Gamma Ray Spectroscopy - small, cheap

The goal of this project is to design a light weight gamma ray scintillator in a 1u CubeSat frame

Similar projects worth following
A group I have fun with has been launching near-space balloons for the last 25 years and is active in Amateur Radio Satellites ( In the past we have flown just about any payload we could think of and that someone could build. My last contribution was the system that did the first WiFi from near-space and, along with hot wings and libations, they helped me come up with my next project/challenge. We have had a couple of flights where we measured radiation changes with a geiger-mueller tube, but none of us had ever done any quantitative gamma measurements (just counts/min). Since I have never built anything like that it sounds fun, so this will be my imperfect documentation of how I am going about the project.

I plan to tackle this using a methodology I learned from Frank E Bird a long time ago. It always works well for me when starting anything I have never done. I will be using a small spreadsheet to break up my thoughts and process into five major steps:

1. Identify the objective and Investigate the situation.

2. Evaluate the options

3. Develop a plan

4. Implement the plan

5. Monitor, measure, and improve

If you do these steps in a loop it becomes the Demming approach of Plan, Do, Check, Adjust (PDCA). Alternately you can replace step 5 with PDCA.

My objective(s) will continue to morph over the initial bit as my starting research unfolds, but once I hit step 3 it is critical to not change the objectives. I have a huge problem with the "Babbage Syndrome' (Les Kilpatrick-1977) and it is crucial that I do a solid lock down before step 4.

  • Not completely inert, yet

    sparks.ron01/20/2019 at 04:32 0 comments

    Even though I have been focused on getting my lab back together and working on my weather station project, I have not been fully idle on this project.  The circuit I designed for the weather station anemometer "time between pulses" is a very slow speed version of what this project will use to determine the pulse width and the effective energy for any true Gaussian pulses. That is why I designed it without timing elements or a capacitor based one-shot.  For the high speed version required here there will be a lot of layout and component criticality, but the concepts will still hold.

    I also began a "dry fit" of the PMT tube and crystal into the diagonal of the cubesat frame mock-up I was given. It doesn't quite fit. That will leave me with a few choices: 1) shift to a 2u form factor, 2) look for a shorter PMT, or 3) attempt to roll the PMT leads down the sides to gain a few millimeters. I am not sure if option 3 will be enough though.  The diagonal of the cubesat seems to be long enough, but then when the diameter of the assembly is considered, it interferes.  I am also worried that if the fit is too tight there will not be any room for needed shock and vibration mounting.

    I'm looking forward to getting back on this project, but realistically it may have to wait for summer when it is too hot here to do the outside things that are currently being attended to.  It seems interesting to me that we have an opposite climate issue to those in more northern latitudes.  They do their inside projects in the winter and are outside in the summer. Here in the steamy gulf coast we do inside projects in the air conditioning during July-September and are outside the rest of the time.

  • On hold for a few more months...

    sparks.ron10/12/2018 at 20:09 0 comments

    I have had to put this project on hold for a few more months while I finish my weather station and my lab renovation projects.  In the meantime I am continuing to do research and experiment with different circuit configurations in LTspice.  My projects never really die, they just age to "vintage".

  • Hurricane Harvey

    sparks.ron09/09/2017 at 03:11 0 comments

    Life is amazingly adept at throwing "curve balls" that can bend you out of shape and ruin your plans.  It has been a lot that way here lately.  The latest massive disruptor was named Hurricane Harvey.  It revved up in only 3-4 days from a disturbance to a Cat 5 hurricane.  That is precious little time to prepare anyway.  But in this case preparations were not enough.  No forecaster had ever dreamed of an event this large, let alone one that stuck in basically the same area for 4 days.

    Thankfully I did a lot of research before building our place and it paid off.  We had no flooding around our home or shop.  The amount of water was unbelievable.  At our bunkhouse/ham radio clubhouse we have an 18ft round above ground pool.  It was empty on Friday when the storm began.  By the following Monday it was full.  That is 42 inches (1.06 m) of rain in 3 days!  Many of my family had evacuated the lower areas of Houston and came here to ride out the storm.  Most hectic and tense as they watched the news to see if their houses were still standing.

    The good news is all of my family is ok and will be able to get back on their feet quickly.  The sad news is that 50,000 homes and 1,000,000 cars were damaged or destroyed.  Only 15% of them have National Flood Insurance.  They will be suffering for years.  Please do not forget them 6 months from now.  Most people will get distracted and not remember.

