My little sensor assemblies are gaining some ground from R&D. I had to toss a few ideas out, and rejected a number of epoxies, but settled on these little things in the latest batch.
Real cable is by far the most expensive thing here. This is a PUR jacket, PVC inner, 3 conductor job. At a little over $1.50/ft, 8 feet of cable adds up fast. There are even fancier jacket materials, but they run $3+ a ft. Over a generic PVC cable, this is quite the expense but should make up for its self in long term reliability. I'm also trying to source as much as possible domestically, and import cables are right out for that (not going to touch them, lest I end up with some copper clad aluminum).
The capsules are are a thin wall acetal/delrin construction (0.5" diam), with bottom caps also of delrin. Inside is a _real_ DS18B20 TSOP-8 mounted on a carrier board, and a power decoupling capacitor.
Finding the right potting compound took awhile. I'm currently using Loctite 870FT - not cheap, but it has the right viscosity and the cartridge dispensing is perfect for this, even if it means I lose some to the mixing nozzle - just incentive to make bigger batches. I had a number of failed potting attempts previously, and pot mix was just simply too much work.
The commercial guys are using this 5k thermistor pill, potted (with probably a non-thermal modified resin) in an acrylic tube and then heatshrunk over. I may borrow the idea of heatshrinking the cable exit of my version to just clean up the looks.
More tests soon!
Probably the best recommendation for reliability would be enabling multiple pins?
