Needing more than a spark test?

[graham-xrf]

OK - we are agreed that a transparent" shield can be aluminium.
Underneath the sensor will be lead. Pretty much the entire little volume will have a liner of lead.
The "measurement cell" will also have mu-metal. Mu-Metal, as a metal box, schieves electrostatic screening as well.

Concerning the mounting structure of the Am241 sources.
I had planned a gadget looking shaped a bit like a flashlight from the days of D cells with a open end that one puts up against the metal to be tested. If the metal sample is small enough, it gets put down over it, onto a small sheet of lead.

My first CAD modelling of it sucked. I started again. I hoped to mount 6 or 8 sources in small drilled recesses arranged around a circular lead tube, all pointing at an angle onto the "hot zone". The holes just deep enough so as not to waste radiation, but masking it from making to the sensor. In the centre of all that is the tube with the sensor up it, again arranged to collect the maximum of what pings upwards, but shadowed from what comes out the sources.
I thought the sources could be put into their settings, about 5mm deep, down onto a dab of epoxy.

It is a bit of a fiddle to get at them.
_ _

The ion chamber covers them up. They are facing "upward" away from the board. You eventually get down to a mounting disc that is itself peened into a contact ring. Maybe the arrangement differs depending on where you got them from.

_ _

I had thought the most suitable would be the approx 5mm diameter peened disc which has the approx 2mm diameter metal-looking thing that is presumably some alloy with Am241 thrown in. That means losing the contact ring with connect solder tab.



I think I can cut the outer peening with the little die grinder diamond wheel, and pop the middle disc out. That would be compact enough to set around the radiation zone target region. We have quite a lot of lattitude to use a wide angle. The photons won't reflect off anything, and they definitely can't go around corners.

The sensor would be set on it's own little PCB, which is attached at right angles to the electronics PCB. At the rear, the LVDS network type cable goes off to the little computer. The region of PCB nearest the data cable take-off has an initially unpopulated section to mount a PIC or something to turn the cable into USB. There may be some case to make it USB from the outset.

Local circumstances make it awkward to do the experiment here right now, so I am hoping someone can tell me what the deal is in attempting to make an approximately 2" end of a Am241 ray gun assembly .. out of lead!

Another question comes to mind. Other than as a calibration source, is the stuff in Thorium mantles useful as a source?
I have to check what is the Thorium activity, to see if it includes gamma.

For the present then, back to A/D conversion.

 

[graham-xrf]

While you have yours working, maybe cut the foil free around the sensor, and check brazil nuts, allegedly rich in K40, but with significant amounts of radium, up to 444Bq/kg. ?? Units what? That is 12nCi/kg.

I think we already junked potassium as a source for being all beta, but I read that in 10.2% of the decays, it makes Argon-40, emits a neutrino which could go through the Earth with little likelihood of being stopped, and a 1.46MeV of gamma ray. 500g of Potassium Chloride "health" salt, which is bad for you, but anyway..

K_mass = 39.0983u Cl_mass =35.453u
K share of 500g -> (35.453 * 500g)/(39.0983 + 35.453) = 262.2grams.
So - if we packed salt from the jar into a tube - I dunno, an alternative use for a condom or something, and arranged it in a torus shape under a hood funnel of lead sheet, would that ever work? 1.4MeV would let is see pretty much everything, but unfortunately, it would also bring up the lead characteristic.

Most of the numbers I can see on internet search go on about microsieverts, doses, and banana equivalents, instead of saying what the radiation type is, and it's energy.

 

[graham-xrf]

About Aluminium
I had a passing thought..
kα1 = 1.487 keV kβ1 = 1.557keV
These, if seen by the X100-7 photodiode, would yield more than 100k electrons each, so provided we are still above the noise floor, they could be seen, but not very often. The probability of either being successfully absorbed by the X1---7 is about 2%, so the counts would be low, but still a signature that would be there if the test time is long enough.

Now we get into what software does.
In calibration, we show the gadget some stuff with the element in it, like (say) ferric chloride etch granules.
The signature is 100% iron. The amplitudes will be collections of pulse height numbers from 6.4keV and 7.05keV.
Of course, our samples would would make some buckets either side to make the signature shape. I don't expect to get "single spike" buckets, but given we are accurately trying to preserve the pulse shape and duration, we might hope for better resolution than Thermino#1

Signatures.
We can build up our own set of calibration signatures. They would correspond to published signatures only on where the peaks are for energy.
So do we get too "process" the buckets counts, to compensate for the probability, so as to be able to estimate the proportions of metals in an alloy?