    As for the project: I have been working on the design for the pulse stretcher and the pulse height digitizer.  This is the trickiest part of the design.  It is easy to stretch the pulse and then use the longer time constant to allow reasonable A to D conversion.  The hard part for me is to be able to ensure that this process does not alter the overall energy of the pulse so it can be an accurate measurement.

    Pulses with picoseconds of rise time and nanoseconds of duration are not easy to simulate or handle (I have no access to the mega-buck software needed).  That means I have to do the math as well as possible and then prototype it and work from there.  For me that is a lengthy process.

    Suggestions are welcome.

    Stay safe (particularly if you are in a hurricane area).

  • Back at it

    sparks.ron02/07/2017 at 17:31 0 comments

    The US holidays and other matters have kept me out of the lab for the last few weeks, but I think I now am able to focus a bit more time on the project.

    At Theremino's suggestion I looked in depth at their circuits and did a few simulations with their LTSpice models. I am deeply thankful for their open source attitude and hope my stuff will be helpful to others.

    For a simple gamma counter it appears that a standard Schmitt trigger approach will give the pulse detection and duration data that would be needed for that. The peak detector required for realistic energy calculations is a bit more troublesome for me. It looks like the video industry has pushed technology and costs down to a point where a high speed peak detector is feasible. If the pulses were captured/recorded in an analog form it would be possible to do this in post processing. However, to digitize it things get more complicated. A standard 44KHz audio capture would likely be too slow to prevent aliasing. The random, non-repeating nature of the pulses prevents a "stacking" approach.

    One way I have looked at, which seems overly complicated to me, would be to have a very high speed A/D converter that is triggered by the Schmitt trigger in a one-shot manner. Since the pulses are very short (nanoseconds) the A/D converter would need to be able to capture at several hundred Megahertz. Both TI and Maxim make a 250MHz sampling unit that would work for about $25. For a 200 nS pulse that would allow about 50 samples to be acquired from the pulse.

    But most of the lab devices I have read about are using some form of Pulse Height Detector (PHD) rather than sampling the entire pulse. I have tried a few designs for a PHD but my simulations are not working reliably.

  • New Crystal Arrived from Ukrane

    sparks.ron01/05/2017 at 17:24 0 comments

    Based on the discussions I decided to change my approach from using an organic scintillator and go with a crystal scintillator. It will add a bunch to the size and weight of the overall unit, but will definitely improve sensitivity and may allow me to do pulse energy measurements. Presuming I can learn enough about the mechanisms to understand how to build the detector.

    I will do the crystal unboxing and put photos (or maybe video) up in the near future.

    Unfortunately recent events have shown me how virtually every instrument or networking device in my lab is in poor condition. A lack of time over the last 10 years has encouraged my unhealthy habit of doing "quick fix" patches and not ever getting back to "doing it right". After all if it works, why fix it? Well, I am fed up with spending more time patching than on projects and am undertaking a remodel/refurbishment of the lab.

    If anyone is interested in how I do that, just let me know in the discussion and I will consider posting it as a project here. It may be useful for someone wanting to build their own lab. At the very least they could learn what not to do (ha, ha).

  • Rethinking approaches

    sparks.ron12/12/2016 at 18:11 2 comments

    From some great suggestions and discussions I found several alternative approaches to the one that was my initial focus. THAT IS FANTASTIC. This project is still in "Step 1 -- Identify and Investigate (gather data)". It means that I should not be limiting options at this point. That comes only in the next step of Evaluation.

    Two more options I began researching for this project was the possible use of SiPMT devices instead of a vacuum tube PMT. The second option was to use a different scintillation crystal.

    On the surface (I will detail more when I post the project plan spreadsheet), it seems that an SiPMT of the quality and sensitivity needed would be too expensive. It would have some fantastic benefits though. But if I cannot afford it, then it becomes "unobtainium".

    Using the BC-408 organic crystal that I have was based on an initial goal of simply having a more sensitive counter. However, based on further research and the project discussions it seems that actual gamma spectrography may be possible. As with the SiPMT, that would be based on my ability to obtain a proper NaI(T) scintillation crystal for a reasonable price. This option is looking to have potential. I have located a 40x20 crystal for a reasonable price. My actual requirement would be a 40x30, but the loss of 33% in capture volume would be made up by the increased sensitivity.

