Needing more than a spark test?

[graham-xrf]

I have not yet given thought to filter discrimination tricks to mask the unwanteds, but what you say does trigger a thought.

In messing with the geometry, I tried to minimize photon routes from lead shielding surfaces that could arrive at the sensor. If the outer tube is large enough radius, then the photon routes at the diode are shadowed to all except the test piece, and the surface under it. That might be something low density, thick enough, and made of stuff that we can't get a glow from. I was thinking a bit of polystyrene insulation might do. Anything will do, so long as it's all carbon and hydrogen and maybe chlorine (PVC). PTFE is nice. Flourine X-rays are too low to care!

If we do not have the test piece sat on a sheet of lead, the lead peaks go away The diode only sees the back of it's tube lead shield, not direct top surfaces that saw a 59keV photon arrive.

The lead, if excited, is capable of delivering L-shell Lα1 = 10.55keV, and Lβ1 = 12.61keV, which is right in the middle of energies we would collect.

The values for lead are:
For Pb
K shell --> Kα1 = 74.9694 Kβ1 = 84.936
L shell --> Lα1 = 10.5515 Lβ1 = 12.6137

The energy is not high enough to get a glow from the K-shell, but we would see the lead there from L-shell responses. It may not matter, because, out of your list of test metals, only zinc comes close with 9.57keV. If we have preserved our pulse shapes well enough, we should still be able to see the separate peak.

The obvious question is - if we do have lead responses, are there any filtering tricks one can use to let the wanted X-rays build up count, to tip the scales against the lead response?

Going where I have not yet fully understood it yet.
On the getting of calibration pure elements, I suppose I may have to gather some. On calibration, one can have software setups to deliberately spread the X-axis buckets over a more limited range, to "zoom in", using calibration elements nearer to what is suspected to be in the alloy.

e.g. Setting the "zoom" to have more buckets devoted to collecting energies in range 2-8keV, the cadmium L-shell would have handy 3.1keV and 3,3keV responses. We would have ignored the 23.1keV and 26.1keV. Even at the low 2% probability of getting a diode response at 3keV, we may see it if we wait some seconds (minutes??)

Do tell what Amazon search term you used?
In my design, there is now provision for inserting a disc of metal "filter" up the shadow tube to be over the sensor, where Am241 energy cannot directly see it.

 

[graham-xrf]

The Mouser electronics hardware has arrived!
We shall see what this stuff can do.

 

[homebrewed]

Do tell what Amazon search term you used?

I was searching for "Manganese metal" (to use as an energy filter) in Amazon's "industrial and scientific" subsection and came across the set of cubes. To save you some hunting around, here's a link to the ones I bought: metal cubes.

I still haven't found an affordable source for thin manganese sheet. You can get it in powdered form, which may turn out to be about the only choice. Mix it with a binder and cast it. Getting good uniformity of both density and thickness would be the tricky part. Another idea would be to make a solution of a manganese salt, like MnSO4 or MnCl2, and use that.

 

[graham-xrf]

Mix it with a binder and cast it. Getting good uniformity of both density and thickness would be the tricky part. Another idea would be to make a solution of a manganese salt, like MnSO4 or MnCl2, and use that.

I get it that for filter purposes, you want something of uniform thickness. A salt in solution is problematic. I find liquids anywhere hard to control.

Thanks for the Amazon link.
I see that on amazon.co.uk search, there are sets for educational density testing purposes. One set is for 12 cubes for £31.29 (£41.57) which got a 4 star rating because the cube's sizes were slightly mismatched. Also, one of them is "marble" and one is "perspex".

--> Set of 12

Also, it's a chancer scammer's paradise. You have to look carefully to be sure the description says it's a set.
One of them manages to choose a price + postage, which when added together, equal the legit product price, and the description is very wordy, designed to obfuscate, and calling the color "10mm Cube Bismuth". Even though the picture shows a bunch of 8 of them, you only get one!

--> Misleading scam description cubes

I think the educational sets may generally be a better deal than the ones aimed at periodic element table collectors. I guess they could get more expensive if precision machined and laser etched or engraved with Periodic Table info.

- - - - - - - - - - - -

I have to now devote more time to getting things set up for the coming building operations. I have made up the profile stakes and boards, and I have to set them up on the site. Also, I have to do major organize and tidy up, de-clutter, and relocate a whole lot of stuff. I am not so good at that.

Also, I have to put the woodwork hat on, and move over to routing the frame jambs and installing the fire door into the garage + more other domestic stuff than I want to know about.

 

[homebrewed]

The element samples arrived today:

 

[graham-xrf]

Biggest-Ass Photodetectors!

I don't exactly know how this works, but it seems to be a huge lump of liquid scintillator, surrounded by a bonkers spherical array of photo-detectors.
I don't care it's PMTs or PIN diodes, that lot looks expensive!

 

[homebrewed]

I don't recall hearing about that particular neutrino detector. Neutrino detectors typically do use large numbers of huge PMTs, along with what amounts to a massively-large liquid scintillator.

On our current subject, I found I already have some aluminum guilding leaf (left over from a work-related experiment). So while I was ordering some stuff from Mouser for a different project, I added a small quantity of conductive ink to stick it down over the hole I cut in the copper shield to expose the detector and LMC662. It's considerably thinner than kitchen-style aluminum foil so it should work pretty good for passing low-energy x-rays while shielding the detector from powerline noise.

On the subject of avoiding the 10.55Kev signal from lead, I had a thought that might be worth exploring. A composite shield, made of a fairly thick aluminum sheet ~1cm will completely absorb 10.55Kev. Since that much aluminum only reduces the primary 60Kev gammas by about 50%, back it with a lead sheet. The 10.55Kev's it sends back through the aluminum are absorbed. This way, only the relatively poorly-detected aluminum peak at 1.5Kev will make it to the detector. And of those, only the top 30 microns will contribute a significant number of those photons.....the rest get absorbed in the bulk of the aluminum. But guess what: the 10.55Kev photons that will cause the fluorescence @1.5Kev are reduced by a factor of 1/3E-31 before they get to that top 30 microns! I think that will pretty much take care of everything but the 60Kev-->1.5Kev conversion at the surface of the aluminum, and it's a long way from the energy range we're interested in.

The relatively high absorptivity of ferrous metals at ~6Kev means that we're only going to be sensing the surface of the sample down to a few tens of microns, no matter what. That's just baked into the cake, so to speak.

 

[graham-xrf]

The filtering is interesting. I think we aught to consider whether we care if there are responses for lead in the middle of responses for the stuff we are trying to test. I had a quick look at the energies in the elements list, looking for any that could get masked by the lead. 10.55keV and 12.6keV. If those are there, we know to ignore them.

A software digital filter?
To some extent, my notion may be completely stumped by the bucket resolution we can achieve. To be able to have sufficient measurement accuracy in counting up the samples for the duration of a pulse event, so that we know buckets values "near", but not exactly "on" the lead signals is what I am thinking about. So long as the counts are consistent enough for there to be clearly distinct buckets for non-lead things, we could implement a "software filter", and simply remove the 10.55keV and 12.61keV buckets from the results.

Even test for lead in alloys?
Getting more fine-tuned, it might be possible to establish a base count in calibration of the lead buckets that are always there, known to be coming from the shielding. Then, if there is lead in the sample, or maybe an alloy known to have lead in it, we subtract the base line count values, and might be left with counts revealing the lead content in the tested sample or surface. These, of course, are proportionally scaled to the test duration calibration. This sounds fine in theory. One hopes that our measurement is not so noisy and messed up that we cannot separate them.

High L-shell values combined with specific signatures near lead allow a logic to take hold. e.g.Iridium has a value at 10.7keV. Unless we can resolve a bucket 10.61keV as being separate, we would not know. Bismuth, next to lead has a 10.8keV response, so might not be seen, except it also has a 13.0keV. I doubt we would ever encounter Actinium, or if we did, we are in a big mess!

Going back down the list, everything lower number than Iridium is unambiguous, the L-shell becoming feeble near Molybdenum, but Moly's K-shell comes to the rescue. Arsenic has a response at 10.54keV which might have it lost among lead, but the K-beta shows it up at 11.72keV.

- - - - - - - - - - - - - -
The 662 will be off shortly, and replaced. This leads to a decision about my lead shield holding the Americium pellets. I had imagined it as having the circuit board in the tube behind, which meant the sensor needed to be set at right angles to the end of the PCB, possibly on it's own board being simply the cut off end of the Geiger board. This notion was driven by thinking the final gadget would need it's own board anyway.

Staying with the notion of using, in part, a modified Geiger board means it sticks out at right angles out of what would have been the tube behind the shield. So at least for the first lash-up, I will do something like that. I took the copper screen off, because the PCB will be in a metal enclosure anyway.

Are we saying that under the lead skirt outer shield, 60Hz interference can still get in?

 

[homebrewed]

Are we saying that under the lead skirt outer shield, 60Hz interference can still get in?

I'd have to say that we don't know yet. Gain at 60Hz will be high, and there will be a hole in the lead shield that may allow some electrical noise in. I'm just being proactive when it comes to keeping noise out, particularly since noise is going to be our enemy when it comes to energy resolution.

 

[graham-xrf]

I'd have to say that we don't know yet. Gain at 60Hz will be high, and there will be a hole in the lead shield that may allow some electrical noise in. I'm just being proactive when it comes to keeping noise out, particularly since noise is going to be our enemy when it comes to energy resolution.

Pre-emptively, the second gain stage can be configured to deep-notch 50Hz or 60Hz with a resistor selected tuning option.
That could be a case of bolting the stable door after the horse has gone!
I agree that stopping the 60Hz before it gets into stage #1 is the best way.

Screening, and paying attention to common mode current paths can get electric field pickup below the noise floor. Magnetic pickup can be minimized by limiting the area of the pickup loop, which is pretty small anyway, at the diode connections. If the conductors are brought close together before they go to the amplifier, and guard rings are used, that might help. I am unsure whether a piece of mu-metal under the diode would help, but the mind imagines the interfering power line field taking the preferential route through the metal, and failing to couple to the diode connect conductors.

A piece of mu-metal, placed under the diode, held under the copper tape even, if you retain it should work. I don't think you need the copper tape if the circuit is in a metal box. It was only the expedient because the Geiger was in the little plastic case.