Measuring low light levels

[daveshah]

I want to have an array of photodiodes measuring the light output of a scintillator - which will probably be very low, around 420-430nm, but am slightly confused about what to do. I understand that I need some kind of high gain, possibly transimpendance, amplifier, however, I am unsure about multiplexing the signals. Can I put the photodiode signals through an analog multiplexer before the amplifier, or does there have to be an individual amplifier for each photodiode then the outputs from the amplifiers are multiplexed (or is there an even better way of multiplexing the photodiodes)

Thanks

[mikeselectricstuff]

You'll probably need some special UV sensitive PDs to have response down to 400-ish nm. As currents will be very low, you'd probably need to put the TI amp very close to each and mux the output. Anotehr issue is at the very low currents you'll have response time issues if you try muxing the input.

Hamamatsu are specialists in this area, both PDs and photomultipliers - may be worth looking to see if they have any appnotes

[w2aew]

If you intend to simply sum the outputs of the photodiodes, you can simply parallel them into the input of a transimpedance amplifier.  Each PD will look like a light dependent current source, and they'll sum nicely by wiring them all to the input of a single TIA.

[TerraHertz]

Interesting question. Are you trying to retain pulse amplitude information for individual particle interactions with the scintillator, or just accumulate an average?

I'll be interested to see what people suggest too. I haven't tried diode photodetectors with scintillators at all. But I do see quite a few diode detectors intended for direct particle detection come up on ebay.

[daveshah]

If you intend to simply sum the outputs of the photodiodes, you can simply parallel them into the input of a transimpedance amplifier.  Each PD will look like a light dependent current source, and they'll sum nicely by wiring them all to the input of a single TIA.

Ideally I want individual data from each of the sensors, so this won't work

Interesting question. Are you trying to retain pulse amplitude information for individual particle interactions with the scintillator, or just accumulate an average?

Given the power, cost, and weight restrictions for the project (it will be travelling to 30km on a weather balloon), I have a feeling that obtaining information for each particle would be difficult. Therefor I am currently trying to take an average, but very quickly (hopefully up to ~1000-10000 times a second) - would a separate lowpass filter be necessary, or would the GBW of the amplifier and stray capacitances be enough?

[TerraHertz]

Given the power, cost, and weight restrictions for the project (it will be travelling to 30km on a weather balloon), I have a feeling that obtaining information for each particle would be difficult. Therefor I am currently trying to take an average, but very quickly (hopefully up to ~1000-10000 times a second) - would a separate lowpass filter be necessary, or would the GBW of the amplifier and stray capacitances be enough?

Ah! I can't answer, as I'm just learning about this stuff myself. But have you seen this forum:
  http://www.coultersmithing.com
  http://www.coultersmithing.com/forums/

There's a lot of information there about all kinds of particle detector operation.

Also see: http://www.ebay.com/itm/160932531827?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1438.l2649

The device is too heavy for a balloon, but the _principle_ is interesting. He's using a USB audio codec to capture the scintillator output.
See http://www.physics.usyd.edu.au/~marek/pra/ for a freeware data capture & analysis utility. The guy that wrote that is at Sydney University.

If you can build a lightweight scintillator, add a miniature flash-PC and you probably _could_ do a complete balloon-carried gamma spectrometer logging system.

[mikeselectricstuff]

Avalanche photodiodes may be an option of you need low weight, they do appear to be rather expensive  - i wonder if any normal PDs are useable in avalanche mode?
This is the cheapest one I could immediately find for the UV region :
http://uk.mouser.com/ProductDetail/Pacific-Silicon-Sensor/AD800-8-S1/?qs=sGAEpiMZZMtWNtIk7yMEscpTFN8GTr6j%2f4Ht%252blUzzz8%3d

Digikey have some but hundreds of $

Hamamatsu make them but can't find any pricing online.
there are a couple on ebay.
http://www.ebay.com/itm/Photo-Detector-Diode-Silicon-Avalanche-Photo-Diode-SiAPD-/200859585312?pt=LH_DefaultDomain_0&hash=item2ec42a0b20

[daveshah]

 Ideally we want an 3x3 or 4x4 array of sensors, so I think avalanche photodiodes would be beyond our budget. Assuming I am careful with amplifier positioning (at the moment I am planning to put the TIA directly behind the photodiodes), do you think normal photodiodes would work in their standard configuration?