    One commenter suggested transmitting the audio from the analyzer real-time to be captured and processed at the ground station. That is an easily implemented item and would solve the problem of speed for any digital downlink.

    Great stuff! Onward, and upward.

  • I get it - Applied Quantum Physics works

    sparks.ron12/07/2016 at 16:21 0 comments

    I did a bunch on the Project Management (PM) yesterday and continued reading on the physics of the detectors. My thinking had been to much influenced by Radio Frequency (RF) in my past. What I missed until last night is that the Gamma wave is actually a high energy quanta (forgive me if I do not have the nomenclature correct). Because of that there is no loss or attenuation of the energy in them until they hit something and release their energy into photons.

    What that means to this project is that the shape and amplitude of the PMT pulse important. The thing that matters is the integral of the pulse to obtain the energy it contains. That eliminates the need for very high speed A/D. The problem then becomes one of accurate integration of the fast pulse, with all the associated issues of an integrator -- such as baseline offset, noise, accurate accounting of integrator losses, and such. There is a wealth of history on that due to work done on radar pulse reception.

    So while all the losses in the detector affect the overall sensitivity of the device, they do not directly influence the accuracy. For example in the PVT scintillator I have, the sensitivity is specified as about 25,000 photons per MeV. That is about 2,500 for a low end gamma at 100 KeV. If you lose all but 10% in your detector arrangement you will only have, maybe, 250 photons hitting the PMT cathode. However, those 250 photons will all generate the same 100 KeV each when they hit the cathode.

    So the pulse height may be very low compared to the baseline noise, but it will always have the same total energy. I am visualizing it like a mountain in the ocean (Hawaii - ahhh - oops, drifted off for a second). The total size of the mountain would be unchanged regardless of the sea level, only the amount sticking out above the water would appear to change.

    That means if we integrate the amount of pulse above the noise base for a known emitter (Cs137), it will allow you to calculate the baseline value to subtract from the pulse energy of any pulse to get the actual pulse energy. This seems to be roughly how the professional devices use the calibration sources.

    I want to thank anyone who is reading these blog entries for being patient and allowing me to "think out loud".

    Hopefully any insight as to how I work through a problem and project will be a help (it could be either as a good example or negative example - readers choice).

  • Too much fun - work to do

    sparks.ron12/06/2016 at 17:15 0 comments

    I realized I am, as usual getting distracted by the fun stuff and letting the "formal" planning slip. I know, but fun is fun. Last night I got to thinking about how the energy level of a pulse from the PMT could be measured. From the various data sources it looks like the pulses will be a few nanoseconds (up to maybe a microsecond) wide. I also saw some YouTube videos of people using DIY probes to measure calibrated sources.

    From that it looked like the highest count rate any of the sources gave was around 400,000 cpm. That would work out to a pulse every 150 uS -- IF they were all evenly spaced. But they will not be. Obviously their distribution is the definition of a random Gaussian distribution. A nice discussion of pulse adjacency is covered in the paper I mentioned in my last entry.

    But, back to the exploration at hand. How do you determine the output voltage level of each pulse? Some of the DIY systems use a fairly coarse (2 or 3 bit) discrete flash digitizer. But I wondered how much a very fast equivalent analog to digital (AD) chip would cost. Since this project is a one-off (up to a max of 10) a cost up to about US $10-15 would be acceptable.

    I was pleasantly surprised - I found a couple of them in that price range that can do up to 80 MSP at 10 bits! Most likely that will be far more than I need, but it is a fun thought. That would give me a 10 bit sample as quickly as 125 nS.

    Okay, now back to the project design spreadsheet to figure out when, or if, I need to go any further down that road.

  • A real nice engineering start

    sparks.ron12/05/2016 at 20:35 0 comments

    In digging through my old notes I found a paper that looked like it would be a big help. Unfortunately it was in Italian and I do not speak enough Italian to read something that complex.

    So I had put it aside for "future use". Last night I dug into it again and began the process of converting it to English. First I used a Linux command line program to extract the text from the posted PDF. Then I used another program to convert the PDF into an ODT (LibreOffice Text - open standard equivalent to MSWord).

