Needing more than a spark test?

[homebrewed]

Teensy 4.1 is apparently £34.99 ($42.47) from Amazon.co.uk.
Is Adafruit the place to be getting these?
The "out of stock notice" looks similar to when searching for Raspberry Pi(s)

Your situation likely is different compared to mine. The PJRC web site is selling their T4.1 w/o ethernet for $29.60 and has them in stock. But since you ran into some problems when buying the pocketgeiger from Sparkfun you might encounter similar issues when ordering a teensy from them.

PJRC's "corporate headquarters" actually are in Portland. I could drive there in about 30 minutes if they have a retail outlet.

If you are really after a Linux SBC you might want to take a look at offerings from Hardkernel. They have some pretty good stuff but it's going to be more expensive than an RPi. Last I checked their 8-core SBC was running about $53, and that's without a memory card, which you need for the OS. They have a bunch of GPIO's brought out to connector pins but I can't say whether or not they're siimilar enough to the RPi to meet your requirements. The 8 cores are divided into 4 A15's and 4 A7's. System clock is 2GHz.

I'm using one of their XU4's as a media server and it works a treat for that. I have a couple of their lower performance boards as well, one is being used as a print server for a photo-quality inkjet printer and the other is a DIY call blocker. It listens to the incoming phone line and if the phone number or caller ID are on the blacklist it picks up the call and then immediately drops it.

 

[WobblyHand]

If you look up PJRC store, I think you can find local stocking dealers for PJRC products in the UK. There's around 3-4 stockists, ie authorized companies. Others may mark up the prices more. Adafruit stocks in the US, to the best of my knowledge. At least in the US, their Teensy pricing was ok. When Teensy 4.1's were of limited availability, I bought one off Amazon, and it was easily the most expensive one I bought. (Have 3 of them.). PJRC now has them in stock, so I just order them directly. Might order one and a display just for this application in the next few weeks.

 

[graham-xrf]

PJRC's "corporate headquarters" actually are in Portland. I could drive there in about 30 minutes if they have a retail outlet.

I thank you kindly for the generous suggestion, but I would not wish to put you to such trouble on my account.
Getting hold of one from a nearby store may be easy enough, but then getting it all the way to me is a whole ridiculous nonsense that can easily end up costing much more than the item itself. It should be "post-able" in a padded envelope as a very light item, but because the value is more than a few pounds, it hits a wall of paperwork and fees that were the last ditch admin rapid setup after a Brexit débâcle.

If nobody else says it, then I will. We were promised the bright trading future, both with European nations, and US /Asia / Pacific Australasia. The reality sucks, and I fully account the effects of COVID and supply chain problems in saying this. At the present, stuff that arrives in container ships at minimal cost because of bulk scale loading, mostly from China /Taiwan, has increased in cost a bit, but not markedly, with certain major exceptions. Everything else is a import organizational mess!

If you are really after a Linux SBC you might want to take a look at offerings from Hardkernel. They have some pretty good stuff but it's going to be more expensive than an RPi. Last I checked their 8-core SBC was running about $53, and that's without a memory card, which you need for the OS. They have a bunch of GPIO's brought out to connector pins but I can't say whether or not they're siimilar enough to the RPi to meet your requirements. The 8 cores are divided into 4 A15's and 4 A7's. System clock is 2GHz.

I had in mind to get the PCB done with tracking to allow as much experimenting as possible. The circuit would use links to be able to have the low noise TIA, with ability to change components to set various gains. have an ADC - or not, Have a window comparator trigger available - or not. Have deep notch 50/60Hz rejection - or not. The plan is to have three button cell lithium battery mountings. One can have 3.3V, or 6.6V or 9.9V for low noise bias sans switching regulators.

I plan to have the PCB made as as many as I can fit on the standard minimum board (12" x 12") on step-and repeat. Then I can post them off to the members that want one, and feel able to get some components on.

I thought to have ready-made connection areas to mount Raspberry Pi Zero-2W.
That same GPIO can also mount any other Raspberry Pi, either direct on header, or cabled to a separate mounted Raspberry Pi.
Then also, connections for a Teensy 4.1, or perhaps some connection setup to best suit Teensy of several sorts, if compatible.
Allowing for the more available Beaglebone was also in mind.
I don't yet know the kit from Hardkernel, but I see ODROID C4 at $54, and an apparently much higher performing ODROID-XU4 which comes standard with mounted fan for a special offer of $53.

