homebrew multi-channel analyzer

This is bound to get mixed up with things in Electronics, check both. Physics-specific stuff here, mostly.

Re: homebrew multi-channel analyzer

Postby Philipp Windischhofer » Wed Sep 24, 2014 3:40 pm

Here's the software the version I sent Doug runs with. Created using MPLAB-X 2.05 and xc16-gcc 1.21.
Download to the pic can be done with any compatible dongle, I used an older "PicKit 2".
Attachments
Sctintillator_Doug.hex.zip
ready-to-use hexfile
(4.59 KiB) Downloaded 398 times
Sctintillator_Doug.zip
software for MCA; uses serial interface for commands / data
(23.35 KiB) Downloaded 402 times
Philipp Windischhofer
 
Posts: 21
Joined: Wed Aug 13, 2014 2:12 pm

Re: homebrew multi-channel analyzer

Postby Doug Coulter » Fri Oct 03, 2014 8:51 pm

Here are some simple software tools I'm using. Gmail wouldn't let me send this straight to Phillip, nor would a pm here. Ignore if you're not a software type.

Tools.zip
(685 Bytes) Downloaded 404 times
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: homebrew multi-channel analyzer

Postby Philipp Windischhofer » Sun Oct 26, 2014 3:03 am

Doug sent me (thanks again!) some Na:I heads among other stuff, and I finally got around to try the smaller one of these. (I still need to put a BNC plug on the bigger one before I can wire it into my MCA.)
I didn't change any settings on the MCA, just replaced the BC-412 by the Na:I and wondered how it might turn out. Since my design uses even slower opamps (OPA-111) than the one I sent to Doug, I'm using excessively high series resistance for the phototube (either 100k or 1M) to give me slower peaks. Of course this makes it prone to problems with pulse pileup but for not-too-hot sources it seems to work well.

Here's a spectrum of Doug's Cs-calibration source (spark gap tube) with live times of 5 minutes and 3 hours, respectively.

cs5min.png
first spectra taken with the new head, 5minutes live time

cs3h_1.png
same source, but 3 hours sample time


Note that the scaling is a little different for both graphs -- that was because I used a somewhat higher bias voltage (~50V more) for the 3h-spectrum.

Doug, I couldn't get that misbehaviour of my board doing these weird fluctuations between adjacent bins reproduced here so far (although I didn't try with the big head). Probably peak length plays an important role (considering the not-quite-highspeed components on the board), and since I stretch my pulses that much, I don't run into those problems?

Philipp
Philipp Windischhofer
 
Posts: 21
Joined: Wed Aug 13, 2014 2:12 pm

Re: homebrew multi-channel analyzer

Postby Doug Coulter » Sun Oct 26, 2014 4:31 pm

I'm not sure what's going on there...noise of that amount usually isn't an issue here, so ???. Try with the same sensor I used. I see here that you've set the threshold high enough to not see the low energy line of Cs137. I didn't toss the boards out, if we learn something, that's good. I did add some pulse stretching in the form of 22 pf across the 220k feedback resistor on your opamp. Sure looks like a lot of compton scattering, more than I had here on the bigger sensor, which is to be expected. To avoid some of that, the big boys will sometimes put detectors outside the NaI and put in anti-coincidence logic so that anything scattering out of the side of the NaI means "no count". But in practice that's really hard to do - even lead may as well be transparent at the higher gamma energies, so how to keep the incoming beam collimated enough?

Still working on my version. Fixed a bone-head layout problem with switched gate drivees to the fet, and a latchup in the comparator, but now I have something weird going on with the chipkit a/d shorting the peak detector to ground all by itself at odd times - something screwy in their adaptation of the arduino stuff I'd guess, since I've found other mistakes in their port (and fixed them) already. It can be a real challenge to go through tons of files that include files that get #defines like magic from ? and do different code and constants, depending. I may just go to MPLAB and their compiler to eliminate the uncertainty. At this point, the amount of work seems similar. Though fixing chipkit's errors would do more for the community at large...
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: homebrew multi-channel analyzer

Postby johnf » Tue Oct 28, 2014 2:29 am

Philipp / Doug

philipp
try putting up the volts on the PMT

We do something called Bias scan on our detectors to find the sweet spot
ie low noise and not taking off. noise tends to be a problem with low pmt volts
your volts seem a bit low for the normal pmt scintillator setup we use "low 800 medium 950 high 1100 volts" on most of our pmts.

They plateau out with rising volts so your check source is ideal.
keep the pmt volts just below the knee of the curve
johnf
 
Posts: 433
Joined: Sun Aug 08, 2010 3:51 pm
Location: Wellington New Zealand

Re: homebrew multi-channel analyzer

Postby Doug Coulter » Tue Oct 28, 2014 3:05 pm

While I've heard otherwise from some in the biz, I have to agree with John here - it's always worked for me. The phototube gain vs voltage is a really steep curve, and getting closer to where the voltage makes less difference (bonus, more signal) seems wise, and has worked out that way here many times. Your main worry is overheating the resistive divider chain if you go too far. We don't know how many stages are in those bicrons, but usually ~>~120v/stage is good. The reason the curve rounds over at high voltage is that the electron multiplication of whatever dynode coating they use has a maximum-limit vs electron velocity. I run my 12 stage (going by resistor count) "gallon jug" NaIs at 1837v here, which is what gave the best spectra on an expensive borrowed MCA. I have re-wired it so that the anode "floats" and is a ground level, providing the HV as negative polarity DC to the photocathode. That works out to around 153v/stage, which made Paul at URSA wince, but it's what worked the best in real life.

Shown here is also the voltage doubled 1kv CCFL inverter I use to drive this - it's in its own box that's kept fairly far away for noise reasons.
BigJug.JPG
The big jug...we're lucky, we have two of these!


And here's the actual wiring in there, with the lid taken off so you can see it:
JugWiring.JPG
Phototube wiring


This lashup, with a good MCA, gave better spectra of CS137 than the example on wikipedia, but not by a whole lot. There are limitations in every scintillator to resolution, intrinsic and constructional - how many times does a photon have to bounce around before it hits the cathode? This is why my tiny NaI, only a couple mm thick, works better at the low energy end of things.
In this pic, blue is the negative HV, black is ground, and the orange is the output signal (goes to a bnc).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: homebrew multi-channel analyzer

Postby Philipp Windischhofer » Sat Dec 06, 2014 3:10 pm

I finally got a chance to do some cross-checking between the two heads Doug sent me. I took spectra of the Cs-137 check source and small radium watch hands.

Here's the raw data (only first and last channel removed from displayed dataset):

radium_small.jpg
Radium with small detector, 1hr live time

radium_big.jpg
Radium with big detector, 1hr live time

cs_small.jpg
Cs-137 with small head, (about) 20 minutes live time, (much) more noise

cs_big.jpg
Cs-137 with big head, 1hr live time


Two things are noteworthy: of course, the big head is much more sensitive, hence the improved SNR compared to the small detector. On the other hand, the peaks on the spectra taken with the big detector seem to come out not nearly as clear as on the corresponding shots with the small one. Any hints to what the cause could be?

Philipp

EDIT: Doug, I just looked again at the noise characteristics of the spectra you took with my board and the big probe (added below for sake of completeness).
Screenshot-1.png
Doug's results using my board and the big Na:I head

Judging just "by the looks of it", the noise in your spectra (at least to me) looks much more artificial than the bin noise on the above Cs-137 spectrum. What do you think?
Philipp Windischhofer
 
Posts: 21
Joined: Wed Aug 13, 2014 2:12 pm

Re: homebrew multi-channel analyzer

Postby Doug Coulter » Mon Dec 08, 2014 2:46 pm

Well, Philipp, I'd have to do some guessing, which is a domain I'd rather not be in.
I did not change the trimpot or try different time constants on the AC coupling, either of which could account for the differences I see here. I fear that the larger of the two detectors has a flaw of some sort - I did see some junk on its output waveform here. I really doubt any excessive noise was the issue, at least from what I saw on the scope. You did seem to run longer than I did - which averages things better, of course. I was running this over a grounded metal plate here, and in a place that is low-noise to begin with.

Your scheme of waiting for a signal to drop entirely below threshold means variable droop in your S/H if the time varies or the threshold, and that alone would account for (most of) the differences we're seeing here.

I see WAY too much compton scattering in these plots (both of ours) which John Futter tells me is mainly due to how small both of the detector crystals are - they are from PET scanners and were only used for detecting coincidences above some threshold. Indeed, in tests here with those and a "big jug", the big jug shows a far better spectra, with very low compton "noise" between the low energy gamma and high one from Cs-137, quite a few dB down there, if you will. Hard to beat a huge chunk of detector - if a gamma hits the middle, it gets stopped by more of the xtal before it can scatter out. It could be you don't have a "good geometry" experiment there - other nearby things scattering gammas back into the detector, or something between the source (like air) and the detector. It's hard to get right.

