I had a chat with a friend of a friend during Christmas time about a correlation of outside temperature and the natural gas bill and how good it would be to forecast these things. After this conversation, I thought that it would be great to collect temperatures in the different places in the house. The main reason for this is that the house is not as warm as I would want and when we insulate it we need to have numbers to compare. "It's got a bit warmer" sounds a way too subjective.
I was looking for something that can be connected to and controlled by raspberry pi which I already got. After a bit of googling, I went for this sensor mostly because it is cheap, there is no problem how to power it and there are open source libraries that can decode signal from the sensor.
I left buying receiver until the very end because I have seen there are quite a lot available on Amazon. However, to my surprise, it took me almost two weeks to find the receiver that would work adequately.
This one was receiving a signal in a range of 1m.
This one I haven't been able to get working at all.
The one that I eventually used for this project is RXB6 433Mhz Superheterodyne Wireless Receiver
This was probably the easiest part of the project. Mostly because there is a brilliant guide from aquaticus which has a step-by-step instruction and great explanation of how to connect 433mhz receiver and configure a service for decoding sensors signal.
There are a couple of things I want to mention here. the first one is that I had a problem to configure libgpiod library because of the missing Linux kernel headers. There are workarounds described in this issue and they worked for me.
Another thing is that if somehow you would get hold of Digoo DG-R8S 433MHz (on the picture above) it won't work out of the box because it uses different protocol but there is a PR that works for Digoo DG-R8S 433MHz and for Digoo-DG-R8H 433MHz sensors.
When everything was wired and configured I hooked it up to homeassistant and it looked pretty cool!
However, I wanted my dashboard to be a bit more flexible and I wanted to have all sensors temperatures on one graph. I couldn't quickly find how to do it in the homeassistant (or, what is more likely, I just wanted to play with grafana) so I started exploring how I can present this data on grafana dashboard. I have to mention that I have never configured grafana before this so I am sure many things can be done better or in a more straightforward way but I learned as I went.
I set the following requirements to my dashboard:
-
All sensor data needs to be stored and not only last 24hr or 3 days.
-
I need to get an outside temperature for my location (I wasn't keen on leaving the temperature sensor outside because I wasn't sure it would survive rain).
-
Data graph should be easy to configure.
-
Everything should work on raspberry pi.
The main grafana how to source for me was this post This post gives the idea how to set up Grafana + Influx + Telegraf. I included docker compose file in this repository with all configuration I have. This is how my temperature dashboard looks like now
Couple of things, which I think, is worth mentioning is that I configured Telegraf MQTT input plugin to collect data from sensors and push them into InfluxDB. I also added two ruby scripts. One to get an outside temperature from Weather API and then publish it to MQTT broker. Another one for testing internet bandwidth using speedtest.net and then publish results to MQTT. The last thing I want to mention here is that I tried to wrap nexus433 service into docker container but due to lack of knowledge I haven't been able to complete it.
You can see some gaps on the graph above and it's happening because the signal from the sensor can't reach the receiver. I was trying to move sensors around to solve this problem but this strategy was neither effective nor sustainable so I eventually created a simple 17.3 cm antenna. It solved the problem of lost signals from some sensors.