These are my "I would like it to be" notions. Right now, I want to first continue the first build to test and prove the TIA analog stuff, and capture pulses. I don't yet have the mechanical mounting for the smoke detector sources as envisaged in the first CAD model. I was thinking to go for something simpler and easier to make, with the sources facing straight forwards.

I'm using one of their XU4's as a media server and it works a treat for that. I have a couple of their lower performance boards as well, one is being used as a print server for a photo-quality inkjet printer and the other is a DIY call blocker. It listens to the incoming phone line and if the phone number or caller ID are on the blacklist it picks up the call and then immediately drops it.

I have had a Raspberry Pi 4 hooked to 500GB SSD overkill working as my mailserver for some years now. It can also do webserver, but the space is mostly useful to let folk download big files unsuitable for email, or display the videos I put there on a browser. The network it is on is only about a metre from the fibre-optic modem.

 

[graham-xrf]

odroid.co.uk -> https://www.odroid.co.uk/
Check out the C4 with no frills for €85.64
So that would be $90.97.
I can't discover the shipping cost without going almost all the way to buying one from the cart.

We know by now the SBC on the end is not really the problem. All sorts have the power to do the job, and folk can use whatever they feel happier with, or can get in at reasonable cost, or maybe just happen to have handy. The real preforming stuff that needs serious attention is all about the TIA, and pulse capture, and getting working mechanical mountings together.

In passing, have any other candidates for the photo-diode been seen? Are we still into buying a Pocket Geiger, and robbing / re-working it's innards?

 

[homebrewed]

In passing, have any other candidates for the photo-diode been seen? Are we still into buying a Pocket Geiger, and robbing / re-working it's innards?

I have found a few other silicon PIN diodes that might be suitable. They're not as large in area, but plusses and minuses there in terms of performance. Less capacitance but less detection area. One example from Advanced Photonix: the PDB-C107. 17mm^2 area, $27.93 from Digikey-USA. The detector is protected by clear epoxy so it might be suitable for detecting low energy Xrays. Thorlabs has their FDS1010 photo detector. It is 100mm^2 and costs about $60 (but it's not clear if it is a PIN diode or not). They also have a 13mm^ photodiode for even less but, again, the data sheet says nothing about the diode technology.

The latter is in a metal can that might need to be de-lidded so lower energy Xrays can find their way to the diode. That sounds intimidating but it actually isn't that big a deal if one has a lathe -- make yourself a specialized collet to grip the body and use a sharp tool to cut off the lid. I did it all the time when I was de-lidding optical receivers at work. In that case they contained TIAs my company had made & sold as dice and I needed to analyze them to figure out how (or IF) they had failed.

The downside to all repurposed photo detectors is similar to PMTs -- they have to be shielded from ambient light. But that might be accomplished by using black silicon RTV or 2-part epoxy with black pigment in it. As @RJSakowski has pointed out, carbon doesn't absorb Xrays very strongly so black plastic wouldn't necessarily be a deal killer.

If I was in the mood to buy a different photo diode I probably would look hard at the SiPM offerings from ON Semi and couple a CsI(tl) or NaI scintillator to it. But the latter, due to their hygroscopic nature, are typically packaged in a manner that negates their use for detecting low-energy Xrays.

Digikey sells their ~9mm^2 SiPM devices for about $56, in an SMT style package. They probably won't be as quiet as an old-school PMT but the power supply requirements are greatly simplified.

First things first. Let's see if the pocketgeiger detector can do for this.

 

[graham-xrf]

I have found a few other silicon PIN diodes that might be suitable. They're not as large in area, but plusses and minuses there in terms of performance. Less capacitance but less detection area. One example from Advanced Photonix: the PDB-C107. 17mm^2 area, $27.93 from Digikey-USA. The detector is protected by clear epoxy so it might be suitable for detecting low energy Xrays. Thorlabs has their FDS1010 photo detector. It is 100mm^2 and costs about $60 (but it's not clear if it is a PIN diode or not). They also have a 13mm^ photodiode for even less but, again, the data sheet says nothing about the diode technology.

OK - thanks for doing the digging. I will have a look. "About $60" is OK, because, at least for me, it puts it about level with the Pocket Geiger as a source for the main detection item. Even if all else about the Pocket Geiger is not so useful, and an expense, at least one gets a thing mounted on a little PCB, with ways to connect to it, and power it. Interesting that you worked for a company that made TIAs

The downside to all repurposed photo detectors is similar to PMTs -- they have to be shielded from ambient light. But that might be accomplished by using black silicon RTV or 2-part epoxy with black pigment in it. As @RJSakowski has pointed out, carbon doesn't absorb Xrays very strongly so black plastic wouldn't necessarily be a deal killer.