Due to one of my scopes getting "built in" to the fusor setup, and another in a place where it's hard to work on this, I bought another and am reviving my own little effort at this. I hope to have some worthwhile results, better or worse, soon. I had some issues with my comparator reversing sense at its common mode input range, since fixed, and a couple of others. And now I see that something in the arduino-uno (the chipkit variety, not atmel) shorting my "hold" periodically, which could be either a very short glitch out of the comparator, or more likely, something odd in chipkit's adaptation of the arduino code and IDE for their own use - I'm currently suspecting the latter, since it does work quite well on a fixed level or low impedance signal generator.

But there I'm just guessing. I will know soon, and report.

I did find with a replacement version of the larger bichron that you can pry off the back fairly easily (it's merely tight-fit and weak silicone). Some of the noise I saw in it's output (the one I sent you) could be explained by a bad resistor or connection in the phototube divider chain - it would be hard to attribute it to the xtal itself, which BTW, is possible to reuse if you take the old phototube off it and supply something else (as I also sent you). It's a pretty major piece of detailed surgery, though.

One I have here I simply cut right behind the crystal on a metal band saw and then popped off the phototube (crazy glue or similar was used, very brittle) and put on another, and then made a light tight interface again. That's kind of brutal. If you could get that tube loose without cutting the main box - and put in another - that would be a much nicer way of doing things. As far as I can tell, there's no reasonably safe way to just pull the crystal out the front. 5 sides are metal encased and 4 are securely glued to the outer envelope. The phototube end has a glass window that is epoxied onto the otherwise metal case to be hermetic (NaI doesn't ever want to see humidity, it degrades very quickly). So hacking off the entire thing was actually the safer way - no risk of breaking the seal, but you definitely ruin the rest of the box doing that.

You might try those BGO xtals, end on, on that dual phototube I sent as well. Though it's not as good as NaI in theory in resolution, it's not bad either - the pulse is faster though - and due to its very high density (and high Z of most of the chemistry), has less compton scattering vs size than NaI does.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: homebrew multi-channel analyzer

Postby Philipp Windischhofer » Mon Dec 15, 2014 11:53 am

Hi,

the thing is - apart from getting rid of the preamp, I did not make any changes to the circuit. Given the intrinsic problems of the design, I'd say it works remarkable well (for my practical requirements, I can do away with the noise by just letting it run longer).

Here's another one I did the other day. This is a small chunk of pitchblende (the smallest I had, but still hotter by far compared to the Cs-137):
pitchblende.png
small piece of pitchblende from Germany. About 1hr live time, few outliers


Yes, I will most definitely try that BGO crystals of yours on that dual phototube! I only have to throw together a new power supply (negative volts, and more power, my current supply is limited to about +800V) and find some truly light-absorbing material to house the crystal.

Looking forward to more results from your design!
Philipp
Philipp Windischhofer
 
Posts: 21
Joined: Wed Aug 13, 2014 2:12 pm

Re: homebrew multi-channel analyzer

Postby Doug Coulter » Tue Dec 16, 2014 6:42 pm

Well, I wish they were better than they are. To make it work close to right, I have to current compensate the NPN, and that causes sag between the time it reports a peak and the a/d in the pic finally gets to it (interrupt latency etc).
I see a pattern in the data that seems to point to nonlinearity in the pic a/d...hmmm - every 4th bin seems to be very hard to "hit". This might be a pic hardware issue, dunno. I'm going to have to do a couple things differently it appears - move one of the fet "shorting out" elements to the opamp output, rather than the between transistors spot, so the current comp on the NPN (right now, a big R and a -9v battery) won't cause a sag during that time. The pic doesn't seem to always take the same time to respond to an interrupt, which could be because I'm running other junk in the background to flash a led with delay()...and other interrupts (the timer). This is a place where the "simpler" arduino setup is not in your favor, as that stuff is kinda hard to find and turn off so it doesn't interfere with hard real-time things. I might have to go with the "real" mplab IDE, or make the circuit perfect so it doesn't care, one or the other. Or both.

Right now, I'm working with a very nice "Exploranium" GR-130 Bill just got us, only 255 bin resolution, and writing code to grab spectra from it and put them in a database and plot them. When I finish that - which is adequate for my intended use anyway, I'll get back on this one since it will have 4x the resolution and that will eventually matter with better detector heads. For what's in the GR-130, it's not an issue - it's about the same as the bigger NaI I sent you, only laid out a little better as regards light efficiency & compton scattering, but the same weight of NaI as far as I can tell.

Here's some plots of it:
firstlight.png
Cs137

Thorium5min.png
Thorium

uranium.png
Uranium (recently purified, natural, no daughters)

Torbermite.png
Torbernite ore from African natural fission reactor, very old, full of Ra too.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Previous

Return to Metrology

Who is online

Users browsing this forum: No registered users and 2 guests