Needing more than a spark test?

[WobblyHand]

My LTSpice is LTspice XVII(x64) version (17.0.35.0)
When I first downloaded it as .exe I think, linux Mint automatically popped up and gave me the choice to have the Wine install happen.
When updates are ready, LTSpice just brings the pop-up to let one do the update.
It may be that before you do anything with LTSPIce updates, you do the update for Pop!OS

The download should complete. I am not sure what is happening there.
The text file attached in post #1231 has a bunch of YouTube links, which is where I learned the basic ways this can work.

One can actually drop an entire model file into the circuit drawing as a "comment", and have the simulation use it, because it ends up in the netlist.
This is yuk! It takes up lots of space, and clutters everything.

The exact same result can be had with .inc to "include" it, as if it was in the drawing.
Then, the way to make using it easy, is to use it as a .lib file.
I did not make the .lib SPICE directive directly. I started with a .inc include.
The .lib statement got made automatically when I clicked on the correct button choice.

I am suspicious of the update not completing. It may be better to get your previous designs stashed somewhere, and delete all trace .of LTSpice, and load up again. This means losing the c_drive in wine, and then also removing the LTSpice folder in /hjome/bruce/Documents. That said, I am wary, because that seems unnecessary if you can already make an old filter simulate.

I have 17.0.36.0 now. Installing wine by itself showed some ntlm_auth issues. Eventually I installed winbind, which should have fixed that.
So I think the wine install of LTspice did not go smoothly. However, I'm not sure how to uninstall LTspice - even after saving all my old sim files. How does one break the symlink without messing other stuff up?

 

[graham-xrf]

I have 17.0.36.0 now. Installing wine by itself showed some ntlm_auth issues. Eventually I installed winbind, which should have fixed that.
So I think the wine install of LTspice did not go smoothly. However, I'm not sure how to uninstall LTspice - even after saving all my old sim files. How does one break the symlink without messing other stuff up?

You can make the stuff in .wine disappear by deleting the entire folder /home/bruce/.wine/drive_c/Program Files/LTC
Then have a go at /home/graham/.wine/drive_c/users/bruce/Documents/LTspiceXVII
There should be an examples folder, and a lib folder in there.

If you have not been saving anything of your own models into them, or have always done your own stuff from a completely different folder, then there is no harm in having a new LTSpice install re-create them, with all the latest library. Just remove the whole LTSpiceXVII folder.

Now, you should have a machine that knows no LTSpice.
One hopes you have a 64-bit machine.
There is a LTSpice called LTSpice XVII for Windows (End of support)
Perversely, this may be the "good" one. The "other" choice is for Windows 10 64-bit and forward.

I don't know how Wine manages Windows 10 stuff, Windows is famous for continuously making itself versions incompatible, even with itself, and often needs software that runs on it to have some kind of relationship with Windows "telemetry". It may be that is not the one to use, and we don't give two hoots that Microsoft will not "support" LTSpice in that way.

In the end, we don't even need LTSpice. We can use ngspice, or a number of other Linux based circuit simulation CAD that can read and use all the models. It's just that LTSpice happens to have a huge collection of analogue models in one place. We are not beyond simply saving the LTSpice downloaded library, and letting ngspice use it.

That all said, I think the best choice is the 17.0 (End of support) version, unless we can be sure the Windows 10 version cvqan run under Wine. I will be having a quick search about that.

Anyway, before you go there, we check whether you want a clean new LTSpice. I am still thinking maybe no, just because it is "working", and it is a later version than mine.

Dark thought! - I could trash my LTSpice, and attempt to install the latest Windows 10 version, just to see if it will.

 

[graham-xrf]

@WobblyHand
I just tried the LTSpice sync update. A whole bunch of devices is going into the library. This should be interesting.
It is now the same version as yours.
Once I set up the Voltage and Current tolerances, It works on the LMC662C.asc file

 

[graham-xrf]

SmokeZilla-241 ??

This is a trail sketch to find the geometry when sources are placed zig-zag, and all shielding is 2mm.
The central shield fits around the PIN photodiode body with 0.52mm (0.0205") clearance at the diagonal corners
The sources are standard 6.35mm with 0.5mm clearance to the little cylinder shield they are placed in (glued).
The circles you see are all supposed to be shielding buckets. There is not "tilt" on this one (yet).
Exploring the necessary overall size, the outermost circles are 2mm lead shield.
One has to add the outer body beyond all this, plastic or aluminum around the whole lot.
Near 81mm is the final OD of the outer shield.
Add the containing body wall, say 2mm thick, we end up with 85mm total (3.35").
That is a bit fat, but with a handle swiped from a Stanley No. 4 , it might justify a zap sound effect!



