Needing more than a spark test?

Back to the desiccant approach, my online searching has found that both molecular sieve and activated alumina media can reduce the relative humidity enough that the "dew point" is below -100F. Water freezes at 0C so any water would go down as frost, so that's why I put dew point in quotes. it ain't never going to be dew, at least not when it forms. If it warms up, _maybe_ it will become a liquid -- but the ambient RH will be so low that it would quickly evaporate. Both types of desiccant are pretty cheap.

To regenerate them, they need to be heated up to 400-480F for alumina or molecular sieve material, respectively. I'm giving the temperature in Fahrenheit simply because that's what my heat-treatment oven (AKA toaster oven) understands.

Long back when I was in high school I did a project to test the idea of using water ice as a semiconductor. I'd read an article in Scientific American that suggested it might work, using dilute acids and bases as "dopants". Based on that experience, I can say that very cold water ice is a pretty darned good insulator. Needless to say, my project was a bust.
 
I may have found an alternative to the X100 xray detector. Its specifications look like it might be a (better) drop-in replacement, although its active area is one fourth smaller -- 25mm^2. It's the Hamamatsu S5980, available from Newark for $24.80. Its capacitance/mm^2 is the same as the X100. There also is some information via Open Physics Lab that gives me some hope -- see here. The S1223 is smaller in area and its capacitance/mm^2 is higher so the S5980 should work even better. Its dark current also is pretty good, given its area.

The TIA/charge amp they used has poorer specs than the LTC6268/9 so their results, even with the S1223, could be improved.

These diodes come in metal cans, so it wouldn't be all that difficult to remove the top so lower-energy xrays aren't absorbed by the borosilicate glass window. I'd make a collet-like holder that matches the can's OD and remove it with my lathe. Easy-peasy. I've done it with much smaller fiber optic transceivers, no problem.
 
Coupla things. I found a perhaps-better equivalent to the X100, offered by Hamamatsu. The S3590. Its active area is 100mm^2. They have an app note #SD37 that describes using it, in combination with their charge amplifier, showing the spectrum from Am-241 obtained by directly converting the gamma rays to charge. The capacitance is lower, about 40pF, but with a reverse bias of 70V. The price is about the same as what the X100 was. Its dark current still is pretty high, in the nA range, which may ultimately limit its energy resolution for XRF of things like iron and its neighbors in the periodic table.

The title of the app note is "characteristics and use of charge amplifier".

The other useful item is the TIA circuit shown on the LT1028 data sheet. I substituted a TI JFE150 JFET and simulated it in LTSpice, and it performs very well. The TIA's output noise voltage is in the 100's of nV/sqrt-hz, despite using a 10Mohm feedback resistor.

The LT1028 has very low En but not-so-good In. That's where the JFET comes in, because its output R is very low. The circuit as given uses the JFET as a source follower. I'm playing around with a variation that uses the JFET as a common-source amplifier to see if the noise performance can be further improved. The problem is that such an amplifier has very poor PSRR so the noise coming out of the power supply becomes an issue. Most voltage regulators are not very good when it comes to their noise voltage specs.
 
homebrewed said:
Coupla things. I found a perhaps-better equivalent to the X100, offered by Hamamatsu. The S3590. Its active area is 100mm^2. They have an app note #SD37 that describes using it, in combination with their charge amplifier, showing the spectrum from Am-241 obtained by directly converting the gamma rays to charge. The capacitance is lower, about 40pF, but with a reverse bias of 70V. The price is about the same as what the X100 was. Its dark current still is pretty high, in the nA range, which may ultimately limit its energy resolution for XRF of things like iron and its neighbors in the periodic table.

The title of the app note is "characteristics and use of charge amplifier".

The other useful item is the TIA circuit shown on the LT1028 data sheet. I substituted a TI JFE150 JFET and simulated it in LTSpice, and it performs very well. The TIA's output noise voltage is in the 100's of nV/sqrt-hz, despite using a 10Mohm feedback resistor.