    By cutting reasonable chunks of the text file and pasting them into Google Translate, I got their English equivalent. Then I pasted each in the ODT file replacing the corresponding Italian. It is amazing how far Google's machine translation has progressed. There were only a few places where it "got lost" and I had to do a manual correction. But a lot of those rough spots were technical and easy to fix, while the rest were close enough to Spanish that I could figure out what the translation should be.

    I am now the proud holder of "Tecniche di condizionamento del segnale per la Spettrometria Gamma" by Livio Cicala, but in English. The title in English becomes "Techniclal Signal Conditioning for Gamma Spectrometry". Sig. Cicala has done great work on the output from the PMT and I will have very little, if anything, to do in that section of the system.

    I recommend you check out his site:

  • Gamma Spectrometer (hopefully)

    sparks.ron12/05/2016 at 00:14 0 comments

    This all started with brainstorming about two years ago. I did a lot of the preliminary research and began to learn what a gamma ray scintillator does and how it does it. Then in early 2015 I started roughing out a design with the (later unsuccessful) goal of having something we could launch in the annual August flight. Due to not following my methodology closely I got distracted by the ground station tracking system that had been needing improvements since the WiFi project flight. Then life got in the way.

    In November 2015 the second blow in the collapse of the energy industry hit and ate up all my free time. I am now back to part-time consulting and have the time and energy to restart this project.

    In no particular order, here are the things I have done up to this writing:

    • Research the concepts of Gamma Ray Spectroscopy
    • Determine the major components used. A Photomultiplier tube (PMT) and a scintillating material (sensor) are required. The PMT will need a high voltage power supply and its output will need shaping and conditioning of the pulses.
    • Deep dive the web to see what major components cost and what is in/out of reach for this project
    • Research the things required for general High Voltage (HV) design
    • Research HV power supply noise
    • Determine what work in the Gamma Ray Spectroscopy field might have relevance and download any papers, presentations, and datasheets that are potentially relevant
    • Determine what the form factor of the enclosure will be (approximately)
    • Perform enough research to determine what type scintillating crystal would have good availability, minimal degradation for used parts, and could be purchased on a hobbyist budget. I decided on an organic material.
    • The organic that seemed the best match to the requirements up to this point was a Polyvinyltoluene (PVT) sensor. One of the industry standards is the BC-xxx series made by Saint Gobain Crystals in Ohio, USA.
    • Some of this material was available surplus on e-Bay.
    • The same process was used to decide on a Hammamatsu PMT. The R9420 had a good mix of size, sensitivity, and e-Bay availability
    • So I purchase a PMT and a scintillating crystal.
    • From the PMT and crystal choice I was able to determine the HV power supply requirements.

    The things I need to do now to get the project back on track and well defined are:

    1. Go back through all my notes and get "back up to speed" and where I was.
    2. Design the power supply and get parts. Then simulate it, tweak as required, and build a breadboard prototype.
    3. At this point I will have done the informal parts of my project steps and I need to go back and formalize them to build a proper plan of how to move forward to meet the balloon launch objective of August 2017 (along with all the needed test points and payload integration.

    This is the first time I have experimented with "going public" on a project as it is being done. The old saying is, "sausage is a tasty treat, but you don't want to have to see it being made." Because of that I have been reluctant to share things until they were fully cooked. My hope is that by showing the process, it will keep me motivated and hopefully help others.

    Onward to the edge of space!

View all 10 project logs

  • 1
    Step 1
    1. Figure out what to build
    2. Get parts
    3. Build it
    4. Test it
    5. Fly it

    These instructions are all I can figure out right now <chuckle>. As in the old cliche, the problem is in the details. <chuckle>

    I'll add more detail as I work through the project.

View all instructions

Enjoy this project?



sparks.ron wrote 12/11/2016 at 03:35 point

Thank you theremino for the information.  It seems that the most cost effective approach is to stay with the vacuum PMT but change to a NaI(TL) crystal if I can find a 38mm one for a reasonable price.

Downlinking the audio is a great idea, because we are very good with audio.  I can also record the audio onboard as a form of datalogging as well.  Since I will also inlcude a GPS to provide precise time data, I could also time and altitude stamp the audio log easily.

That would allow any post flight analysis to reconstruct the information into an altitude versus gamma type & level -- which is the primary objective of the project.  It would also allow us to see the gamma count "inversion" point which is the second key objective.