Maybe with 8 sources, one might take the trouble to make up a series of little pots for source shields, but I think just lose these extra little parts. Make the whole inner part out of lead, and just drill some holes for the sources. Also, without the need for supporting material between cylinder shield pots, one can move everything a whole lot closer, to get a much more compact thing. In the total concept, the PIN diode can only ever see what has either hit lead, or the material under test. We can make it that it only ever sees the material under test, provided only that material is under the aperture.

For when one has only a tiny piece of test material, then I guess the only practical thing to do is put the stuff down on a lead sheet, and accept the count might show some 10.5KeV and 12.6KeV that does not belong to it.

 

[WobblyHand]

SmokeZilla-241 ??

This is a trail sketch to find the geometry when sources are placed zig-zag, and all shielding is 2mm.
The central shield fits around the PIN photodiode body with 0.52mm (0.0205") clearance at the diagonal corners
The sources are standard 6.35mm with 0.5mm clearance to the little cylinder shield they are placed in (glued).
The circles you see are all supposed to be shielding buckets. There is not "tilt" on this one (yet).
Exploring the necessary overall size, the outermost circles are 2mm lead shield.
One has to add the outer body beyond all this, plastic or aluminum around the whole lot.
Near 81mm is the final OD of the outer shield.
Add the containing body wall, say 2mm thick, we end up with 85mm total (3.35").
That is a bit fat, but with a handle swiped from a Stanley No. 4 , it might justify a zap sound effect!

View attachment 434452

Maybe with 8 sources, one might take the trouble to make up a series of little pots for source shields, but I think just lose these extra little parts. Make the whole inner part out of lead, and just drill some holes for the sources. Also, without the need for supporting material between cylinder shield pots, one can move everything a whole lot closer, to get a much more compact thing. In the total concept, the PIN diode can only ever see what has either hit lead, ot the material under test. We can make it that it only ever sees the material under test, provided only that material is under the aperture.

For when one has only a tiny piece of test material, then I guess the only practical thing to do is put the stuff down on a lead sheet, and accept the count might show some 10.5KeV and 12.6KeV that does not belong to it.

Seems you are having fun. I think 6 or 8 sources is enough, at least as a design goal. You are making one heck of an emitter there. We'd want that pretty well enclosed. 16 emitters do increase your exposure... And cost. But it is an interesting exercise to see how one can fit it all in.

I wouldn't want to machine that much lead, chances of a mess up increase with every operation/hole. Just turning copper in a lathe gave me fits. Everything was fine, until it wasn't. Saved it, but it ended up kind of an ugly piece. Lead sound even more fun.

 

[homebrewed]

Seems you are having fun. I think 6 or 8 sources is enough, at least as a design goal. You are making one heck of an emitter there. We'd want that pretty well enclosed. 16 emitters do increase your exposure... And cost. But it is an interesting exercise to see how one can fit it all in.

I wouldn't want to machine that much lead, chances of a mess up increase with every operation/hole. Just turning copper in a lathe gave me fits. Everything was fine, until it wasn't. Saved it, but it ended up kind of an ugly piece. Lead sound even more fun.

If the main point behind milling pockets into lead is to make it easier to arrange the sources, I'd be inclined to make a template out of thin plastic sheet. Punch or drill holes in it with the desired pattern. A thin layer of epoxy would be transparent to the 60Kev xrays so there isn't a downside to just leaving the template in place. Or, for those who have one, a relatively thin template that's been 3D printed should work equally well -- if it can be released from the printer platen without damaging it.

 

[graham-xrf]

Seems you are having fun. I think 6 or 8 sources is enough, at least as a design goal. You are making one heck of an emitter there. We'd want that pretty well enclosed. 16 emitters do increase your exposure... And cost. But it is an interesting exercise to see how one can fit it all in.

I wouldn't want to machine that much lead, chances of a mess up increase with every operation/hole. Just turning copper in a lathe gave me fits. Everything was fine, until it wasn't. Saved it, but it ended up kind of an ugly piece. Lead sound even more fun.

I think 8 sources will do. I don't think they would be a hazard. Even if you walked around with 100 or so in your pockets.

Regarding turning goopy soft material, like copper. It can be done, and to a good finish. As I understand it, lead is way easier. It also needs the right attention to cutting tool, and technique, but I think the original motivation for adding lead to metal to make it "free machining alloy" was because lead had the desired machining ease. The main issues are supporting it well. Folk will make mandrels, or even use wooden dowel.