The LT1028 has very low En but not-so-good In. That's where the JFET comes in, because its output R is very low. The circuit as given uses the JFET as a source follower. I'm playing around with a variation that uses the JFET as a common-source amplifier to see if the noise performance can be further improved. The problem is that such an amplifier has very poor PSRR so the noise coming out of the power supply becomes an issue. Most voltage regulators are not very good when it comes to their noise voltage specs.
Click to expand...
TI also sells a dual JFET, the JFE2140, that could be used to improve the PSRR in a differential amplifier configuration but at the expense of adding a bit more noise. That could be offset by the diff-amp gain.
 
I came across an open-source gamma spectrometer here, if folks are interested in using it to make a SiPM-based XRF system (and perhaps are offput by the heavy bias of a roll-your-own setup that's dominated this thread).

However, the recommended detector setup is a 2 x 2 array of 6mm^2 SiPMs, which, going by current prices on DigiKey, will cost in the neighborhood of $120. Given that, to hit the claimed system price, around $200, it will be necessary to find a cheap scintillator. The best bang for the buck is a NaI(Tl) scintillator. Ebay usually has a number of this type. You want one that has NO orange coloration, that's a sign of a degraded detector. But keep in mind that OST will sell you a new 1" NaI(Tl) scintillator for $65. Sometimes ebay is NOT the best deal you can get.....

I haven't looked too closely at the recommended SiPM so can't say how challenging it might be to solder them to the carrier board. I think you'd want to be very confident in your assembly process, since you'd be installing four of them and potentially risking some significant $.

This approach uses a hardware-based peak-hold scheme so a fast ADC isn't required. It uses the ADC in a Pi Pico, so no added cost there.

If using EasyEDA's PCB partner (JLCPCB), their default shipping cost is pretty steep but you can choose a slower shipper and get your boards for less than $2 in shipping. EasyEDA is a free online PCB design application so that's zero cost.
 
  • Like
Reactions: rwm
From reading the info on Github, my impression regarding the preferred 2x2 SiPM array is to get a better match, area-wise, to the scintillator. BUT there are smaller scintillator crystals out there. For instance, the CsI(Tl) scintillator I bought off ebay is a 4x4x20mm job--so it would be a great match for ONE SiPM.....
 
The 6x6 SiPM can be found in a 4-pin SMT package, which should be much easier to solder compared to a WBGA. DigiKey has them for $33.94 apiece. One would be an excellent match for one of the 4x4x20 CsI(Tl) scintillators.

The 4x4 SiPM is only available in a WBGA and costs about as much as the 6x6. Unfortunately, this is the one I bought so I get to try my hand at soldering a WBGA....
 
I had come to the conclusion that I need more gamma ray reference sources to use for calibration purposes, but many are pretty expensive. There are a couple of exceptions to that, however. LYSO scintillation crystals are only slightly radioactive so it can take quite awhile to get a reasonable spectrum. But ebay has some vendors that sell pure Lutetium for collectors, and a gram or two isn't all that expensive. I haven't ordered any yet: but I DID order some thoriated tungsten rods (sold for TIG use). They arrived yesterday and today I had a chance to see if I got anything that looked reasonable for a spectrum. And I did:


IMG_2334.JPG


For comparison, folks can go here to see someone else's result. I'm getting all the peaks except the one labeled "x-rays", but that may be due to its low energy -- my detector's efficiency may not be all that great down that low (~60Kev). The peaks shown in my spectrum look to span 238Kev to a little under 1Mev.

A little side-note, the count rate from the box of thoriated rods is pretty high. My system was accumulating 100 counts in about a quarter of a second. That was with them right next to the detector (I subsequently moved them further away). I just left them in the box so that count rate was the gamma rays that penetrated the plastic. That MIGHT be part of the reason that the 60Kev peak is pretty weak compared to the others.

One interesting notion that occurred to me was that the higher-energy gamma rays could permit XRF analysis of higher atomic-weight elements. Or perhaps activate higher-energy xrays from our iron, nickel, cobalt etc. into a range that's more suitable for the commonly-available scintillator/PMT combinations out there. Most really aren't designed to detect low energy (~6-8Kev) gamma rays.

Due to their radioactivity, thoriated tungsten rods are falling out of favor Given the count rate I'm getting from my box of 10, that doesn't surprise me at all. There are alternatives that are less problematic in that regard.
 
