I've put together a PIN-diode based radiation counter for balloon launches. My prototype in the photo, built on 0.050" sq., 0.8mm grid board, weighs 6 grams. This is loosely based on a design put out by the CERN Open Hardware Lab for student outreach, described here:  https://www.mdpi.com/1424-8220/19/19/4264

One shortcoming of the source design and similar ones, is its extreme sensitivity to external EMI - the front end is a transimpedance amplifier: essentially a high impedance scope probe preamp. As such, it's sensitive to noise injection by capacitive coupling. Shielding can be used, but this is difficult to do properly and it adds weight. In this design, the signal is sensed differentially and noise is subtracted out as a common-mode signal. This makes for a low enough sensitivity to noise, that shielding is unnecessary. The amount of noise rejection is proportional to the component matching of each half of the differential path. In batches of 1% resistors and 5% capacitors, I can measure component values and achieve 0.1% match for R and 1% for C, yielding a 40dB rejetion of the common-mode noise (CMRR).

Details:

The sensor is five Vishay BPW34 photodiodes in parallel. The amplifier chain delivers a fairly healthy 500mV pulse to the U3 comparator. An LM2665 is used for voltage doubling to supply diode bias. Since the diodes must be heavily reverse-biased to increase their detection volume (the diodes' depletion region depth), it's necessary to AC couple the diodes to the front end. The BJT capacitance multiplier at the bottom right of the schematic, was not used.

The prototype will fly on NU7B-22, as soon as the U4B can support transmission of external sensor data. It was a real pain to build on grid board though. As soon as this flies, I will fab a board at Oshpark or similar, and share the board design after I get some units back. The eagle-eyed looking at the schematic may spot that the function of U1 can be combined into a single op-amp. I haven't tested that configuration, but there's good reason to try it: it saves an op-amp and fewer components have to be matched. When I fab the boards, I'll fab both kinds, but the prototype is the posted schematic.

Steve NU7B

Oshpark will fab a board on 0.8mm substrate, but for this they use 2 oz copper, which adds weight. Does anyone know a PCB fab that has a lighter option - thinner?

 

Steve NU7B

Thanks Ross:

Yes I noticed 0.6mm was also much cheaper than 0.4 with pcbway as well. This is the first time I've fabbed a PCB - there will be a learning curve with KiCAD, and hopefully the component library is complete enough on that tool.

steve

JLCPCB can do 0.6 mm thick boards with 1 0z copper.  0.4 mm thick boards get really expensive.

72/73,

Jim, K8IQY

Microwave substrates are available with 8- or 10-mil dielectric thickness with 1/2- and maybe 1/4-oz copper, but it will cost you.  However with small patterns a single 12 x 18 or 9 x 12 sheet will yield a lot of boards.
For simple circuits you can hand-cut the pattern and peel the unwanted copper with a soldering iron and tweezers.
Also, try AWG30 wire wrap wire interconnects to save some weight.  Magnet wire is lighter yet but more difficult to strip.

73, Don N2VGU

Don,
I have been looking for wire wrap wire. Who still sells it?
--
Colin - K6JTH 
http://tangokeys.com

Not to butt in, but mine came from Amazon:

https://www.amazon.com/gp/product/B01LWI20M0/ref=ppx_yo_dt_b_search_asin_image?ie=UTF8&psc=1
--
Cheers,
Chuck

...do you intend to let it hover over Alaska? Make sure you paint some nonsense meaning, far eastern looking characters on it: they will go nuts trying to decipher the message ;-)

https://smile.amazon.com/Solder-Coated-Plated-Copper-Wire-Wrapping/dp/B083352WFF  8-color spool, all but white and grey
https://www.adafruit.com/product/4730 8-color spool, all but white and grey
and if you are OCD like me,
https://www.ebay.com/itm/123831170093 white
https://www.ebay.com/itm/363100710321 grey

73, Don N2VGU

I wanted thin, 30 gauge tinned wire insulated with thin teflon, as I use these for short point-to-point soldered jumpers.  I found some on eBay. Teflon insulated single strand tinned 28 to 30 gauge gauge wire tends to be rare.
Mark K9TR    

Replying to self and all. eBay has tons of 28 to 30 gauge tinned wire in both teflon and kynar insulation types.

KiCAD was easier than I thought it would be. Here's the final result for the radiation particle counter: 1.9 in.sq. which will weigh 1.9g (unpopulated) with Oshpark's 0.8mm board. After I get some back for verification, I'll share the project on Oshpark's site.

Steve NU7B

Detector now verified, and it's about to fly on NU7B-22. Oshpark Link for shared project is here: https://oshpark.com/shared_projects/X0ZccRAU

Github site with files, photos and schematic is here:  https://github.com/stefanoaz/Particle-counter-V1

And here's a picture of the board, with six PIN diodes... PCB is 0.8mm, assembled weight 3.2g.