first thing that comes to mind is a beer ball--do they even make those anymore?

also look at "floating pool lights" Maybe you can get them wholesale somewhere.

Hi, I think the enclosure could be a lot smaller, if you build on some of the other work folks have done on thermal fishing bob designs, for example, the 555-based analog design work, which'd be super tiny, or the Digispark based version I posted on @kgrevera's post a while back:

http://publiclab.org/notes/kgrevera/07-06-2015/cape-cod-thermal-fishing-bob-field-test#c12101

A TinyDuino could also run off a coin cell battery, making it even smaller.

The original idea for the TFB was to have it fit inside an actual fishing bob:

That way, it can be cast from the shore and dragged back slowly for a thermal image. This also means that you need to be measuring only one temperature at a time, because the color indicates a single temperature at a single depth. This simplifies the design dramatically, making it much easier to construct. A single thermistor and a single RGB LED plugged into a DigiSpark or a TinyDuino has only 4 parts:

• Small arduino ($9)
• Battery ($7)
• Thermistor (<$6)
• LED (<$4)

= <$26

Instead of adding complexity to the device, we could keep construction time and cost very low, and cover a large area quickly by casting it with a fishing rod. To measure at depth, you can put the thermistor deeper with a fishing weight, just like you'd do with a hook.

For a container, I think an old film capsule could work, since it'd seal nicely (you could use plenty of silicone glue or hot glue to seal the thermistor through a hole cut in the bottom) or you could go the @donblair route and use a tiny Dasani water bottle. Don's done a lot of work to get a good seal on a bottle cap, and is also working with thermistors.

Scratching the outside of the bottle with fine sandpaper would also help diffuse the light to make it easier to see.

What do you think?

I know you're looking for a sphere, but I'd instead look for a hydrodynamic shape that can be pulled without a ton of actual "bobbing." on the things-that-need-machining side, I'd look at some diy "towfish" for side-scan sonar and similar activities.

I second @warren's suggestion of following the Riffle project into PET bottles, and think some weighting and fins could turn a bottle into a functional enough towfish. if you need to run wires out of the bottle I can't recommend punching straight through a bottle cap. its hard to keep the sealant sealed to the PE caps. try this instead: http://publiclab.org/notes/mathew/02-26-2015/sensors-in-soda-bottles

OMG I forgot about that post! One of my faves. Awesome, @mathew!

The Beer ball was in fact the original idea Summer 2012 -- do you have photos of Eymund's original prototype handy? I'll try to find them. It's fantastic.

This whole folder is fantastic - @eymund is a craftsman https://www.dropbox.com/sh/n2s0b7lgqxfkl1o/AABDMUeDGhp1g2dqHTSCHbPYa?dl=0

Thanks for the suggestions everyone (@warren, @mathew, @liz, @DavidMack)! I've tried a number of things in the past few months and wanted to post an update.

After more time with the Arduino setup, I've shifted to custom electronics based heavily on @donblair's Riffle design (thanks!). Assembly and debugging can be somewhat time consuming, but it's a huge plus to get away from the loose wire fussiness that I always hit with Arduino set-ups. The smaller compact size and lower cost (probably ~$30 total?) is also nice.

I've started casting the electronics in paraffin wax for sealing (as suggested via Don, though I'm not sure of the original source!). This can be somewhat time consuming to do/re-do, but it is very reliable. I found that the various object + silicon/hot glue/epoxy/tape solutions worked okay in calmer water settings for short periods of time, but those solutions were not consistently reliable in real life. They didn't hold up well when placed in flowing water for long periods of time, dragged behind a boat, or placed on the edge of the Charles where they'd hit against a wall from time to time. The aesthetics also felt a bit unsatisfying.

In terms of packaging, I tried a number of modified commercial spheres. My favorites were casting the board in wax using an ice cube mold and sandblasting a Christmas tree ornament and casing the electronics in wax in the bottom of the ornament.

The bigger Christmas ornament spheres worked best for for long exposure photography documentation and for the real time experience too. The smaller wax spheres are pretty fun but they weren't very visible from a distance. I think the traditional fishing bobs would run into this issue as well, unless they were used at a very high density.

The "how much should it do" question is really interesting. I like the idea of data as performance and as "traditional" scientific tool concurrently. I think they complement each other very well and invite various groups of people to the conversation. That motivated my use of the faster, more digital temperature sensor as the inclusion of an SD card--the coarse data can be seen in real time and fine-tuned data is documented for later use. Since the bigger spheres work best for documentation and the custom electronics are small it is easy to include all of this functionality in the design.

Ooh, i love the christmas tree ornament; didn't know those were available in plastic:

Do you have some pics or plans for the riffle-based board? Love to see them.

Here's a picture of one of the riffle-ish boards (currently called "fishnet"):