Needing more than a spark test?

[WobblyHand]

Some toaster ovens don't get hot enough for lead free so one does have to watch out for that. If you get stencils, that's great, they are much, much better than doing it by hand. Simplified version: Line up the stencil. Squeegee the paste on, lift the stencil, place the components and pop it in the oven. In anycase, lead or lead free, don't ever use the toaster oven again for food!

 

[graham-xrf]

Some toaster ovens don't get hot enough for lead free so one does have to watch out for that. If you get stencils, that's great, they are much, much better than doing it by hand. Simplified version: Line up the stencil. Squeegee the paste on, lift the stencil, place the components and pop it in the oven. In anycase, lead or lead free, don't ever use the toaster oven again for food!

OK - a heater of some better sort has long been on the cards. Anyway, at my place, if I even look like I am about to use the oven for anything other than food, I get the steel glare from the good lady!

For the XRF experiment, I used an adapter like those shown in the picture. The large one on the left mounts quad flatpacks of all sizes, in either 0.5mm or 0.8mm pitch depending which side of the board you use. For £1.37 with free 3-day postage, it seemed obvious to take advantage.

 

[WobblyHand]

If you plan to process whole boards I'd go with an oven. Especially if you can find one in a second hand shop. I use a hot air gun for stuff like this. Solder paste is quite forgiving. You can even smear it over all the pads without being very careful. The solder melts and flows on the pads first. It tends to ball when on the FR4. If you get a short use a sharpened toothpick to brush it away. A fine blade works ok as well. Since you have a stereo microscope it will be easy. Much easier than manipulating a soldering iron, wire solder and hoping the part doesn't move!

 

[homebrewed]

A friend of mine uses 3D-printed stencil holders to align boards and stencils, and a hot air gun to solder components down. He got the stencil holder design from Thingverse.

I bought a cheap (ding and scratch) analog toaster oven from Amazon to do reflow and "light" metallurgical things like annealing brass and home-wound springs. It cost me less than $30 AND I got free shipping. I'm not a Prime member, either. Like some other (ahem) projects, the reflow oven mod is "in progress".

One thing I like about using the toaster oven is that it's relatively easy to implement temperature/time profiles with them: DIY reflow oven. A ~100C pre-bake eliminates my concerns about delaminating parts due to the so-called "popcorn effect", where absorbed moisture turns into steam inside the package, expands and breaks bond wires. Don't think this is just a theoretical possibility, it's a real issue, especially for larger IC's!

 

[homebrewed]

One thing I didn't mention about the toaster oven mod is that it uses a bang-bang controller -- basically, the toaster oven is turned on and off using a relay shield. You don't even need to replace the power cord. That's why it is necessary to use an analog toaster oven. To repurpose a digital toaster oven you would have to bypass all the electronics and directly control the heater elements. If doing so, I'd make sure to include a mechanical thermal cutout switch in case something goes wrong with either the relay control or S/W (I did much the same thing to turn a slow-cooker into a sous-vide setup) . It would be a lot easier to just find yourself an analog toaster oven.....

 

[WobblyHand]

One thing I didn't mention about the toaster oven mod is that it uses a bang-bang controller -- basically, the toaster oven is turned on and off using a relay shield. You don't even need to replace the power cord. That's why it is necessary to use an analog toaster oven. To repurpose a digital toaster oven you would have to bypass all the electronics and directly control the heater elements. If doing so, I'd make sure to include a mechanical thermal cutout switch in case something goes wrong with either the relay control or S/W (I did much the same thing to turn a slow-cooker into a sous-vide setup) . It would be a lot easier to just find yourself an analog toaster oven.....

That's why I recommend an old toaster from a 2nd hand shop. Or an old hand me down. Far more likely to be "old school" analog.
Here are the melting points of various lead free solders

So it would be good for the oven to go over 230C.

Trick I learned. If you have trouble removing a part, use a little low temp solder paste on it. It will lower the melting point and stay liquid longer. Clean up the pads with solderwick, or what ever you use, and put the normal paste down and resolder. My version of ChipQuik. A little syringe of 15 gms lasts a pretty long time. Far beyond the expiration date. The only bad thing is the needles get clogged after time. I just replace the needle. I use a 28 or 30 gauge needle, if I recall correctly. A 20 gauge will spew solder paste everywhere. A 20 or larger is fine if you are squeegeeing. For individual pads, the smaller needle is better.

 

[homebrewed]

Some progress, at least as far as the hardware goes:



This is an array of eight AM241 sources, epoxied around the aperture behind which will be the 100mm^2 detector. Yes they're ugly and canted this way and that, but since I don't have any lenses to focus the x-rays anyway they'll do . Having 8 of these things shooting 60Kev xrays out made me pretty uncomfortable, so most of the time the source plate is inside a .125 inch thick lead shield.

