Needing more than a spark test?

[graham-xrf]

For most people, even dealing with the PIN diode will be a struggle. When I designed my ELS controller PCB, I deliberately chose through hole construction so it could be DIY assembled. For these boards and the op amps that have been chosen, (for performance and availability reasons) 98%+ of the interested users couldn't assemble them. 0402's are just too small for most. It seems we agree.

I think Mark's board is an intermediary board between a modified Pocket Geiger and a Teensy4.0, as far as I know, it does not have a TIA on it for the PIN diode.

Our "supply" of the PIN diode is, at present, a ravaged Pocket Geiger, so getting them on a pre-assembled PCB is not so likely. The PIN diode itself is quite large, but again, just getting it off the board is a hassle. The think the most any of us can do is take a little hacksaw, or Dremel to it, and just cut off that piece of the Pocket Geiger. Connecting in a way that is short, very high impedance, low capacitance, and possibly magnetically shielded is also a challenge.

Given that I would end up with a diode on a hacked away chunk of Pockt Geiger anyway is where I imagined setting it at right angles to the amplifier PCB would be OK. The places where the diode solders to are at the edges, but "under" the diode body. It's just not easy to even unsolder it. I thought it may be easier to just leave it on it's board, and connect to the tracking. This stuff is not set in stone. I agree to any scheme that makes it easier for folk not experienced at all with electronic to be thinking "yes - I can do that"!

Yes - Mark's signal conditioner board is a stage between the TIA and the ADC

 

[WobblyHand]

Our "supply" of the PIN diode is, at present, a ravaged Pocket Geiger, so getting them on a pre-assembled PCB is not so likely. The PIN diode itself is quite large, but again, just getting it off the board is a hassle. The think the most any of us can do is take a little hacksaw, or Dremel to it, and just cut off that piece of the Pocket Geiger. Connecting in a way that is short, very high impedance, low capacitance, and possibly magnetically shielded is also a challenge.

Given that I would end up with a diode on a hacked away chunk of Pockt Geiger anyway is where I imagined setting it at right angles to the amplifier PCB would be OK. The places where the diode solders to are at the edges, but "under" the diode body. It's just not easy to even unsolder it. I thought it may be easier to just leave it on it's board, and connect to the tracking. This stuff is not set in stone. I agree to any scheme that makes it easier for folk not experienced at all with electronic to be thinking "yes - I can do that"!

Yes - Mark's signal conditioner board is a stage between the TIA and the ADC

Can you explain the advantage of making the PCB at a right angle to the detector? Does this help with packaging the whole thing? I could see an advantage to a single PCB. Considering the emitter ring is 65mm in diameter (my model) why can't everything sit on a common PCB? I don't think the PCB would be bigger than 65 x 130? Personally, I'd rather a single PCB, detector, all analog stuff and the processor, if the processor were a Teensy. I guess if there was an RPI on board, then I'd opt for a single PCB with detector and all analog stuff. What are your thoughts on this?

 

[WobblyHand]

Ok. Figured out a couple of things which were not obvious to me. There's a 3.7 - ? volt input. 5V seems reasonable. There's a switching regulator making ~ 13V feeding a doubler or something like that for the PD bias, and it feeds a 9V regulator. On the back side of the board are some test pads. They are labeled A, 9V, and PDV, and some corresponding ground pads. "A" is connected the the output of the analog circuit! Which is what I want! 9V is the output of the 9V regulator, and PDV is the pumped up PIN bias voltage. So... for the moment, I will connect 5V and see what happens. I just want to see if anything is alive.

Then, I have to disconnect the switcher derived PIN bias voltage, and attach a battery bias. Think I would start with two LiOn cells.

 

[graham-xrf]

Can you explain the advantage of making the PCB at a right angle to the detector? Does this help with packaging the whole thing? I could see an advantage to a single PCB. Considering the emitter ring is 65mm in diameter (my model) why can't everything sit on a common PCB? I don't think the PCB would be bigger than 65 x 130? Personally, I'd rather a single PCB, detector, all analog stuff and the processor, if the processor were a Teensy. I guess if there was an RPI on board, then I'd opt for a single PCB with detector and all analog stuff. What are your thoughts on this?

Everything was about easily contriving shielding to go around it, meaning electrostatic, electromagnetic, and radiation. If all the bits were turned, it has an axial symmetry and one would not have a PCB board area sticking out to the side, complicating the enclosures. I also wanted to include a lead shield behind the PCB. all that would leave it would be two PTFE wires going through two little holes, and then coming together to expose minimal loop area, while maintaining very low capacitance between them.

There is no reason why it has to be this way. One can have a diode planar on a PCB. It was an initial preference that came of knowing that I would likely have a PIN diode part straight onto a PCB, if I had it separate, and easily solderable. That would not be the case if I had already hacked it off a Pocket Geiger. There would be every advantage in leaving on it's little PCB, turn it through 90°, and fix it to the edge of the main PCB. The whole module would then fit straight up a tube that could be quite compact radius. It was just the first preference. I can easily be seduced into all sorts of other arrangements. My only requirement is that it addresses the features I think are necessary to work well.