In effect, to scale the buckets counts to be the inverse of the one the photodiode response. (i.e. 100% minus probability).
Does this happen automatically just as a result of the calibration process? I think it might. Sometimes, I only get there slowly.


The aluminium, as you mentioned, reduces incoming X-ray intensity by 0.6%. Here we know the Al over the sensor is not participating in being hit by photons that would get it excited, it is just letting through X-rays. Also, it only lets through 99.4% of X-rays arriving, but the energy of those lucky enough to get through is not diminished at all! That's one of the nice things about quanta.

 

[homebrewed]

The aluminium, as you mentioned, reduces incoming X-ray intensity by 0.6%.

Unfortunately, for standard-weight kitchen aluminum foil it's not .6%. Exit intensity @ 6Kev = .6*(incident intensity). That's why I'm mulling over the idea of using aluminum leaf. My investigation has found that it's about 1 micron thick, which would transmit 96.9% of incident x-rays @6Kev.

Beryllium foil is available from ebay (not cheap though). Anyway, it's too thick -- there's no benefit in using it compared to much-thinner aluminum foil/leaf.

 

[graham-xrf]

Unfortunately, for standard-weight kitchen aluminum foil it's not .6%. Exit intensity @ 6Kev = .6*(incident intensity). That's why I'm mulling over the idea of using aluminum leaf. My investigation has found that it's about 1 micron thick, which would transmit 96.9% of incident x-rays @6Kev.

Beryllium foil is available from ebay (not cheap though). Anyway, it's too thick -- there's no benefit in using it compared to much-thinner aluminum foil/leaf.

I think not to worry about it. The little aluminium can in the scintillator is much thicker, and I guess they just live with it.

 

[graham-xrf]

Re: Mu-metal magnetic shielding.
This now comes about, only because I happen to be putting together some physical construction concepts, and I remember you saying the high gain + high impedance nature of the amp could pick up mains fields, and power line interference, and needs shielding.

Mu-Metal, being conductive, handily also doubles as electrostatic Faraday shield and RF block, but could be used for magnetic only, with extra copper or aluminium underneath for the electric fields. The stuff can be expensive, but one only has to look around a bit.
- - - - - - - - -
Here gets a initial flavour of what is out there.
It comes in sheets as thin as 0.002" --> 8" x 12" for $34.50 with crap review Amazon i.e.0.05mm

The better stuff is 0.1mm --> Woremor WMF200 --> 4" x 24" for $35.99 Amazon.
This one mentions specs u4=25,000, umax=100,000, Saturates at 0.55Tesla
It seems to be a mix of metals stirred up in a plastic goop, so not real mu-metal.

3ft length od 3/4" ID Flexible conduit
Probably not full annealed and hydrogen furnace treated.
--> $68.93 for for 3ft from Amazon

Genuine Mu-Metal with mixed units. Claims 0.8Tesla. $8.36 to ship to UK.
0.1mm thick, 91mm wide x 2 linear feet (609mm) --> £119.00 MS-91P Amazon
- - - - - - - - - - - -
A Shielding Calculation Formula
Calc Here --> http://mumetal.co.uk/?p=103

It can come as sheet, or ready-made treated cans --> MuCans These are sweet, but I don't know how much they cost.

We can go with or without, or leave it to individuals to raid old scopes, or get sheets from eBay.
One can take apart an old current transformer maybe.

I remember at school (physics lab) being told to roll the foil very gently, and shove it into a tube (toilet roll former), so as not to destroy it's mu value. The annealed mu value is good, but the heating in vacuum furnace in stages with low pressure hydrogen raises it by 3 orders of magnitude.
After that any unwanted excessive bending, or mechanical handling is not good for it.

The eBay's search scheme truly sucks, but I did spot one for the audiophile fraternity £13.88 + £3.09 postage from USA for 8" x 5" x 0.004" (0.1mm).

Possibly one only needs a small piece over the sensitive area, on each side of the PCB. My thought was to minimize the input loop area anyway.

A thought that occurs is - do we include a 50/60Hz deep rejection filter from the outset?
- - - - - - - - - - -
I don't actually have a clear plan for including Mu-metal shielding, but what I am trying for includes a physical layout plan that maximizes the irradiation of the target area, and having geometry to shadow the photon route to miss the sensor. The thing is used either by putting the test sample onto a sheet of lead (like builder's flashing) and placing the XRF end over it, or simply sticking the end up against the bit of metal to be tested. I wanted to include whatever we use in the way of Mu-Metal, but I am relatively ignorant about what we need.

