Needing more than a spark test?

[graham-xrf]

Robert has it right; different shells. This may help to clear things up.

Electromagnetic spectrum - Wikipedia

en.wikipedia.org

At least for the sodium yellow visible lines, they are both from ground state to outer orbital 3 (or M).
You get two lines because the energy to get to 3p3/2 is slightly different to get to 3p1/2
The real stuff we are after has X-Ray wavelengths.
Sorry it took a while for me to find that one out.

 

[graham-xrf]

Detection range:
For the lowest energy in the materials we might put this thing against is from Lithium putting out 22nm.
Beryllium does 11.4nm
Carbon would deliver 4.4nm
Magnesium gets shorter at 989pm
Good old Iron does 193pm
Moving along to the highest value on that page is Silver Ag with Kβ1 at 24.9424 KeV, which causes 48.7pm.
Beyond that, like up to Gold (Au) 77.98KeV, the numbers get silly.
The above numbers come from energies in the XRFSpec_ENG.pdf in posts #9 and #77
The little spreadsheet did it (blame anything)!

These wavelengths cover a range well past what the scintillators we have been mentioning apparently can do.


OR - I have messed up again!
Sorry folks, but when I do try for something that cost more than about $2, it will be a bit hardball.
All the wavelengths I get so far are too short for this kit. Something must be wrong!
Other guys have got a glow out of this stuff.
The numbers have to make sense at least once.
Maybe someone can spot the fumble

 

[homebrewed]

Still a bit more than what I'd want to pay, but check this out: cheap gamma spectrometer. And there are various modules for less that you probably could assemble into an XRF system.

If folks still want to roll their own detector, you also may want to look into so-called SiPM's, solid-state photomultipliers that look to be much easier to operate than traditional PMTs. For example, the microfc-30050 from ON Semiconductor, $55.70 at Digikey. Or for $68.75 you can get an evaluation board at Mouser. Wow! There actually may be a path toward making a reasonable DIY XRF after all...

 

[graham-xrf]

$600 is out of court. I,m sorry - but a flat no.
My internal evaluation reasoning is a bit phsychologically complicated, but the product loses my trust from the beginning.

If I had a product even with with most excellent high tech stuff inside, and was asking that much for it, I would not feel the need to package it in a crappy plastic box with USB lead that would not do justice to a bargain basement phone backup battery.

Pricing at $599 introduces a second element of presentation manipulation covering arbitrary speculative price pitching. The other thing I note is that it is oriented to pulse counting, not pulse sample capture. Many of these radiation detectors use high gain threshold trigger logic transitions, losing all the analogue information it the cause of counting a clear click.

I dunno. I am the type who might go to a shop three times to peruse and mull over the merits of something, and still walk away!

While PMT tubes are very cool in their own way - I like this SiPM idea.
I am OK to try one after checking out the data sheet.

BUT - if I am going to get into the datasheets, choosing something that best suits, I need first to resolve the "wavelengths too short" question. I did not think it would be so tough to wring this one out. All the products data are about the wavelengths responses they are good for. All the data we have is about the energy of the photons that might land on them. The equations are simple, to get at the conversion factor. Still not making sense!

How do we get at the wavelengths that are going to happen coming back out of a steel alloy we have hit with radiation?

 

[RJSakowski]

The wavelength is given by the equation wavelength (nm) = h*c/E where h*c is 1.23984 KeV nm and E is the photon energy in KeV. The lightest element of interest , Mg, has an fluorescence x-ray energy of 1.253 KeV which would have a wavelength of .798 nm. We would not be able to excite fluorescence above the source energy which is 59.7 KeV or .0208 nm. If you wish to see carbon, its emission is .277KeV or 4.476nm. I expect that the carbon will be fairly transparent to the source photons though.

 

[homebrewed]

The wavelengths you are calculating are correct. The graphs you show in message #82 are the EMISSION spectra of the scintillators, NOT the wavelengths the materials will detect. The graphs can help you choose the right PMT -- obviously, if the scintillator is emitting light that the PMT can't "see", you've got a problem.

The parameter of interest in scintillators is the number of photons generated for a 1Kev photon, which for Ti-doped CsI, is 54/photon. This is the one that sets the amplitude vs x-ray wavelength.

 

[homebrewed]

Here's an Instructable describing a DIY gamma spectrometer that uses an arduino and a few other components to implement the pulse height analyzer. I think more of the functionality could be implemented in S/W (like the "mono flop" -- also called a 1-shot multivibrator), but if not it's still a pretty simple circuit. I don't like the 12V signals being routed into the Arduino's logic pins though. I'd clamp them to whatever the Arduino's Vcc is (some are 5V, some are 3.3 depending on the flavor you have).