  • Wow
Reactions: rwm
homebrewed said:
I had come to the conclusion that I need more gamma ray reference sources to use for calibration purposes, but many are pretty expensive. There are a couple of exceptions to that, however. LYSO scintillation crystals are only slightly radioactive so it can take quite awhile to get a reasonable spectrum. But ebay has some vendors that sell pure Lutetium for collectors, and a gram or two isn't all that expensive. I haven't ordered any yet: but I DID order some thoriated tungsten rods (sold for TIG use). They arrived yesterday and today I had a chance to see if I got anything that looked reasonable for a spectrum. And I did:


View attachment 496820


For comparison, folks can go here to see someone else's result. I'm getting all the peaks except the one labeled "x-rays", but that may be due to its low energy -- my detector's efficiency may not be all that great down that low (~60Kev). The peaks shown in my spectrum look to span 238Kev to a little under 1Mev.

A little side-note, the count rate from the box of thoriated rods is pretty high. My system was accumulating 100 counts in about a quarter of a second. That was with them right next to the detector (I subsequently moved them further away). I just left them in the box so that count rate was the gamma rays that penetrated the plastic. That MIGHT be part of the reason that the 60Kev peak is pretty weak compared to the others.

One interesting notion that occurred to me was that the higher-energy gamma rays could permit XRF analysis of higher atomic-weight elements. Or perhaps activate higher-energy xrays from our iron, nickel, cobalt etc. into a range that's more suitable for the commonly-available scintillator/PMT combinations out there. Most really aren't designed to detect low energy (~6-8Kev) gamma rays.

Due to their radioactivity, thoriated tungsten rods are falling out of favor Given the count rate I'm getting from my box of 10, that doesn't surprise me at all. There are alternatives that are less problematic in that regard.
Click to expand...
BTW, if one goes the extra step(s) of combining a SiPM and CsI(Tl) scintillator, the low-energy sensitivity can be greatly improved. The issue with NaI(Tl) based detectors is the necessity of encapsulating the crystal in a hermetic package. The opaque (external) window usually is fairly thick, which smashes the detector's low-energy sensitivity. On the other hand, a thin aluminum-foil window will keep the light out but will pass a high percentage of the low-energy xrays we're interested in. An extreme approach would be to use "silver" gilding, which in reality is very, very thin aluminum. Yes it's fragile but at least the CsI scintillator won't be damaged if the gilding is broken.

Ye Olde black electrical tape also might make for a reasonable window. And it would be less susceptible to damage compared to aluminum foil.
 
I have "first light", in terms of an XRF spectrum:

IMG_2343.JPG

It looks cleaner than previous gamma ray spectra because I'm using the background-subtraction capability I added to my Teensy code. This XRF spectrum is for Cadmium, which has three fairly closely-spaced peaks, two near 23Kev and the third near 26Kev. I see kind of a bump on the right side of the peak that might represent the 26Kev line. Or maybe not....

I had become suspicious that the method I was using to trigger the pulse acquisition (based on a rolling average scheme) was causing problems, so I changed my approach to a simpler majority one: if a majority of samples exceeded my trigger voltage, a trigger flag is set (in the ADC's interrupt service routine, ISR). That in turn initiates the pulse-processing sequence.

The problem affected low-amplitude pulses more than high-amplitude ones so that's why my older code still was able to produce the spectrum for my thoriated-tungsten sample. I'll have to try that one again to see what it looks like now.

BTW the pulse amplitude for the Cadmium peak is less than 100mV. The 60Kev photons from my Am-241 x ray sources generate ~240mV pulses (w/current gain settings). That suggests I will need to increase my system gain if I want to get the XRF spectrum for iron. If my detector can go that low. My first attempt at Fe with my latest S/W wasn't encouraging but I can increase the gain a bit more without much additional effort.

BTW, I really wasn't expecting my particular PMT/Scintillator combo to be able to "get" the iron peak, most garden-variety NaI(Tl) scintillator capsules aren't spec'd for energies that low. It takes a thin window, usually beryllium, to get that.
 
You must log in or register to reply here.
Back
Top