Needing more than a spark test?

[graham-xrf]

@WobblyHand I have a Pi4 being my mailserver (3 years now) and another on the network, handy to mess with when I want to wiggle SDi and suchlike. It has the overkill IceTower silent fan on a bonkers set of fins.

My son has invoked two Raspberry Pi 5(s) to control a industrial grade 2-Canon photogrammetry image capture. For a tiny little computer, they are awesome! One would have been enough, but the Cannon (relatively crappy) software would only allow a single access at a time to the software library it uses. Given the bunch of other threads they were also running, it was easier to "just get another P5", instead of writing a work-around. Then, by serendipity, the extra Pi helps also share the fast networking load. The Pis only have to work the user touch screen interface, take the pictures, and shift the images into the main computing kit via network, but I was surprised at just how many situations one has to program for. Gestures? Did he touch screen in two places? etc.

I do now also have the Raspberry Pi Zero 2W, the intended little computer for the XRF, but Teesy, and others can also do it. In theory, even a micro-controller like Pico can do the A/D, but I wanted to put all the processing right up to getting the plots, on the main PCB, just so the display to external screens becomes very flexible, and without needing a bunch of software for many platforms. Anything that has a browser, or can access a USB or WiFi would do.

A 79c part sounds good, but for me, I usually use Mouser, where I need to rack up £30 worth of parts to get the free shipping. I think it be much the same situation from Digi-Key. When next I get (probably another ADC to replace a fried one), I will include some crystal oscillators. Now that I have tried the "hot air from the top trick", maybe a BGA mounting is not so scary. I always think the little things will blow away. I have not yet invested in the "modified temperature controlled oven".

Getting some of the stuff from Mouser mentioned by @homebrewed may be on, but I think 3 x 3mm significantly reduces the captured photons. A CsI(TI) scintillator crystal for $30 bucks, with a large area photodiode may be a better idea. Not needing the high voltages, and having a photodiode that need only be sensitive to to visible light? Maybe a re-purposed cheapo webcam imager? Would the scintillator I already have work with a Si photodiode in place of the high voltage tube? Perhaps, but we should stay with available stuff as Mark suggested.

I don't know - maybe I am grasping at straws when it comes to alternative sensors, but I will try anything that has a chance provided you folk have given it a think-through and pointed out the flaws. Whatever we use, the light is faint. We will always need a load of low noise gain.

 

[homebrewed]

Would the scintillator I already have work with a Si photodiode in place of the high voltage tube?

It appears so, that's the combination device I found on Mouser. It may or may not use the same PIN diode that was in the X100-7.

If you want to stick with PIN diodes as the photodetector, the best bang for the buck looks like some offerings from Osram, but they're on the small size, comparatively speaking -- about 8mm^2. Their capacitance and dark current are much lower than the X100, but that's just due to the area difference. This scheme is most compatible with the development work to date, and they sure are a lot cheaper than the commercial PIN diode+scintillator offerings.

I'm not dead set on using my PMT for the "final solution", at this point I want a known something-that-works reference to test my code with.

Since the light pulses coming out the scintillator will be very short, I think it would be best to stick with PIN or avalanche (A.K.A. SiPM) diodes. From there, it depends on the choice of going with the Theremino scheme that permits the use of a slower ADC: or trying to capture the pulse as close as it is to its raw form as possible and perform post-processing on it.

 

[WobblyHand]

FWIW, at least for me, in the US, I have found Digikey to be less than Mouser. That being said, sometimes paying for postage helps one get working sooner, rather than later. But, yes, I squeak whenever I have to do that. Better to wait up for a bigger buy most of the time.

 

[homebrewed]

FWIW, at least for me, in the US, I have found Digikey to be less than Mouser. That being said, sometimes paying for postage helps one get working sooner, rather than later. But, yes, I squeak whenever I have to do that. Better to wait up for a bigger buy most of the time.

I usually go with Digikey, but sometimes Mouser has something that Digikey doesn't. The X100 parts were like that. They also stock some low-noise JFETs that Digikey doesn't carry.

 

[WobblyHand]

I usually go with Digikey, but sometimes Mouser has something that Digikey doesn't. The X100 parts were like that. They also stock some low-noise JFETs that Digikey doesn't carry.

Both are decent suppliers, but when I was doing my ELS work, and sourcing amplifiers, I found Mouser to be consistently more expensive by 5-10%. That may be different now, I haven't checked lately.

 

[graham-xrf]

It appears so, that's the combination device I found on Mouser. It may or may not use the same PIN diode that was in the X100-7.

If you want to stick with PIN diodes as the photodetector, the best bang for the buck looks like some offerings from Osram, but they're on the small size, comparatively speaking -- about 8mm^2. Their capacitance and dark current are much lower than the X100, but that's just due to the area difference. This scheme is most compatible with the development work to date, and they sure are a lot cheaper than the commercial PIN diode+scintillator offerings.

I'm not dead set on using my PMT for the "final solution", at this point I want a known something-that-works reference to test my code with.

Since the light pulses coming out the scintillator will be very short, I think it would be best to stick with PIN or avalanche (A.K.A. SiPM) diodes. From there, it depends on the choice of going with the Theremino scheme that permits the use of a slower ADC: or trying to capture the pulse as close as it is to its raw form as possible and perform post-processing on it.

We were trying to exploit the X100 bandwidth sensitivity (if one can call it that) response to incoming X-rays. The thing about scintillators is they do the conversion to visible light, which we might get at with much lower cost photodiodes. I guess the pulse might be a bit "second hand", but it may be worth a try. How sensitive are webcams? Duh - the mind should not do this to me!

