The purpose of this project was to have a low-cost temperature sensor solution for home use. While investigating temperature sensors it seemed that most sensors start at around $35 and are limited to N number of sensors. Many of them included subscriptions which I don't like.
The solution that I came up with based on my recent learning of time series databases and moving metrics across the wire has been really stable so far. I've tried to make it as simple as possible with as few moving parts.
At 10 minute interval the 2500mAh provides ~2 months of battery.
After creating the backend and dashboards I found it painful to constantly have to open a device to see the temperature. Using an AdaFruit PyPortal I was able to easily visualize from a great distance the temperature. I am pretty happy with the result. You could probably accomplish the same result with much cheaper hardware, but this is what I had around!
PyPortal Temperature Visualizer on GitHub
My local InfluxDB runs on my main computer because it's on 24x7. I use Docker because I run Windows, but didn't want to limit myself. Thankfully InfluxDB is extremely well-made, and is not very impactful on the host.
The following command can be used to easily create a backup of the very small sensor database for off-site storage purposes. This will cost pennies per month in any cloud blob storage offering.
Why backup this way? There is no down sampling. We have perfect resolution in every backup and the backups can be used to restore any time ranges from within each backup.
docker run -v /path/to/backup/dir:/backups influxdb:1.8 \
influxd backup \
-portable \
-host host.docker.internal:8088 \
/backupsSensor:
Battery:
Relay:
Battery Recharging: