Need a few recommendations for a gamma spectrometer

[Highvoltagefox]

HI all,

I am a nuclear engineering student trying to make a handheld gamma spectrometer. I am not a complete novice in electronics but I have never designed anything myself mostly just improvising off of others schematics, but I wanted a project to challenge myself and grow in understanding. So I have a few questions I hope you can answer.

First off, instead of a photomultiplier and a large NaI crystal I have opted for a small LYSO and a PIN photodiode recovered from a PET scanner. The diode I have is NOS stock but doesn't have any data sheet. From what I found it seems like single custom line from advanced photonic systems made specifically for siemens scanners. I have tried to get a response from the company that made it but haven't heard back. So how would I go about driving  and testing a photodiode I know nothing about? I'm just worried that i couldo destroy it. If anyone knows a good resource I might be able to find regarding these PIN photodiodes or has an idea on how to non destructively determine operating voltage I would be ever grateful.

Second I am looking for a suitable fpga to use as a fast data collector for a raspberry pi to then interprete but cost is a bit of an issue. I want to get a spartan 6 because of it features and being still manufactured I can get used to it and use it in later developments. But it's expensive for something to practice on ando I'm am confident that most of its abilities and power consumption will be wasted on this. So what is a decent highish speed fpga that would be suitable for this project? Also I plan on using matlab to program the fpga since I am very familiar with programming in it. Has anyone used matlab to program fpgas and has some advice?

Thanks for reading, I appreciate any advice 

[KE5FX]

Especially if power consumption is important, you should use an Artix FPGA, or wait for Spartan-7 to show up.  Almost nobody should be using Spartan-6 for new designs at this point.

One photodiode is probably more or less like another, in that you'll end up feeding it to a relatively simple transimpedance amp.  There are subtleties involved -- not the kind that will blow out your diode, but the kind that may keep you from getting the best performance possible.  If your book budget can tolerate the hit, I'd strongly recommend this one. 

[Kleinstein]

If the diode is just a PIN diode, you don't need that many data from the DS. They are all relatively similar - just find a suitable bias voltage as a good compromise of extra noise and speed. It gets way more complicated if the diode a an avalange type diode - so with internal amplification. Anyway the performance of diodes is often not that good compared to PMTs.

For the way of look at the pulses, there are mainly two options: one is using an analog peak detector and than a rather slow ADC to measure peak hight. Here no FPGA is needed - just a small µC with it's internal ADC is good enough. I have build such a system based on an Atmel tiny26 (with classical NaI(Th) and PMT) quite some time ago. It works rather good despite a NaI crystal with a crack.

The other more advanced method is using a really fast ADC to sample the pulse curve and do pulse detection / area calculation in software / digital. Ideally one would want something like 10s of MSPS and sufficient precessing power. For a start one can get away with a ARM type µc with a relatively fast internal ADC (e.g. 5 MSPS). Unless the data rate will be very high, there should be plenty of time for the µC to process the data - ony storage could be an issue. The task is more something suitable for an µC than an FPGA, unless you have very high levels.