Needing more than a spark test?

[rwm]

It is kind of interesting that they built the hardware separately and plugged a HP tablet into the gun to do the processing and data point fitting. That makes sense in the way you are approaching this. They are almost separate projects.

This thread is now so long that I have lost track of where we are and what has been decided!
Can someone summarize the thinking so far? What is the current thought on the best type of detector, signal processing, etc.
I for one, would really like to build this but I do not have the software and computing background to do so. I can put the parts together but I will be dependent on you guys for the rest!

 

[homebrewed]

Making high voltages at low current is (relatively) easy.
40KV at 7mA is not, and I would rate that as lethal, or at the very least, bloody dangerous!
Several Cockroft-Walton voltage multiplier stacks, all in series, each separately excited by that one-turn toroid, and all dunked in oil is great to look at, or play with, but is complicated, and expensive!

Having the high voltage alone is not enough to make X-Rays. It needs electrons to happen, and get get accelerated, and target collisions, and that generally does not happen easily in air. I don't know enough about that "tape effect". If the effect is real, then something special is happening at the interface of the tape separation, perhaps within the tape. Apparently patented, (not the same thing as being a real worthwhile thing), it's still moving parts and complications. There is not much that can come close to a smoke detector source as a convenient way of getting a X-ray response out of stuff you put in it's way.

I am making the effort to get back to the KiCAD, maybe to catch up to Mark a bit. Unfortunately, it still has to be in between hangover-like spells, but things are improving.

7mA @40KV comes to 280 watts, a lot more power than needed for XRF. I thought the narrator said 7 micro-amps but if he said milliamps, he misspoke. We had a real-time Xray unit at work that could put that kind of power into a target, and the Xray gun was NOT a cute little glass vacuum tube.

The capacitance at the final output stage of a voltage multiplier capable of sourcing 7 uA still could give you a nasty jolt!

BTW I totally agree with the convenience factor of using smoke detector capsules. It's interesting to see what lengths inventors sometimes will go to in order to come up with a unique (patentable) item, but that approach often comes to naught.

 

[graham-xrf]

7mA @40KV comes to 280 watts, a lot more power than needed for XRF. I thought the narrator said 7 micro-amps but if he said milliamps, he misspoke. We had a real-time Xray unit at work that could put that kind of power into a target, and the Xray gun was NOT a cute little glass vacuum tube.

The capacitance at the final output stage of a voltage multiplier capable of sourcing 7 uA still could give you a nasty jolt!

Hmm - sorry, my mistake. He did say "micro-amps", so then 0.28 Watts. I rate that at the top end of nasty!
I guess I got conditioned by the eBay common definition of what is a "milli"amp when claiming 50,000mAh for phone sustainer USB battery packs

 

[WobblyHand]

Wonder what the net capacitance is in that circuit. 1nF charged to 40kV is 0.8 J, which might kill you.

I was shocked by a Van de Graaff generator that had about 0.1J and I collapsed on to the floor due to involuntary muscle contraction. I'm darn lucky it was through my right hand and right leg. Spark was well over 3 feet long in air to my idiotically pointed finger. One moment I was pointing to corona, the next moment I was on the floor because nearly every muscle on my right side contracted at once. I simply fell over. Was sore for a while...

 

[graham-xrf]

It is kind of interesting that they built the hardware separately and plugged a HP tablet into the gun to do the processing and data point fitting. That makes sense in the way you are approaching this. They are almost separate projects.

This thread is now so long that I have lost track of where we are and what has been decided!
Can someone summarize the thinking so far? What is the current thought on the best type of detector, signal processing, etc.
I for one, would really like to build this but I do not have the software and computing background to do so. I can put the parts together but I will be dependent on you guys for the rest!

Sorry Rob - you have been patient. Allowing that you have kept up with us, and already understand the principles, Let me try for others..

1. We have had an expanded discussion that was all about self-learning, teasing out the rubbish, and evaluating all the ways this can be done. We are agreed that using some smoke detectors, and an available large area PIN diode is a feasible way to excite stuff into making X-rays, and detecting them. The stress has been on simplicity, and lowest cost, consistent with getting a really worthwhile result.

2. We have minutely explored, dissected, criticized, and simulated circuits that will do this stuff. One very viable method involving high gain photomultiplier detector tubes, and scintillators that make measurable flashes of light is put aside. Good as it is, the bits are hard to come by, and need high voltages. We go for the PIN diode, stolen off a Pocket Geiger project, because that's still the cheapest way.

3. We have figured out, and fully understood how to quantify the energy in the X-ray florescence flashes, in a relative sense calibrated against known sample materials. We have some fully developed (free) software used in Universities for studies at CERN and similar projects called PyMCA. It runs OK on the tiny computing boards we intend to use - if we use it. We also understand enough to hang together much of our own code. The software will accumulate statistical counts of returns of similar energy, providing a way to estimate the proportions of elements seen.

4. We are agreed on the importance of capturing and preserving as much information as possible without it becoming lost in circuit noise, whether from unwanted interference, or simply from the molecules in the amplifier components bashing about. It turns out the ADC (Analog-to-Digital) component performance is as critical as the high-gain low noise amplifier.

5. Mark has put together some experimental arrangements to get actual pulses to happen, and done extensive work in using software to aid evaluating the detected pulses, both to filter out the useless, and to measure the good ones. Mark and I end up using ADC chips that are so nearly identical that they can fit on the same tracks, with only some differences in specification. I think Mark's choice is better value. Mark now has a PCB board design that carries the critical components, and has ordered one with the difficult surface mount assembly already done.

6. I think we are at a fairly advanced experimental stage. Even as we may get some setups functioning, we have to pull together a physical mounting that is reasonably easy for anyone to put together - but as a suggestion. There is plenty of room for anyone to get creative about how to package the whole thing. I am taking a similar path as Mark. In many ways, our view of the critical parts has converged, to the extent we almost end up with the same basic circuit. The small computing board that manages the data count collection can be any that is capable. I happen to be using Raspberry Pi, but it could as easily be a Teensy.

I know - I have left a lot out!

 

[homebrewed]

I know - I have left a lot out!

Sure, but there really has been quite a sorting process to get to where we are. I suppose the most striking thing about it is that it's all been on public display.....and perhaps that we are still communicating in meaningful and cordial ways .

 

[graham-xrf]

About the physical front end design.
Taken from the tear-down video when I was not being so excessively distracted by Mayte Mateos (Baccara).
The feature at the sharp end deserving of the zap-ping sound effect had a 45° corner "reflector" looking thing which the presenter said was "silver".

Then we also discover that the front of the zap gun was not open aperture, but instead blocked by a sheet of metal acting "transparently". OK - that would make it thin aluminium, just like with the scintillator receptacles on the front of photomultiplier tubes. This is something I think we should have!

If it were, say, 0.5mm (about 0.020") very pure aluminium, or anything we find handy, there are advantages. Even something from one of those slightly thicker products like the foil tubs used for Chinese take-away food. That kind of foil is pure, as opposed to sheet metal alloy.

Using this allows the entire diode front end with circuit to be shielded. Sure, from light (although ours has black paint anyway), but mainly, it constructs a total Faraday cage that will not allow any external electric fields, and all but very low frequency (60Hz) magnetic fields, to exist inside. We would not be waving around a diode antenna with 80dB+ gain. You could bring your hand up to it without stray capacitance coupled effects.

Aside from having built in a 50/60Hz deep notch filter in the design, I think if the connections to the diode were low capacitance, fine wire, PTFE insulated twisted pair threaded through small ferrite toroids, then 60Hz magnetic fields might well be substantially excluded, and with way less bother than trying for mu-metal shields.
--------------------------------------

Making more photon collisions
This gets more subtle, and I admit, I don't really know what goes on here. This is just what I be thinking..
Back to the tear-down design.
My first thought was that the whole point of standing something metal at 40°, or perhaps somewhat less, would be to "redirect" the X-ray beam, but we know that X-ray "beams" do not get "reflected" like light off a mirror. If the intention was to get a shower of lower energy photons, say from silver, then getting them from the first surface, as opposed to those that might have to make it right through, does make sense. This, I think, would also be "second-hand xrf working on the leftovers of a first silver collision xrf" ! Could this actually be a great idea?

The excitation question(s).
So I think on back to a fundamental question I never did get to learn the answer for.
"When a higher energy (60KeV) photon arrives and excites some xrf photon out of an atom, is that the end of it for the 60KeV photon"?
What if it only provoked (say) some 5.4KeV flash from chromium. What happens to the rest of the energy?
Can we expect we might get both 5.4KeV AND 5.9KeV xrf photons out of the same atom of chromium?
Crucially, might these happen at the same time? Perhaps not.
It may be that an entire 60KeV incoming photon may be expended (wasted?) in only producing one flash at a time.

There is an actual delay before the elevated electron drops out of it's excited state. and I have no idea whether a single incoming photon can excite both Kα1 and Kβ1 states in one go, nor whether they flash out simultaneously.
Can a single higher energy (60KeV) photon ever wring more than one xrf flash out of a pile of atoms?

Back to the "silver" target on the zap gun
Here, of course, we do not have the occasional single arriving photon. We have a "beam", like light, of presumably zillions of them, all colliding into the tilted piece of "silver". If that is what it is, it makes a shower of 22.16KeV and 24.94KeV photons. Given that the zap-gun only had 40KV to start with, these can only be one photon kind at a time.

I had first thought that this might be a handy way to get a huge increase in xrf returns from the material under test, but it cannot be! It also excludes getting returns from all elements heavier than silver. Have we missed something? Can X-ray "beams" ever be "reflected".

From all the above, I end up very much liking the aluminium front cover, and I fail to understand the motivation for the "silver" plate deflector-looking thing.

 

[rwm]

If I am reading the comments on You tube correctly, the metal foil is 8um beryllium foil. Very fragile. Check my work.

My understanding is that the incoming gamma photon excites multiple electrons in the sample in series until it loses all energy. Many of these are outer orbital electrons and we will not see the low energy emitted from these. However, some of these will be characteristic xrays (k shell) and that is what we are picking out.

The xray tube in that device creates a polychromatic beam with 40kV peak energy. The curve is not Gaussian but is skewed to the higher energies. My guess is the idea behind the silver is to create a monochromatic beam (secondary xrays from silver) so you are bombarding the target with only one (or 2?) discrete energies. Again, this scheme is unnecessarily complex if you have a radionuclide gamma emitter as in your design.

 

[homebrewed]

I too have been mulling over the function of the silver foil. I don't think it is used to increase the x-ray flux because the maximum energy from the x-ray tube is only high enough to get one 22Kev photon from an incident 40Kev photon. It _might_ be used to reduce the energy of the photons hitting the sample, to avoid exciting other transitions that would muddy the water, so to speak, but that is pure speculation on my part.

If the point is to have a monochromatic x-ray source, our Am241 fills that role nicely, without needing any silver to do it thankyouverymuch.

One phenomenon that affects all types of x-ray based elemental analysis tools is secondary fluorescence, where additional x-rays are generated by higher-energy fluorescence x-rays coming from elements with higher atomic numbers. If we look at iron's k-alpha line, which is 6.4kev, that could be excited by an x-ray coming from something like tin, which has a k-alpha line of 25.2kev. So a lot of tin would screw up a simple calculation for iron concentration. I view that as a complication that wouldn't affect the kind of setup we're working on because ferrous alloys aren't going to have much, if any, silver present. Nickel's xray line is a _little_ more energetic than iron's, but the difference is small enough that secondary fluorescence from that interaction would be down in the mud.

Effects like Compton backscattering will be present no matter what kind of source you're using. For this kind of thing, system geometry is the primary determination of how many backscattered photons get into your detector. Some papers I've looked at suggest our geometry is about the best for reducing the amount of backscattered xrays.

 

[graham-xrf]

If I am reading the comments on You tube correctly, the metal foil is 8um beryllium foil. Very fragile. Check my work.

My understanding is that the incoming gamma photon excites multiple electrons in the sample in series until it loses all energy. Many of these are outer orbital electrons and we will not see the low energy emitted from these. However, some of these will be characteristic xrays (k shell) and that is what we are picking out.

The xray tube in that device creates a polychromatic beam with 40kV peak energy. The curve is not Gaussian but is skewed to the higher energies. My guess is the idea behind the silver is to create a monochromatic beam (secondary xrays from silver) so you are bombarding the target with only one (or 2?) discrete energies. Again, this scheme is unnecessarily complex if you have a radionuclide gamma emitter as in your design.

Excellent! Thank you for getting at the detail, and for providing the explanation that an incoming gamma photon can expend it's energy into more than one spin, and more than one shell, and even into more than one atom "in series", if there is enough energy left over.

Beryllium, having only one possible electron shell, has traditionally been the material of choice for X-ray transparent windows. Crude (and hazardous) precipitates cost about $4/gram for element collectors on eBay. Actual metal foils 3/4" diameter, 0.0005" thick I see for $400 for six, not including $16 for shipping. This is maybe not for us!

What might we detect?
Between beryllium and aluminium are several that we would love to get some counts from, these being carbon, oxygen, flourine, sodium, and magnesium, but I don't think those are likely. I think we may have to infer the presence of these if we identify a probable alloy from the bigger returns.




Suppose we go for it, and consider whether we might see aluminium. In theory, we might. One might have to increase the gather time to a couple of minutes or more. The PIN diode response probability for 1.48KeV is down to about 2%, meaning a count of only a couple of hundred out of several thousand incoming smoke detector photons.

Letting the bias become 6.6V (2 lithium cells) instead of 10V (3 cells) lowers the dark current from around 2.5nA to about 2nA, and the capacitance increase is from about 82pF to 100pF. The benefit in dark current noise, is hardly worthwhile. The change in capacitance is not going to lower the signal significantly. From the response of the PIN diode, and with the amplifier hoped-for noise figure, and the 91dB possible SNR from the ADC, we might marginally detect aluminium, sometimes. Magnesium is (just about), theoretically possible. Sodium is at probability 0.1%, would yield a count of about 20, even if 20,000 photons were expended on it. That could take much more than a couple of minutes!

Realistically, I think we can happily re-purpose the Chinese take-away aluminium foil tub after eating the contents, and use it for shielding. If I thought we could ever detect aluminium, magnesium, ans sodium, I would think to a conductive carbon shield window, instead of beryllium.

The more we mull this stuff over, the more I become convinced we have got it exactly right!