The basic design of our envisaged kit to put the material in the way of some Am241 excitation does exclude other light anyway, so using a "bare naked" photodiode need not be a problem.

First things first. Let's see if the pocketgeiger detector can do for this.

Absolutely agreed! It's affordable, and everything we have come to know about it has not shown up any reason it would not work.

 

[graham-xrf]

Re: Teensy ADC
It's me just getting into checking out Teensys, looking at pinouts

At PJRC site, the Teensy comparison page shows Teensy 3.2 through Teensy 3.6 having 13 bits usable resolution, and the much faster Teensy 4.0 and 4.1 as having 10 bits usable resolution. I take it that Teensy 3.x series is what we are into if the internal ADC is intended.

 

[WobblyHand]

Re: Teensy ADC
It's me just getting into checking out Teensys, looking at pinouts

At PJRC site, the Teensy comparison page shows Teensy 3.2 through Teensy 3.6 having 13 bits usable resolution, and the much faster Teensy 4.0 and 4.1 as having 10 bits usable resolution. I take it that Teensy 3.x series is what we are into if the internal ADC is intended.

Yes and no. The Teensy 3.x are very rare due to IC shortages. NXP isn't making them available to PJRC, and isn't making any more chips available for about another half year. PJRC had a small batch of 3.x and they sold out very quickly just recently. I think the 3.5, 3.6 are M4 devices.

I think you can get some M4 based chips maybe from Adafruit. They used to have a Feather M4 Express that was nice. Made my Doppler chronograph with one. Personally, I started out with Adafruit, but in some ways, the Teensy libraries are more advanced. But Adafruit supports a lot of devices. Either is decent.

 

[graham-xrf]

Yes and no. The Teensy 3.x are very rare due to IC shortages. NXP isn't making them available to PJRC, and isn't making any more chips available for about another half year. PJRC had a small batch of 3.x and they sold out very quickly just recently. I think the 3.5, 3.6 are M4 devices.

I think you can get some M4 based chips maybe from Adafruit. They used to have a Feather M4 Express that was nice. Made my Doppler chronograph with one. Personally, I started out with Adafruit, but in some ways, the Teensy libraries are more advanced. But Adafruit supports a lot of devices. Either is decent.

OK, and thanks. I get it now
The support for easy connection to the computing part for most devices folk might want to play with is all I was trying for right now.
For the Raspberry Pi(s), I envisaged using the little Raspberry Pi Zero - 2W stood on the same PCB as daughter board, but that one is in the same kind of supply situation as the Teensy(s). For now, it's perfectly OK for me to just hook up one of the boards I already have. It seems OK with an older Raspberry Pi 3B.

It occurs to me that some might want to use the 600MHz Teensy 4.1, but using the separate 16-bit ADC. I think the internal ADC on Teensy 4.x 10 bits worth of usable resolution is not enough, but folk who like Teensys might want to stay with them.

I have bought the AD7622, but this choice need not be set in stone. The price has now gone up to near £50 (£61). Before I put too much more on my KiCAD circuit, I am going to have another last look at available ADC choices, with a view to finding something lower cost, yet still "good enough".

[Edit: Hmm.. is the plural of "Teensy" called "Teensies" ? ]

 

[homebrewed]

I recently ordered an AD7667 to play with in the LQFP package. It is a 1MSPS version and looks pretty similar to the AD7622, functionally speaking. One thing I like about these is the fact that they can be pipelined to reduce the effective acquisition overhead. If using it in the parallel mode, you can quickly read out the result of the previous conversion command, and that also triggers the ADC to start another conversion. So no waiting around for the converter to go through its SAR business. In addition, the byteswap pin allows you to mux the upper 8 and lower 8 bits. This makes it possible to use the T4.0, because its board was designed so there are only 8 bits available for parallel I/O (and there is some bit shifting needed as well). It will be a little slower than the T4.1 because it can handle a full 16 bits in one read cycle. The data isn't going to be on "nice" 16 bit boundaries so there will be some shifting required, but at 600MHz that won't take very long!

The AD7667 cost me $33.65 from Digikey, not counting shipping.