Needing more than a spark test?

[homebrewed]

Can I ask about the PIN diode selection? We have selected the device from the pocket geiger (X100-7) however there seem to be a lot of choices available. What is the main reason for choosing this? I know dark current and sensing area are important factors. Could there be a better and cheaper choice? There are some diodes with lower dark currents but they seem to have smaller sensing areas.

There are several reasons why we're concentrating on PIN diodes. The silicon detectors used in high-end x-ray analysis systems are PIN types. The large intrinsic region (the "I" part) increases the detection volume, which should improve the energy resolution of the detector -- if an incoming xray photon isn't completely absorbed by the silicon, it generates a lower-amplitude pulse so the spectrum is smeared out . So we want the depletion region as wide as possible, best achieved with a PIN diode. And finally, since the depletion region is pretty wide compared to a plain-jane PN junction, the capacitance is minimized.

Dark current is going to be proportional to the sensor area because the current is due to thermally generated carriers -- more area, more carriers.

It would be instructive to see how a large-area _non_ PIN style photodetector performs, As you've noticed, there are many less-expensive photodiodes that are not PIN diodes. Reverse-biasing them to close to their breakdown voltage would give you a wide-as-possible depletion region: but at the expense of high dark current.

FWIW, cooling the detector with a thermoelectric cooler would reduce the dark current (regardless of what diode technology is in use). I note that the teardown of that XRF "gun" revealed something that looked like a heat pipe going to the detector. A photodiode won't dissipate enough current to go above ambient so it seems likely that the detector is being cooled. I have a TE cooler I could use to see if it makes much of a difference but unfortunately there currently isn't enough room in my enclosure for it -- the spacing between the pocketgeiger board and my signal conditioning board is too small. I can move things around but would need to slightly modify the enclosure for that. Not up to attempting a homebrew heat pipe just yet .

I have noticed that some vendors offer two different types of detectors -- conductive and voltage-output types. The conductive detectors are reverse-biased so incoming photons generate a current pulse, while the voltage-output types are more like solar cells. Since they are NOT reverse biased their junction capacitance is very high compared to the other type. For our purposes I think we want to concentrate on the types that are designed to be reverse-biased. They'd feed into a TIA or load resistor.

 

[WobblyHand]

Not up to attempting a homebrew heat pipe just yet .

They are not as hard to make as you think. I made one out of some soft copper tubing, some absorbent paper towel and some freon. Might have been luck, but it worked spectacularly well. The heat transfer was amazingly fast. I compared it with some tubing of the same size. The transfer of the normal pipe was what we expect - relatively slow. With the heat pipe it was in a fraction of a second to equalize the heat over a 8" distance. The tube was sealed to prevent the freon from escaping. I squirted in freon liquid and bent the tube over to seal it. Beats me what one would use today, probably something more environmentally benign. Grabbing one end of the tube made the other end the same temperature, practically instantly. Was pretty neat to see as a teenager. The paper towel was a folded to form a wick that extended the length of the tube. As long as there was some liquid freon in the tube it worked.

 

[homebrewed]

Thought so, but I didn't have access to the backside of your board, only the front. It was hard to tell how it was connected from the picture.

The regulator actually is on the top of the board, labeled as "LDO1". But it IS routed to ODVDD on the back side of the board.

I did a small rev on the board, replacing "LDO1" with the actual devuce P/N but I did that after ordering the boards.

I've attached a PDF of the schematic that will hopefully clarify things. The design pretty much duplicates what's shown on the data sheet so I don't think there's much point in being secretive about it. The most significant variation from the DS is how I separated the analog and digital supplies, sort of doing a star connection rather than a daisy chain. That decision was totally based on my concerns regarding noise coming in from the USB 5V line. I figured a little more low-pass filtering on BOTH supplies would be a Good Thing.

 

[homebrewed]

They are not as hard to make as you think. I made one out of some soft copper tubing, some absorbent paper towel and some freon. Might have been luck, but it worked spectacularly well. The heat transfer was amazingly fast. I compared it with some tubing of the same size. The transfer of the normal pipe was what we expect - relatively slow. With the heat pipe it was in a fraction of a second to equalize the heat over a 8" distance. The tube was sealed to prevent the freon from escaping. I squirted in freon liquid and bent the tube over to seal it. Beats me what one would use today, probably something more environmentally benign. Grabbing one end of the tube made the other end the same temperature, practically instantly. Was pretty neat to see as a teenager. The paper towel was a folded to form a wick that extended the length of the tube. As long as there was some liquid freon in the tube it worked.

Cool! <pun> and also, just COOL!

 

[rwm]

"With the heat pipe it was in a fraction of a second to equalize the heat over a 8" distance."
Amazing. I had no idea about this.
TE cooler = Peltier?
Hey a 5" heat pipe is $3.86 at DigiKey

 

[WobblyHand]

"With the heat pipe it was in a fraction of a second to equalize the heat over a 8" distance."
Amazing. I had no idea about this.
TE cooler = Peltier?
Hey a 5" heat pipe is $3.86 at DigiKey

At that price, a heat pipe can't be all that complicated... My heat pipe was soft copper tubing, a wick, and Freon. The hardest part was soldering the tubing and getting it leak free. Gas molecules have rather high velocities and within the device pretty quickly come to equilibrium. I suspect any refrigerant like material could be used. The key thing is there is some liquid under pressure. The liquid is the reservoir, but the vapor is doing the bulk of the heat transfer, not the physical pipe. Don't need much liquid, too much will prevent efficient heat transfer.