I just bought a CsI scintillator to play with. I wonder if the americium sources in the old smoke detectors I've got lying around are active enough...

 

[graham-xrf]

The wavelength is given by the equation wavelength (nm) = h*c/E where h*c is 1.23984 KeV nm and E is the photon energy in KeV. The lightest element of interest , Mg, has an fluorescence x-ray energy of 1.253 KeV which would have a wavelength of .798 nm. We would not be able to excite fluorescence above the source energy which is 59.7 KeV or .0208 nm. If you wish to see carbon, its emission is .277KeV or 4.476nm. I expect that the carbon will be fairly transparent to the source photons though.

Thanks RJ - it gets us closer.
I got h * c = 1.9864458121E-25 J.m , which is exactly the value found in Wikipedia "Planck Constant"
That Planck constant h = 6.626070E-34 is an exact value, defined, the basis for fixing the kilogram.

To use your value for h * c, and divide by c
1.23984/c = 4.1356610779e-09 which is the number I see in the Wiki.
The given usual value is 4.135667696E-15 eV.s, but that's OK. there were 5 decimals to start with.
The value is also exact, but not expressible as a finite decimal, so approximated to 9 decimal place

The e-9 instead of e-15 is because we go direct to nanometres, and use KeV to begin with.
I get it that the value you used has the eV = 1.602176634E-19 built in.
- - - - - - - - - - -
Looking at some of the solid state sensors suggested by @homebrewed is interesting. Unfortunately, the areas are really small compared to common PMT tubes. Generally multiples of 6mm square.

MICROFC-60035-SMT-TR1 6mm x 6mm, tileable $72,92
MICROFJ-30035-TSV-TR 3mm x 3mm tileable $ 23.59
.. and so on
The selection is here --> Mouser.com ON Semiconductor Photodiodes
There is also a range of TO3 size windowed package avalanche photodiodes. Generally expensive .

The other things I have learned about scintillators is CsI is much more rugged, and a good bit less hygroscopic than NaI. The sensitivity of CsI is much lower. PMT tubes need magnetic shielding, mu-metal as well as lead at the front. Finally, they have to use a holder that excludes all light when put over, or up against, the test sample.

 

[graham-xrf]

Here's an Instructable describing a DIY gamma spectrometer that uses an arduino and a few other components to implement the pulse height analyzer. I think more of the functionality could be implemented in S/W (like the "mono flop" -- also called a 1-shot multivibrator), but if not it's still a pretty simple circuit. I don't like the 12V signals being routed into the Arduino's logic pins though. I'd clamp them to whatever the Arduino's Vcc is (some are 5V, some are 3.3 depending on the flavor you have).

I just bought a CsI scintillator to play with. I wonder if the americium sources in the old smoke detectors I've got lying around are active enough...

Um.. lookin for the Instructable?
The Am241 sources will be blasting away much like the day they were made. The half-life is something like 432 years, slowly making Neptunium I think.

After using Arduinos for a while, I went for Raspberry Pi. Even if the Pi gets given over to a project for which it is overkill, it is affordable enough, and is easily OK to use for compiling software, while it is on the internet, browsing data sheets, downloading software, and driving screen and mouse and keyboard. If after you get the project going, you can use a Pi-zero, stripped down card without video etc.

I just got used to high speed USB3, WiFi, Bluetooth, Audio, 2 x HDMI for fifty bucks or so.
The Pi-4 in the picture had the operating system on the 500GB Samsung SSD. The little microSD card only does a trivial boot-up to launch the OS on the Samsung. This little PI-4 has a hardened mailserver and web server on it, and has stayed on for months except for bad weather power fails, but it just puts itself back to working when it restarts.
I have unplugged the mouse and keyboard and screen.

The older Raspberry Pis are given over to projects - like the one we are considering.



I had the Pi-4 playing on YT while I was figuring out the software



The Pi-3B in the acrylic case got to work with this little A/D converter. it was OK - but multichannel 16-bit with MHz sampling rates are more to my taste.

--> A/D converter sampler

 

[RJSakowski]

The value for h *c that I gave was from wikipedia. https://en.wikipedia.org/wiki/Planck_constant
There many different ways of of expressing h* c but since we are interested in converting KeV to wavelength, it made sense to use that one.
From the table on the right, you will see the value expressed as 1.23984193 eV-um which is the same as KeV-nm.