I milled 8 recesses in a bolt hole pattern to help place the sources. I had the thought that milling the holes deep enough, then using the tabs to tilt them toward the sample volume, would maximize the x-ray flux -- but the end result wasn't quite as nice as I'd hoped. The x-ray emission pattern is pretty broad so I don't think it is a big deal.

I debated over installing all 8, since I'm not really sure what the count rate will be -- but if it's too high I can make some lead shields to block some of the sources. This is a PROTOTYPE so by the time it is producing decent results (if ever) it could well resemble a porcupine. Or (most likely) a real Rube Goldberg assemblage.

For a sense of scale, the aluminum plate is 2.5 inches wide and 4 inches tall. It's .25" thick so the 60Kev x-rays emitted from the backs of the sources should be pretty severely attenuated -- my spreadsheet indicates that a little less than 1% of the x-rays would make it out. If that turns out to be too much, I left enough room on the other side to place another shield -- lead, aluminum, whatever. Another sheet of .25" thick aluminum would drop the intensity to 1% of 1%, i.e., .01%. In the mud.

Some might worry a bit about x-rays from the screws that are used to hold the box together and mount the latch. However, the aperture acts like a collimator to restrict the detector's field of view -- so all it should "see" is aluminum. Since the x-rays from aluminum are expected to only be in the 1Kev range, that's not a problem -- that energy range is at the very low end of the detector's sensitivity curve.

 

[rwm]

This great progress!!! Yes, given the broad angle of emission I don't think the angle is important.

 

[homebrewed]

I've also been thinking about the S/W part, that's at least as important as the mechanical bits. I have a Teensy4.0 processor board, which has pretty respectable performance.. A 600MHz clock. A 12 bit ADC, with about a 2MSPS sample rate at its maximum. It has a floating point coprocessor and also has a number of DSP functions, like 16-bit FFT's (although I don't think I will need that capability). It can be programmed using the Arduino IDE.

I have an old Dell Latitude D600 that I've been using in my "lab" for programming other Arduino boards, but I found that the old S/W setup wasn't able to support the Teensyduino environment. Mostly because its installed flavor of linux was just too old. Several of the Teensyduino functions required un-available libraries so I needed to take a different tact. The D600 is only good for a 32-bit OS, so I installed the 32-bit version of Lubuntu 18.04 and it works with Teensyduino, but the old laptop's CPU is really struggling. I think I can get this to work, but now it's time to outline the approach.

The basic idea is to monitor the digitized input stream and trigger processing when a peak signal exceeds some yet-to-be-determined threshold. A circular buffer will be needed to get a substantial number of pre-peak-value samples, in order to perform a polynomial least-squares fit to the incoming data. I think a least-squares fit scheme is a good approach because it should do a good job of rejecting noise on a per-pulse basis. Once the polynomial fit is done, the next step is to look at the goodness of fit to the data. The idea here is to determine if the pulse is the result of a single x-ray photon or that plus the overlap of a second one. The peak around a single-photon pulse can be fairly accurately modeled by a simple second order polynomial fit, but if a second photon arrives too soon the peak value won't be accurate.

From what I've seen, most simple multi-channel analyzers just use the peak value -- so things like pulse overlap (or system noise) will reduce the energy resolution of the XRF system. The Theremino approach does use some type of pulse rejection scheme, which could be much like the scheme I'm planning on using.

 

[homebrewed]

The electronics are installed, see below for an exploded view. I need to wire up a matching connector for power and routing the detector signal to my processor board. Since I'm planning on using the Teensy's USB port to transfer the data to my laptop, I didn't put it inside the box. It wouldn't be too difficult to connect up a little graphics display to the Teensy -- that would eliminate the need for a laptop. I will consider that IF this thing actually works.



The boards are mounted using a couple of aluminum pieces. I milled slots and epoxied the edge of each board into a slot. The aluminum holders are drilled/tapped (#6-32) to attach them to the base. I had to rotate my signal-conditioning board a little bit to avoid a mechanical interference problem with the connector, which is a 6-pin DIN style. It's a nice way to go because one of the pins on the back of the connector is connected to the body. Makes it easy to ground the box. This is important because the unshielded 10x10mm diode is a fantastic EMI detector.

If I get an iron peak due to the latch's mounting screw I'll use a brass screw.

There also are a couple of holes in the base that are on-axis with the hole in my x-ray source plate (shown in post #567). I will turn a post out of aluminum rod and use it to support samples. The post will be fixed in place with a screw. I'm not sure what the coverage will be from my Am241 sources so I built in a couple of options with that in mind. If the post is ~1 inch in diameter, the screw won't contribute anything to the XRF spectrum so I can use a flat-head steel screw and not worry about it.

The signal conditioning board requires +/- 10V and the Pocket Geiger needs +5. I've got enough bench supplies to run both so I should be getting some pulses pretty soon now. On to the S/W....as soon as I ohm out the connections to make certain I've got everything hooked up right. It would be a real shame to smoke the pocket geiger or signal conditioning board at this stage.