Of course - if you want it to look like Flash Gordon's ray gun, you can adopt a recycled hair drier, with the insides metal plated

 

[WobblyHand]

Everything was about easily contriving shielding to go around it, meaning electrostatic, electromagnetic, and radiation. If all the bits were turned, it has an axial symmetry and one would not have a PCB board area sticking out to the side, complicating the enclosures. I also wanted to include a lead shield behind the PCB. all that would leave it would be two PTFE wires going through two little holes, and then coming together to expose minimal loop area, while maintaining very low capacitance between them.

There is no reason why it has to be this way. One can have a diode planar on a PCB. It was an initial preference that came of knowing that I would likely have a PIN diode part straight onto a PCB, if I had it separate, and easily solderable. That would not be the case if I had already hacked it off a Pocket Geiger. There would be every advantage in leaving on it's little PCB, turn it through 90°, and fix it to the edge of the main PCB. The whole module would then fit straight up a tube that could be quite compact radius. It was just the first preference. I can easily be seduced into all sorts of other arrangements. My only requirement is that it addresses the features I think are necessary to work well.

Of course - if you want it to look like Flash Gordon's ray gun, you can adopt a recycled hair drier, with the insides metal plated

Sounds sensible with respect to shielding. Although I'd rather have Flash Gordons's ray gun than a toilet paper roll!

 

[graham-xrf]

Then, I have to disconnect the switcher derived PIN bias voltage, and attach a battery bias. Think I would start with two LiOn cells.

That is the most excellent idea!
Both Mark and I disliked that switcher. Before anything else, make it stop oscillating!

6.6V bias will likely be OK. It causes PIN diode capacitance to be near 100pF, and dark current about 2nA.
Three cells would make the capacitance about 80pF, and the dark current about 2.7nA

This first capacitance it sees is what the charge from the incoming photon has to spread into, decided by the physical size of the PIN diode. It cannot have any help, and we try to get it into the TIA first device as unharmed as possible. Once it has seen the gain, we have a lot more freedom to get at it. Theoretically, the lower bias (2 cells) gets a slightly lower signal. The 2-cell bias gives 80% of the signal compared to a 3-cell bias.
The difference in dark current noise from 2nA instead of 2.7nA is not very much.

I had considered only 1-cell (3.3V) bias, but that has 170pF or so capacitance. It loses much more signal than the benefit from dark current noise reduction would have on the S/N ratio

 

[WobblyHand]

First power up. As I surmised, these test pads are connected up. A is connected to the analog output prior to the comparators. 9V read out close to 9V. And the PDV reads out about 33V, so it is a tripling circuit, as suspected. Triple would be 39V, but there's losses, so 33V it is. The analog DC value is sitting at about 3V. Next is to set up the scope to monitor the analog output.

 

[graham-xrf]

Sounds sensible with respect to shielding. Although I'd rather have Flash Gordons's ray gun than a toilet paper roll!

I had kinda thought I might have to clamp it onto a retort, if I was going to have to walk away from it for hours.
The count rates that Mark is reporting from 8 smoke detector innards is a bit of a downer.

 

[WobblyHand]

That is the most excellent idea!
Both Mark and I disliked that switcher. Before anything else, make it stop oscillating!

6.6V bias will likely be OK. It causes PIN diode capacitance to be near 100pF, and dark current about 2nA.
Three cells would make the capacitance about 80pF, and the dark current about 2.7nA

This first capacitance it sees is what the charge from the incoming photon has to spread into, decided by the physical size of the PIN diode. It cannot have any help, and we try to get it into the TIA first device as unharmed as possible. Once it has seen the gain, we have a lot more freedom to get at it. Theoretically, the lower bias (2 cells) gets a slightly lower signal. The 2-cell bias gives 80% of the signal compared to a 3-cell bias.
The difference in dark current noise from 2nA instead of 2.7nA is not very much.

I had considered only 1-cell (3.3V) bias, but that has 170pF or so capacitance. It loses much more signal than the benefit from dark current noise reduction would have on the S/N ratio

I have two 2032 coin cell holders I could use, or two 18650 cells. The 2032's seem to run around 3V, whereas the 18650's fully charged are 4.2V each. Coin cells are a lot more compact! So ~ 6V vs 8.4V. The coin cells are a bit safer. Probably use them for the bias. Might use the 8.4V for powering the unit. Got to figure out how to shutdown the switcher first.

 

[graham-xrf]

First power up. As I surmised, these test pads are connected up. A is connected to the analog output prior to the comparators. 9V read out close to 9V. And the PDV reads out about 33V, so it is a tripling circuit, as suspected. Triple would be 39V, but there's losses, so 33V it is. The analog DC value is sitting at about 3V. Next is to set up the scope to monitor the analog output.

Arguably, 33V puts the capacitance at 50pF, so getting a bit more signal. It's as much as double as one gets from 2 Li cells, but the dark current goes up to somewhere near 5nA. With a low noise front end, it should not matter. In my view, once captured as clean as possible, don't throw away the advantage.