[plesa]

Did you consider usage of MCP? They are not cheap but are tiny and the TIA will be much more simple.

[Jeff1946]

I assume what you are detecting is pulses of light, if they are fairly short then you may want to couple the output of the diode via a suitable capacitor to a fast amplifier, which drives a fast comparator, which triggers a counter.  Fairly standard pulse counting stuff.

[mikeselectricstuff]

Ideally we want an 3x3 or 4x4 array of sensors, so I think avalanche photodiodes would be beyond our budget. Assuming I am careful with amplifier positioning (at the moment I am planning to put the TIA directly behind the photodiodes), do you think normal photodiodes would work in their standard configuration?

For UV detection I don't think there is any cheap solution. All the ~400nm PDs on Farnell are GBP10 and up, so APDs may not be a bad option sensitivity-wise.
I wouldn't rule out small photomultipliers if you can find some cheap on ebay.

[daveshah]

I have decided on APDs, at the moment I am waiting to hear the budget so I can decide how many to get (either a single sensor, a 2x2 array or a 3x2 array)

[daveshah]

Although it's been many years since I made this thread; I thought it might be interesting to post an update about what happened (I was reminded about this when someone posted to another old thread about a project of mine.)

Back in 2013 I couldn't persuade anyone else on the project to spend a serious amount of cash buying decent photon counting type detectors; so after messing about a bit longer with DSLRs (wholly unsuccessfully) we decided to fly a completely different project on that balloon.

Last year another HAB opportunity came up; with a bit more weight budget. The first plan was to fly one of Elektor's simple photodiode detectors (which use direct interaction between particles and the photodiode) to measure radiation. However we still wanted to use the scintillator for a larger area and hopefully more sensitive detector, ideally with some kind of direction sensing.

Advansid kindly provided us with a 4x4 SiPM array (with one defective channel so couldn't be sold, but not an issue for our purposes) free of charge. We developed a fairly small and simple system using this and the BC408 scintillator to detect radiation. The first version of this used 16 fast comparators (on the output of an amplifier connected to each SiPM channel) and an CPLD to measure pulse width as this seemed like the cheapest way of measuring the "amplitude" of 16 channels without needing any kind of peak detect or fast ADC.

This setup was first tested at school with a gamma source to ensure it detected something (which it did), and then flown on a HAB last October. Although results looked promising at first; a partial failure in lightproofing at some point during the flight limited the usefulness of the results.

The downside of pulse length method is it struggled to distinguish lots of concatenated small pulses from random photons (due to ambient light) from a large pulse due to a large number of coincident photons. Nonetheless some very long pulses were observed that could not be replicated simply due to the lack of lightproofing; and starting before the lightproofing failure likely occurred. There also seemed to be some as of yet unexplained difference in the behaviour of different channels; though this may simply be down to the way the scintillator was mounted.

Subsequently we built a new board that uses ADCs (but only on 4 channels) instead of the pulse with method to analyse the pulse amplitude. This has not yet been flown on a HAB; but hopefully will be one day as it will hopefully record more useful data.

I've been meaning to produce a proper report about this project with design details and am currently about half way; and I'll post a link here once that's closer to complete.

[kony]

Was that intended as gamma spectrometer, or just radiation detector with very crude binning of energies?

[daveshah]

The first attempt was a crude binning of energies and also some attempt at obtaining directional information (the latter was not successful due to the differences between channels). With the ADC board and covering the scintillator in reflective material hopefully it will be possible to get slightly more defined binning although I haven't done any proper tests yet.

[daveshah]

I've decided to put my write-up so far online (not finished or checked) which describes the SiPM circuitry and the results I have obtained.

http://ds0.me/sipm/index.html

[Kleinstein]

A peak detector circuit is not that complicated - usually just 2 OPs, a FET, a few diodes and some RC and than going to a µC internal ADC. So it might be really worth getting peak hight measurement.

[daveshah]

A peak detector circuit is not that complicated - usually just 2 OPs, a FET, a few diodes and some RC and than going to a µC internal ADC. So it might be really worth getting peak hight measurement.

Yes - that's something I might look at. The problem previously was 16 fast peak detectors looked surprisingly big and expensive; although I might look into it again. Of course it's possible to use fewer channels, however I already have a 4 channel 65MSPS ADC board that can do digital peak detection so any analog peak detection solution would need to be better than this.