necessarily needed... However, you don't want to go from anode straight into a scope input. Instead, connect the anode to coax, and go from there into a 50 ohm terminator, and attach that to the scope. That'll keep the short pulses intact. It will also attenuate them a bit. Going straight to a high impedance scope input will merely charge the scope input (and the PMT anode, which is likely to seriously reduce the PMT's performance) to some DC level.

The supply and resistor chain are critical to get good performance. Take care, though. Some PMTs will work fine at 'low' voltage (about 1 kV), others are at their best when supplied with about 3 kV or more over the divider and won't work at all with a mere 1 kV. But put 3 kV on a small PMT designed for 1 kV and that tube could suffer serious damage. Having the datasheet with the manufacturer's recommendations would help a lot here. Otherwise, perhaps try to get 100 V between each successive electrode as a starting point and work up from there.

More recommended reading:
http://www.hamamatsu.com/resources/pdf/etd/PMT_handbook_v3aE.pdf

[Dajgoro]

The datasheet is available, but my understanding of Russian is extremely limited.

http://www.gstube.com/data/2848/

[plesa]

It is almost impossible to detect photons due to the scope input capacitance. So preamplifier is a must.
Depends on the enviroment magnetic shielding including the photocathode insulation (to prevent leakage) can be handy.

Not necessarily needed... However, you don't want to go from anode straight into a scope input. Instead, connect the anode to coax, and go from there into a 50 ohm terminator, and attach that to the scope. That'll keep the short pulses intact. It will also attenuate them a bit. Going straight to a high impedance scope input will merely charge the scope input (and the PMT anode, which is likely to seriously reduce the PMT's performance) to some DC level.

The supply and resistor chain are critical to get good performance. Take care, though. Some PMTs will work fine at 'low' voltage (about 1 kV), others are at their best when supplied with about 3 kV or more over the divider and won't work at all with a mere 1 kV. But put 3 kV on a small PMT designed for 1 kV and that tube could suffer serious damage. Having the datasheet with the manufacturer's recommendations would help a lot here. Otherwise, perhaps try to get 100 V between each successive electrode as a starting point and work up from there.

More recommended reading:
http://www.hamamatsu.com/resources/pdf/etd/PMT_handbook_v3aE.pdf

I do not want  to start the flame, but what you suggested require high output current of PMT. With this it will be difficult to measure the dark counts  nor detection of sigle photons or some light pulses with e.g. neutral density filter.
Just try it ;-)

[LukeW]

This may be helpful.

https://www.flickr.com/photos/by_wally/5944662525/

https://www.flickr.com/photos/by_wally/5947391691/in/photostream/

This guy's flickr stream contains various other interesting stuff, mainly gamma spectroscopy.

The PMT needs to be completely enclosed in a light-tight dark environment with zero room light, except for your scintillation crystal or whatever the light source is that is producing a relatively tiny amount of photons. Magnetic shielding (mu metal) around the PMT may also be needed.

[Dajgoro]

Thanks for that schematic.

This is what I get on my scope:


It is just full of noise.
I can get something above that, but the amount of light must be much stronger than anything a scintillator would give.

I thinking I am going to try the other tube (feu68). But I can't figure out where pin 1 is. It has round pins with one pin being a bit displaced. On the feu31 the pin next to the displaced one was pin1.

[Wim_L]

I do not want  to start the flame, but what you suggested require high output current of PMT. With this it will be difficult to measure the dark counts  nor detection of sigle photons or some light pulses with e.g. neutral density filter.
Just try it ;-)

I assume you mean sending the output of the PMT into a 50 ohm scope input? Works fine on high gain PMTs, even for single pulses. Low gain devices (mostly not intended for photon counting anyway) may struggle to produce pulses high enough for some scopes scope to trigger on, but a high gain PMT can produce pulses of several millivolts.

http://www.thinksrs.com/downloads/PDFs/ApplicationNotes/SignalRecovery.pdf