Component choices

A project log for Theremino PMT 3.3 based radiation detector

Theremino "the modular in out" software, PMT driver and detector with USB sound card analog in.

BadgerBadgerMushroomBadgerBadgerMushroom 03/07/2015 at 16:560 Comments

See the weird "bent" component unfocused at the back of the board? Technically, that was my failure in reading the schematics and thinking those 4 x 1Mohm resistors were all identical SMD components. Turns out the first one is meant to be a "normal" barrel type. I like the end-result though, so I'm going to claim I always planned to use one of those high-quality 1kV SMD resistors there.

It has short leftover resistor leads soldered on so it serves as a barrel type resistor. However, I soldered the leads on the board first, and then the SMD in between them. This way you don't end up bending and cracking the delicate SMD component. So no soldering leads to it first.

The Theremino PMT adapter shows the L1 flyback coil having a current drop of 120mA, possiibly more, though the specification doesn't state a requirement for high current, just "L1 must have a low series resistance (A few ohms, maximum 5, absolute maximum 15)". The eBay descriptions unfortunately didn't list series resistance for these components, but I opted for the highest current rating available, expecting series resistance to lower alongside it: "3.3mH 300mA 6x8mm 10% Ferrite Core Shielded Radial Lead Inductor Black". With a standard multimeter, they measure about 10.5 ohms. The 1mH chokes that eBay specs said are 0.91Ohm @ 1Khz measure 1.6 Ohm at DC multimeter. So I am thinking these components might be better obtained somewhere else than eBay, but I'm making do with them for now.

The choice of resistors is one issue; in the Theremino PMT prototype pictures a curious mix of metal film and carbon film resistors can be seen, leading me to assume the designers didn't consider resistor choice to play much of a part, so I've followed suit with getting whatever resistors I could best get. For the unpolarized condensators there is bit of an issue with physical sizes. In particular, 2kV cap instead of the specified 1.5kV was the most affordable choice for the signal filter capacitor, and also best considering the target voltage range, but on assembly it turned out I had to bend the lead dangerously and position it leaning over against the noise shield to get it fit within intended encolosure. I worry what this might do for noise, so sticking with 1.5kV capacitor may prove for easiest assembly and casing.

And speaking of noise shield, I built mine from electric guitar copper tape. In the end it may not have been as clever as I expected, and I still soldered the connection posts to it all the way down of course. Mostly the copper adhesive tape seems to ensure no sheet metal cutters are needed, and if there's a danger it makes contact with the casing, it's easy to crumble into lower profile. At worst I've introduced more ground loops by the doubled over (lenghtwise) copper tape, although I'm hoping that isn't a real issue here.

All other components have been populated here, except for the T1 which is alternative to Mosfet BSP300. Since I had the BSP300 from Mouser, T1 isn't populated (and in fact wouldn't fit because BSP300 is blocking the holes on the flipside of the board). The RC circuit intended for T1 is populated, because instructions said it could be left on. Beyond that there were a few gotcha's, like the eBay sourced trimmer potentiometer had to be stretched to max to fit the PCB footprint and the pin-strips are difficult if not impossible to cut without the final pin breaking free. I also worry about how good connection the pin-strips are making; they seem to be failing the test set forth in the build instructions, despite carefully choosing what looked like quality parts.

As of right now, the board powers up successfully with no lost magic smoke, and provides voltage. There's an entirely expected problem, however: The voltage is beyond what multimeters will measure, and the current output is below what the multimeters tend to demand. More on that in another update.