I divide some time between the electronics and the physical stuff. I have (gulp ) not seriously tackled digging into the software yet. That sounds a bit disorganized, and it is, but I need to set down ideas as they happen, so it comes together.

I envisage the main probe is end is about 40mm diameter, and is funneled into larger sizes by using (lead sheet lined) push-on ends up to about 75mm (3"). The top limit can be anything you think you need, but the energy gets spread thinner. Even that may not matter too much if the area of the sample keeps up.

The concentrated range is about 30mm, and expands from there. The concept can allow all manner of push-on funnels and back-sheet blockers to let it go up against odd shapes (like rod stock, tool cutters, pipes, tool steels, etc), or stood over test samples, small containers of powders, liquids. The "heavy end" has lead lumps in it, so it is always stable standing on end.

I hope to shortly post some of this stuff. I was thinking that if anyone else wants to make one, they get drawings, and have to turn up the bits themselves in their workshops, including learning the technique for turning lead! So far, there is no 3D printing involved.

 

[graham-xrf]

Desoldering the diode?
Umm.. upturned electric iron when the lady of the house is out?
I dunno!

 

[homebrewed]

I think not to worry about it. The little aluminium can in the scintillator is much thicker, and I guess they just live with it.

Most scintillators seem to be made for detecting higher energy x-rays than found in our application, so somewhat thicker aluminum is OK. For example, at 60Kev aluminum's u/p value is just .278, compared to 115.3 @6Kev.

Mu-metal shielding:
Oddly enough, I just recently bought a sheet of mu-metal (but for a different reason -- I have another pie-in-the-sky project that needs really GOOD magnetic shielding). So I'm already aware of the forming issues that can affect its permeability, and why it really isn't possible to anneal it yourself. Mmm, hot hydrogen -- what fun could be had there!

Apparently even mu-metal has some remnance so if you need to keep DC magnetic fields out you need to include some de-Gaussing coils in your mechanical design.

I found some mu-metal boxes on ebay but they didn't say if they had been annealed after being formed or not....so I didn't even consider them.

De-soldering the X100-7:
Some hobbyists who regularly assemble their own circuit boards have used a cheap toaster oven + temperature controller to reflow parts. At work we used a temperature controlled hot plate to assemble/rework test boards, and it worked fine. To concentrate heat on the part we wanted to remove (or solder down), we used aluminum blocks as spacers and also to conduct heat. One of the spacers was placed so it was directly underneath the part we wanted to work on and the others were scattered around so the board would balance on them. Bottom-side components near the IC were problematic, so I always laid out my test boards so the bypass caps were on the top side.....or soldered the IC's down first, if I didn't have a choice.

The X100-7 is pretty large so I'd pre-bake it at 80C or thereabouts for at least an hour before de-soldering (or re-attaching it). If there's any moisture inside the package it will expand into steam and might break the wire bonds. I also worried about PCBs delaminating for the same reason so I always baked them before using the hot plate. I didn't worry about that when soldering smaller components like R's, C's or individual transistors.

 

[graham-xrf]

The X100-7is filled all over the die with black epoxy, but I have to assume it can stand a ramp up to solder flow temperature, and a unforced cool-down in air.

I know I am just describing in words now, when a picture would be a whole lot better, but mechanically, it would be extremely convenient if the diode were put on a small PCB only just big enough to solder onto, and the board is circular to locate in a tube. That little PCB has a ground plane on both sides, except for a clearance on the side up against the diode, to have two (very thin) tracks come toward each other, this to almost eliminate the coupling loop area.

There is also a layer of lead on back of the little PCB, stuck on with epoxy. It needs only be about 1mm thick, but if it were thicker, say 3mm, with a 0.8mm little slot across it, that provides a way to mount the main electronics PCB at right angles, again secured by epoxy. There would be two holes in the lead, to bring up the diode connections on thin solid wires up the middle of the holes in a low leakage, low capacitance fashion, to get to the board pins going to first amplifier input.

It is just a concept in my head right now. I'll try and get you a picture. It's just that quite a lot of very separate parts are suddenly coming together, and here I am still going after noise in amplifiers!

 

[homebrewed]

Well, noise could have a big impact on the energy resolution, since each pulse must be processed as a separate event -- there is no way to average pulses to improve the SNR, because you don't know the photon energy beforehand. That's why I'm hoping that a least-squares fitting scheme will improve things. However, starting out with the best SNR possible can't hurt.