What I liked about the PIN diode was the very direct capture of photons into becoming semiconductor carriers, but if decent area PIN X-ray pin diodes are becoming too exotic, then let us try something else. In passing, the scintillator crystal Mark found on eBay is only in the USA. It's asking price competes with the £50 (about $60) to get it posted. There is a picture of some dude with whole handfuls of them!

I am not going to get too concerned right now about diodes and scintillators. There is too much else to get in deep with.

 

[homebrewed]

In passing, the scintillator crystal Mark found on eBay is only in the USA. It's asking price competes with the £50 (about $60) to get it posted. There is a picture of some dude with whole handfuls of them!

Our European friends may have a more difficult time getting these cheap(er) scintillators. I have found an online supplier that probably is OK with selling small quantities of scintillators, and they have a number of relatively inexpensive offerings in addition to NaI(Tl) -- like CsI(Tl) and CdW04. The latter is completely nonhygroscopic and is reported to have energy resolution that is comparable to NaI(Tl). Since it contains tungsten, and we are only interested in xray energies in the 6-8Kev range, the .5mm thick scintillator will work OK.

Here is a link to that supplier. They may sell to European customers: or have a European distributor.

 

[graham-xrf]

Our European friends may have a more difficult time getting these cheap(er) scintillators. I have found an online supplier that probably is OK with selling small quantities of scintillators, and they have a number of relatively inexpensive offerings in addition to NaI(Tl) -- like CsI(Tl) and CdW04. The latter is completely nonhygroscopic and is reported to have energy resolution that is comparable to NaI(Tl). Since it contains tungsten, and we are only interested in xray energies in the 6-8Kev range, the .5mm thick scintillator will work OK.

Here is a link to that supplier. They may sell to European customers: or have a European distributor.

Thanks for the link. As I look at it, I remember I have come across the supplier sometime long ago. At least with the ones you mention, there is a reasonable area for pickup. I may be missing something you might have figured out long ago, but why only 6-8KeV range?

K-shell responses from alloying elements are admittedly more limiting, because interesting stuff like silicon and sulfur are below 3KeV and stuff like Tungsten and beyond (59.3KeV) are out of (K) range, but Titanium, Vanadium, Chromium, and Manganese have K-Shell responses around 4.5KeV to 5.8KeV.

L-Shell responses give us a look at even Lead around 10KeV to 13KeV. There is an abundance of L-shell glows outside the range 6-8KeV for all sorts or rare exotic stuff we would not likely be looking for in alloys, but there are some that are useful. There is Molybdenum, Silver, Cadmium, and Tin confirmation around 2.3KeV to 4Kev. We even get to look at Tungsten with two glows at 8.4KeV and 9.7KeV.

Is it reasonable to think that L-shell glows are just more probable?

 

[graham-xrf]

Logic ICs ?
Yes - this happens when you need to scatter some glue logic between the computing stuff and measurement bits.
In the lower left corner of the A/D circuit of AD7622, there is a D-Type Latch which lets in the low jitter clock.
There is a "Note 7", which did not help much.
It's been a very long time since I used something like 1/8 of a 74273.
3.3V has arrived (and almost moved on!)

So what is the modern choice? Maybe MC74AC273 costing £0.94, or something with a "T" in the part number?
Its a 20-pin SOIC. That's enough space to fit two whole A/D converters!
One wonders what the remaining 7 of those D-Latches might be useful for.

 

[homebrewed]

Thanks for the link. As I look at it, I remember I have come across the supplier sometime long ago. At least with the ones you mention, there is a reasonable area for pickup. I may be missing something you might have figured out long ago, but why only 6-8KeV range?

K-shell responses from alloying elements are admittedly more limiting, because interesting stuff like silicon and sulfur are below 3KeV and stuff like Tungsten and beyond (59.3KeV) are out of (K) range, but Titanium, Vanadium, Chromium, and Manganese have K-Shell responses around 4.5KeV to 5.8KeV.

L-Shell responses give us a look at even Lead around 10KeV to 13KeV. There is an abundance of L-shell glows outside the range 6-8KeV for all sorts or rare exotic stuff we would not likely be looking for in alloys, but there are some that are useful. There is Molybdenum, Silver, Cadmium, and Tin confirmation around 2.3KeV to 4Kev. We even get to look at Tungsten with two glows at 8.4KeV and 9.7KeV.

Is it reasonable to think that L-shell glows are just more probable?

The 6-8Kev range is where the ferrous elements' K-alpha xray lines are. I recall that was the energy region where we looked for ferrous contamination in IC's, but that could have been due to limitations of the detector we had at the time. We generated the xrays with 40-50 KV electrons (in a SEM). If need be we could go up to about 70Kev, but spatial resolution suffered. Since we were interested in localizing the contaminant(s) we tried to use as low an energy as possible.

For XRF, the low energy detection range is both a factor of the detector and the light-shielding materials necessary to separate visible photons from x-ray photons. The X100-7's probability-of-detection bottomed out below 2Kev, and that was at least partly due to the molding compound. The PMT/NaI scintillator I have are not going to be very good for us because of the light (and water) shielding really affect the low energy range. I think yours is a bit different construction so it may be better in that regard.

The theoretical maximum energy is set by the Am241's gamma photon at about 60Kev, but the X100's response starts to rapidly fall above 10Kev. For scintillators the max-energy is determined by a combination of the choice of scintillator material (ones with high atomic-weight components = higher absorptivity), and the physical size of the scintillator -- for higher energies, bigger is better. And more $$.

The simplest answer is that the X100's energy range limitations established the 6-8Kev range as best, in terms of the goal of "more than a spark test". Widening the choice of detectors widens the energy range.

As far as probabilities go, a 60Kev photon should be able to excite any and all x-ray lines below that. As a result, XRF spectroscopy has to employ element-extraction operations to produce quantitative results.