 

[graham-xrf]

If the main point behind milling pockets into lead is to make it easier to arrange the sources, I'd be inclined to make a template out of thin plastic sheet. Punch or drill holes in it with the desired pattern. A thin layer of epoxy would be transparent to the 60Kev xrays so there isn't a downside to just leaving the template in place. Or, for those who have one, a relatively thin template that's been 3D printed should work equally well -- if it can be released from the printer platen without damaging it.

In my concept, it is pretty uncompromising. I want to have as few goes at it as possible, so I seek to ensure stuff we do not want to happen becomes impossible by design.

A plastic shape.
This is one that can be 3D-printed, but also could be made of aluminium. The achieves it's first shielding by pressing into place a 2mm wall cylinder that goes around the PIN diode body, and extends slightly above it's package to deny incoming photons from all but a defined target area angle. My concept is 8 sources, arranged in the original tilted manner. Folk can vary this to suit what they fancy.

The sources shields
These are little cylindrical containers, also pressed into the locating holes on the carrier shape. This need not be deep, or they may even simply be glued in place. I favour the holes deep enough to securely hold them, but they stick up the right amount to deny travel to any photons not destined for the defined test area, so nothing sideways, nor backwards out of their little pots. The walls of the pots are 2mm thick. The base can be a bit thicker.

If I would be mounting a PIN diode on my own PCB, it gets a rectangle of lead sheet underneath it as well.

The inner sides of the main cover.
Again, lined with 2mm or so lead. It need not be a turned cylinder, It can be lead sheet curled up by wrapping it around a beer can, and then fixed in place inside the end of the body tube. That tube can again be plastic, a salvaged piece of smaller guttering, or a bit of drain PVC from the DIY. The tube can, of course, be quality aluminium, or painted up cardboard shipping tube. It does not matter, because everything inside will not let the photons go anywhere except where they must. Having it all metal from the start does help with the electric field shielding, but there will be internal screening anyway.

Magnetic ingress.
I have not yet thought all this part through, but the intention is to exclude these. The 60Hz state-variable filter is optional, but is in the plan to have a place on the PCB.

"Smokezilla-241" only happened because I had already defined the middle circle of sources shields. A rotated version, with bigger circle, was easy to put down on it. It would probably work OK, but I am not sure such a lot of complexity would be worth it. Every 3dB worth of extra extra energy gathered requires doubling the number of sources. We already have two doublings to get to 8. I think it will be enough.

Yes to the template!
Template, or scribed markings and take aim with the drill, should be OK. This part does not have to be beautiful precision machined. We only want to locate some sources. Taking care with a hand drill might even do. This is something where a touch of hot melt glue here and there might be OK. Making a template to guide a drilling operation is just practical good thing to be doing.

 

[WobblyHand]

If the main point behind milling pockets into lead is to make it easier to arrange the sources, I'd be inclined to make a template out of thin plastic sheet. Punch or drill holes in it with the desired pattern. A thin layer of epoxy would be transparent to the 60Kev xrays so there isn't a downside to just leaving the template in place. Or, for those who have one, a relatively thin template that's been 3D printed should work equally well -- if it can be released from the printer platen without damaging it.

I print on a thin coated steel sheet. Post print, I just remove the sheet and flex it, and the object pops off. Not all that difficult, if you let the sheet cool to ambient.

@graham-xrf, here's a rough model of the source ring, roughly to one of your early drawings. Had a bit of trouble making pockets that were tangent to the surface, but more or less solved that using a datum plane. Not sure I got it right, but that's ok. I can widen and deepen the pockets to accommodate the lead buckets that the sources sit in. Might simply make a lead ring around the center.

You can see the big pockets. They are 10.4mm wide and 4mm deep. These pockets can be stamped or cast. They are basically cups. The green can be printed. Don't know how I'd do the ring yet. Cast it? I don't have the ability to extrude it. Seems wasteful to drill it. Maybe wrap foil and clamp it?

 

[WobblyHand]

For what it is worth, I am printing this. It will take an hour at the settings I chose. Sometimes having something in your hand is far more useful than staring at a screen. Pretty easy to pick up problems in the real thing that are not so obvious in CAD. Thinking about a punch and die to form the lead cups. Shouldn't be hard to make that on a lathe.
Edit: here it is. 64mm wide, 10mm high, 8 sources. Some stringing, but will clean up easily. 10.4mm wide source pockets at 55 degrees, 4mm deep. Came out ok for a draft print.