Teardown of a Garmin temperature sensor
It was a hot day, but how hot was a guess, because there was no reading from the Garmin Tempe sensor. Probably an empty battery. But after replacing that, it was empty again in a few hours. Was it a bad battery? Let’s try another. But that one was also depleted in no time.
The CR2032 coin cells used in sensors of this type, normally last well over a year, so this was definitely abnormal.
Opening up the sensor
There was nothing to lose, and who doesn’t like a teardown of a gadget, so I opened the sensor, using a hobby knife, a small flatbed screwdriver and some force.
There are no user serviceable parts inside of course, so I wasn’t expecting to accomplish anything, but who knows, it was a well used sensor, maybe dirt had made it’s way into it and nearly short-circuited the battery somehow.
It turned out there was no lint or debris inside, but some dirty stuff had leaked on the printed circuit board, from what only can have been a capacitor (near the green arrow in the picture).
The capacitor itself had mostly turned to charcoal. Replacing it was out of the question, so with the help of someone more knowledgeable about electronics, it was determined that the capacitor was part of the reset circuit of the processor, and that the sensor might work normally without it – and would stop eating batteries – if the conductive debris was removed.
To the right a detail of the culprit. Getting a good picture of the board with gunk and capacitor removed was too hard, it just doesn’t show clearly.
Closing it up
After that the top was replaced. The casing is now held together with sticky tape. It isn’t pretty and isn’t waterproof, but it does work.
For best results measuring temperature while running, mount the sensor on the underside of the visor of a cap, out of the sun. Fellrnr does it the same way, and he’s an authority on all things related to running, GPS watches and sensors.
Instead of prying a sensor open, you can view pictures of the inside on the FCC website.
Comparing the picture of the back of the PCB on the FCC site (detail of the FCC board in picture below; top) and the one from my sensor (middle), shows that there is an “island” in the middle that has an IC on it in the FCC picture, but not in mine. And it isn’t on the other side of the PCB either (bottom).
The markings are the same on both PCB’s, so the FCC photo is probably not of a prototype. I had always assumed that the temperature sensor would be in the middle, but it clearly isn’t. The MSP 430 CPU has a temperature sensor on board, so a separate temperature sensor isn’t needed, although it does take some more programming effort. Either the separate sensor was deemed unnecessary in a later stage of development, or it is just what it is: a mysterious void.