Yes, a TE cooler is usually a Peltier (or stacked Peltier) device.

 

[homebrewed]

According to information I found here, acetone would be a good working fluid for a heat pipe in the -48 to +125C range. The heat pipes Digikey stocks appear to use water as a working fluid, so the low-temperature end wouldn't be any lower than 1 degree (also according to the table shown on that web page).

Cooling the detector down below the dew point would be problematic from a condensation viewpoint. To get lower, it would be necessary to enclose the detector so it can be in a dry atmosphere. One approach: use two TE coolers, one attached to the detector and one with just its cold side exposed to the interior of the enclosure and make sure that the second cooler is colder than the other. It will scavenge the water. No need to carry around a bottle of compressed CDA or nitrogen; but it won't be an instant-on measurement. And TE coolers are kind of power hungry if we're looking at a portable battery powered system.

A cold finger with dry ice in it would do the trick as well but that introduces a consumable.

Detectors using wide-bandgap semiconductors like CZT circumvent the need for cooling but it looks like they are WAY out of our reach, moneywise. One web site indicated that CZT crystals cost on the order of $2000/cc (cubic centimeter), and that is just for the raw material! I'd look around for an SrI scintillator first, even if that means a PMT or SiPM. PbI is supposed to be a good wide bandgap semiconductor for that kind of thing but is out of favor due to its toxicity. Although cadmium and tellurium aren't much better -- bad press is everything, eh?

Anyway, as in many other discussions we're getting a bit ahead of ourselves.

 

[WobblyHand]

According to information I found here, acetone would be a good working fluid for a heat pipe in the -48 to +125C range. The heat pipes Digikey stocks appear to use water as a working fluid, so the low-temperature end wouldn't be any lower than 1 degree (also according to the table shown on that web page).

Cooling the detector down below the dew point would be problematic from a condensation viewpoint. To get lower, it would be necessary to enclose the detector so it can be in a dry atmosphere. One approach: use two TE coolers, one attached to the detector and one with just its cold side exposed to the interior of the enclosure and make sure that the second cooler is colder than the other. It will scavenge the water. No need to carry around a bottle of compressed CDA or nitrogen; but it won't be an instant-on measurement. And TE coolers are kind of power hungry if we're looking at a portable battery powered system.

A cold finger with dry ice in it would do the trick as well but that introduces a consumable.

Detectors using wide-bandgap semiconductors like CZT circumvent the need for cooling but it looks like they are WAY out of our reach, moneywise. One web site indicated that CZT crystals cost on the order of $2000/cc (cubic centimeter), and that is just for the raw material! I'd look around for an SrI scintillator first, even if that means a PMT or SiPM. PbI is supposed to be a good wide bandgap semiconductor for that kind of thing but is out of favor due to its toxicity. Although cadmium and tellurium aren't much better -- bad press is everything, eh?

Anyway, as in many other discussions we're getting a bit ahead of ourselves.

Just getting everything to work, from end to end, would be a major accomplishment. No need to tweak until that happens in my opinion. Active cooling will make this a lab bench curiosity. I'm looking for something that one could possibly carry into the field, not a bench warmer!

Sometime, I might make another heat pipe with acetone. You know, for the heck of it...

 

[rwm]

Interesting. I may fool around with acetone, stainless tubing, silver solder and a torch.
Regarding diode cooling...would the temperature need to be very constant or just the colder the better? I was thinking about something like a freezer pack linked to a heat pipe. The freezer pack would not really be consumable but would need an occasional recharge. Maybe make a custom pack the size of a D cell with the pipe down the middle?

 

[homebrewed]

Interesting. I may fool around with acetone, stainless tubing, silver solder and a torch.
Regarding diode cooling...would the temperature need to be very constant or just the colder the better? I was thinking about something like a freezer pack linked to a heat pipe. The freezer pack would not really be consumable but would need an occasional recharge. Maybe make a custom pack the size of a D cell with the pipe down the middle?

Thermally generated leakage current will have a fairly large temperature coefficient. But the input to my signal conditioning board is AC coupled so slow shifts in the TIA output level due to temperature variations wouldn't cause pulse amplitudes to move around much. The main impact would be w/regard to the SNR, something a bit more difficult to evaluate.

I'm thinking that getting more samples to integrate under over the length of our pulses will have a larger impact on the system SNR in terms of bang-for-the-buck approaches. But, since the SNR improvement should be proportional to sqrt(#samples), the SNR improvement becomes less and less significant as we increase the number of samples. This is not all bad -- it means that, for better or worse, throwing really expensive (fast & high resolution) ADC's at the problems won't pencil out all that well so there's no point in going there.