  Are you sure? yes | no

theremino wrote 12/07/2016 at 08:21 point

You can download a useful PDF file about SiPM, with wiring diagrams designed by us, here:

But beware that there are various SiPM models. The low-cost models have a square cell composed of a low number of detectors and therefore have an unacceptable noise and a very limited dynamic. There are also models with performance nearly comparable to glass photomultipliers, but they have very large silicon cells and cost many thousands of dollars.

  Are you sure? yes | no

sparks.ron wrote 12/09/2016 at 00:03 point

I did a quick scan of what is available near my budget and found I need to do a lot more research to understand what I am looking at.  Having the tiny weight and getting rid of the 1KV supply would eliminate LOTS of problems.

What would you think of something like the 6mm C or J modules here:

Would I need multiple modules?  Would they couple with my BC-408 scintillation crystal?

  Are you sure? yes | no

theremino wrote 12/09/2016 at 09:21 point

The dark count rate of this c-series is very high compared with glass photomultipliers.

A glass photomultiplier has about zero dark count rate. So the only count rate is produced by the environment background radation and the chrystal.

The SiPM c-series says: 30 to 96 kHz of count rate, while normally we work with some 100 Hz of background (about 100 to 1000 times less).

SiPM types
- The so called "SiPMs micro diodes" have low sensitivity so as to be unusable. They cost about $ 100
- The diodes from 6x6 mm (painful sensitivity) cost $ 150 to 300
- The arrays with good sensitivity costing $ 2,000 to 5,000

We have not written explicity this in the SiPM documents but the spectrometry graphs of the SiPM are very poor. Isotope rows are large and the noise floor is so high that only the main rows are visible. All the little particulars of the graphs are completely absent.

Yes the SiPM could be coupled with your BC408 crystal but there is another problem: BC408 is useful only for counting, not for spectrometry. If your intention is only to count events, then there is a more simple and cheap solution: our GeigerAdapter and a geiger tube (less then 50 dollars for the complete system, very little and more sensible respect to a cheap SiPM)

Otherwise, if you need a true spectrometry, I suggest to:

- Use a true photomultiplier coupled with a NaiTL chrystal
- Record the whole fly to an audio file (about 100 Ksps/sec or more)
- Or send continuosly the audio via radio to the base station and record it
- And finally decode the audio file, to the energy spectrum, with our application Theremino MCA

  Are you sure? yes | no

theremino wrote 12/07/2016 at 08:01 point

We have tested silicon photomultipliers (SiPM). They are very little and the cost is only approximatively double, respect to a glass Photomultiplier detector. You can find our project here:

But you have to take into account that the SiPM have lower performance than a true photomultiplier. They have a strong background noise that can not be eliminated and which prevents the measurement of the low energy gamma rays.

With a SiPM the entire gamma-ray area below 100..200 KeV it is greatly disturbed by the noise and below 50 KeV no measurement is possible. While a glass photomultiplier easily reveals the very low energy rays down to less than 2 KeV

  Are you sure? yes | no

sparks.ron wrote 12/08/2016 at 23:56 point

Very interesting!  A couple of years ago I built the silicon diode detector from Elektor ( but had problems with it self-oscillating and I never got back to debugging it.

The concept of the SiPM looks very similar to how the reversed biased diode was working in the Elector gamma counter.  The only drawback to the single diode counter was its very small capture area.  In the SiPM that problem goes away and the issue becomes one of sensitivity and dark current noise.

  Are you sure? yes | no

sparks.ron wrote 12/06/2016 at 18:12 point

Thanks for the info.  The solid state detector is now definitely in my list of options to investigate.  As to the cost side, this has to be something I will be willing to sacrifice since the probability of balloon recovery has gotten quite high but is still not zero.  My budget limit for the whole thing plus telemetry link is about $100-$150 maximum.  Our group of techno-nuts is great but every project has to be done out of your own pocket since there is no funding for the launches.

  Are you sure? yes | no

David H Haffner Sr wrote 12/05/2016 at 23:13 point

I appreciate your methodology, and projects like these are always inspiring because they are so very cool!

  Are you sure? yes | no

Gene Hacker wrote 12/05/2016 at 22:45 point

You might want to consider using a silicon photomultiplier or other solidstate means of radiation detection. They are getting cheap:

There is even a solid state gamma ray spectrometer kit available for <$600:

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates