Needing more than a spark test?

[WobblyHand]

I have a 3d printer. I can print something as large as 205mm x 250mm x 210mm, although that is a heck of a lot of filament! I am limited to 1.75mm diameter filament, however. Anyways, just another possibility to consider. PCL filament is at least twice as expensive at PLA. Found a 1kg reel at Amazon, they want $55. PLA can be found at less than half that, per kg. PCL is printed at around 60-90C. PLA is printed around 215C. I'd think that either would be burned out at the melting point of lead (327C).

 

[graham-xrf]

I seem to recall that model now. I'll confess that I haven't followed your progress through all the posts.

I don't have a 3D printer myself. As to a low melting point filament, this is what I found. https://filaments.ca/collections/3d-filaments/products/pcl-low-temperature-filament-yellow-2-85mm At a melting point of 60ºC, it could be melted of of a mold with hot water. It would be a simple way to make complex geometry.

As to the design, my assumption is that the sources are essentially isotropic radiators so the orientation of the source is of little consequence.
I would want the sample to be a close to the detector as possible, consistent with proper shielding of the source. I would also want as much lead between the source and the detector as possible. To that end, on your model, I would fill the air space below the detector in the drawing with lead and extend the lead behind the source. More shielding around the detector too.

It would be easy to have the disc of behind the detector be of lead, which might stop incoming from outside, coming right through the rear of the instrument. In the geometry shown, the detector is completely surrounded by the lead collimation cylinder. Also, the sources are completely surrounded by lead, or lea inserts. Photons going straight out the back of a source enter 3mm of lead. I think it's photon-proof.

 

[graham-xrf]

I have a 3d printer. I can print something as large as 205mm x 250mm x 210mm, although that is a heck of a lot of filament! I am limited to 1.75mm diameter filament, however. Anyways, just another possibility to consider. PCL filament is at least twice as expensive at PLA. Found a 1kg reel at Amazon, they want $55. PLA can be found at less than half that, per kg. PCL is printed at around 60-90C. PLA is printed around 215C. I'd think that either would be burned out at the melting point of lead (327C).

The critical bit is nothing like that. About 60mm or less diameter. A couple of inches.

Polylactic Acid
Polymer of lactic acid (PLA).
Hydrogen, Carbon, Oxygen, and that's it!
None of these will make anything we can detect, and there is no path past the lead shield collimators anyway.

Polycaprolactone (PCL)
Made of the same elements as PLA, but it's a polyester
No issues, except it is expensive.

We never have to dump molten lead into nor against any of it. We just push the lead bits into the right places, maybe with a touch of glue under.

 

[WobblyHand]

@graham-xrf I use FreeCAD, so if you were to send me a model, I could print it from the model file. I'd prefer a native FreeCAD file rather than any other file type, so the model could be modified to accommodate any 3d related printing (or casting) issues. I'm using 0.21.0 compiled from source, so I think I can read previous generation files.

 

[graham-xrf]

@graham-xrf I use FreeCAD, so if you were to send me a model, I could print it from the model file. I'd prefer a native FreeCAD file rather than any other file type, so the model could be modified to accommodate any 3d related printing (or casting) issues. I'm using 0.21.0 compiled from source, so I think I can read previous generation files.

OK - let me have a chance at it to modify the model to have the lead buckets, collimator, and what else I can think of at this stage. I will put the model file where you can download it, or place it here, or email it - whatever.

I have to be impressed with "compiled from source". I have done such things, and it's always a bit of a task to be sure all the dependencies are in place. I took the lazy way, and used the PPA maintainer's version of FreeCAD Daily. 0.21
https://launchpad.net/~freecad-maintainers/+archive/ubuntu/freecad-daily
.. and I used the correct Ubuntu version choice for my Linux Mint (Mate).

It updates easy along with everything else from..
sudo apt update
and then
apt upgrade

 

[rwm]

I am thinking that if we later want to cool the detector, the right angle board attachment would be better? I am not sure how important cooling will be.
If you go with PLA and plaster, (like lost wax) you would burn out the PLA in a kiln or oven and then pour in the lead. It is not like lost foam casting where the foam stays in place for the pour. I think any of the approaches would be fine: lead bucket shield embedded in plastic, the entire front end cast in lead, or machining this from a lead cylinder. Having no experience machining lead, I personally might try casting.

Edit: Actually I may have a better idea. You really only need to shield the PIN diode from the sources. Why not just make a lead ring 3mm thick that fits around the diode. It should extend in from of and behind the detector to a short degree. Then print a plastic disc to hold the sources or just glue them to a plastic disc. That would make a good test bed. Later, if you are concerned about shielding the operator you could easily add shielding around this assembly.

 

[homebrewed]

Truly fantastic results thus far.

Another possibility for positioning the sources might be to 3d print the complex shape and use lead inserts. For a single use one could print the punch(es) to press the lead into the complex piece.

This is getting exciting enough to spin a board. @homebrewed what's your front end (analog circuitry) look like again?

To minimize the effects of power supply noise you could run off a small 6V battery. I have a lead acid gel battery for this purpose. Did this for my radar chronometer. 50/60 Hz noise would be greatly diminished. The output voltage is about 6.4V. I ran it through an LDO regulator to get 5V, and a second 3.3V regulator for my analog circuitry. If I recall correctly, I'm using very low noise op amps similar to what @graham-xrf chose.

The analog portion of my setup consists of two separate boards. The first is the pocketgeiger, link to Sparkfun's schematic here. The back of the board has a footprint for JP2, which might be for testing the board; or perhaps an option for a surface mount header setup. I haven't checked the spacing to see if the header hypothesis makes sense or not. Soldering wires to board ground and PS4 on JP2, I tapped into the output of U1B. U3 is a quad open-collector comparator that is used to detect pulses, so its output is no good for our purposes.

Along the way I found that the switching regulator used to generate the 27V bias for the detector (and also 9V for the amplifiers/comparators) was injecting too much noise into the analog circuit path. Quite possibly because I removed the copper shield over the detector. I think it's necessary to remove the shield to avoid absorbing our lower-energy xrays. So I removed the switcher and am now biasing the detector with an outboard linear power supply, using PS6 on JP2. Removing the switcher also meant I had to go with an outboard supply for the amplifiers as well.

While looking at the output of U1B I saw that the pulse amplitude was pretty low so decided to design a signal conditioning board to boost the signal -- idea being to get a better match to the range of a teensy's ADC. Since the pocketgeiger circuitry uses a single supply, its quiescent output was (approximately) Vcc/2 so I also AC coupled the input to my signal conditioner. Here is a copy of the schematic:



The signal conditioner was a pure shot in the dark and shows it. The input's RC time constant is ridiculously long so it takes a long time to settle down when everything is turned on It uses a 10-turn trimmer to adjust offset: and the trimmer's input voltages are right off the +/- 10V supplies so there's no PSRR. The choice of opamps was what I had around, not the greatest in the world for noise performance. The whole thing picks up noise from the mains like crazy so opening up my box to adjust the trimmer is a pain.

Going back to the pocketgeiger for a moment, I also swapped out the LMC662 for a lower-noise dual from Analog Devices (IIRC, an AD8646), but that part could only handle a 5V supply. So I also had to swap out the 9V regulator for a 5V one. I did it that way because I didn't want to have the need for a +10 AND a +5 bench supply.

I think a better approach would be to redesign the signal conditioner to use amplifiers that work with +/- 5V supplies. That would simplify things a bit. Also, I'm currently just using the gain-of-10 signal path so U2 isn't being used. Just as well . If offset adjust _is_ needed, use a digital pot instead and tweak it remotely. However, a good opamp nowadays should be pretty darned good on its own when used in a 10X gain configuration.

One thing to consider going forward: the AD7667 and its cousins have a 2.7V Vref so it, effectively, has higher "gain" than the T4's ADC, whose Aref is hardwired to 3.3V.

 

[homebrewed]

I'll ask a question that may have been discussed before. Is your detector shielded from the source? Xrays travel in a straight line so facing the aperture toward the sample and away from the detector should eliminate any primary radiation. As to making your lead shielding. I would use a variation on the lost wax method of casting. A pattern could be machined from machinist's wax and used to make a mold, The wax us burned out and the metal poured to make the part. Jewelers have used this technique to make complex castings in gold and silver and large bronzes have been cast using this method. With the much lower melting point of lead, it should be fairly easy. IIRC, plaster of Paris can be used for the mold. Another possibility is making a reusable mold from silicone with a plaster of Paris shell. If plaster of Paris won;t work, the mold could be made from potter's clay and fired. Rather than wax, it may be possible to 3D print the master and burn it out. The clay may be required for that. Is there a low melting point filament available? Thinking children's 3D printer here.

The detector is, well, kinda shielded It's clear from what I'm seeing that it isn't shielded well enough!

I think the problem in my particular setup is that I was too zealous in my quest to get the sources as close to the sample as possible, while doing the same for the detector and sample. As a result the cross-sectional thickness of my lead shield probably is too thin around the perimeter of the hole in the lead sheet. I have a lot more lead sheet I can use to add some thickness there, that will definitely be one of the things I try in the near future.

 

[rwm]

I have some lead "cups" that could be useful. They are cylindrical and have an ID of 21mm with wall thickness of 4mm. They are 60 mm tall. It should not be hard to part off a short section of "tubing." What is the ID needed to encircle the diode housing?

 

[homebrewed]

I have some lead "cups" that could be useful. They are cylindrical and have an ID of 21mm with wall thickness of 4mm. They are 60 mm tall. It should not be hard to part off a short section of "tubing." What is the ID needed to encircle the diode housing?

The detector itself is smaller than the package. The detector is 10 x 10 mm so the diagonal would be about 14mm. For my setup, that's what established the ID of the hole in my lead sheet.

According to the X100-7 data sheet, the package is 16.5 x 16.5 mm. The diagonal then would be a little less than 23.34mm so that's the ID you would need. Just a bit shy of 1 inch.