“Very nice. I have installed it on a Pi Zero with both a custom app to send information to Home Assistant for the readings via MQTT and also installed room assistant. The screen has been configured to flash up messages from Home Assistant also using MQTT. The whole thing sits on a small 3d printed stand and works very well”
“easy to set up on pi zero W. just follow clear instructions on the packet from the pimoroni website and you'll be up and running watching basic tasks either in the shell or on the colour OLED. The Pi zero can handle these hats with relative ease.”
“Brilliant little product. Useful for checking moisture levels in shower room. Just watch out you do not fix it in reverse as it starts to burn some of the components on the board.”
“Its a very neat and tidy way of making an enviroment sensing Pi without using a breadboards and welding and wires flying around. Runs fine on both 4B and Zero W, with caveats - details below.
Note that the default library requires root access on Ubuntu. That was not required on Raspbian; I am not an expert as to why. This makes it critical for your script permissions to be secured, otherwise someone who knows where your service is pointed to can replace the script to make a rogue agent. In short, use Raspian if possible. I also don't believe the sensors are the most accurate in the world, especially when it comes to temperature. This is mostly because of the heat from the Pi itself can rocket the readings to Sahara levels. The sample code includes a simple linear compensation T(real) = T(raw) - (T(cpu) - T(raw)) x F where F is an arbitrary factor you have to define yourself. There are two problems with this; firstly - the relationship between the two may not be linear; secondly it takes time for heat to propagate to the sensor, which means that compensating against the current T(cpu) does not make a lot of sense.
My personal experience is that you can just about get away with a small compensation factor (F=0.3~0.5) on my 5 Pi Zeros, but you will need active cooling on 4B for the temperature to make any remote sense. I tried running passive with a heat sink and the F sky rocketed to beyond F=4 which makes it wildly sensitive to any CPU load changes. So for my purpose I am just using the 4B as a development unit with Archiconda, while sticking to Zeros for deployment.
It also appears that the humidity calculations are factoring in the temperature. Since the temperature is completely off, it is now reporting humidity of 25-35%. As I am in London, there's clearly something wrong. I have to do another round of compensation using the compensated temperature and estimated dew point, then I arrived at a humidity that I am expecting. (Also revealed how inaccurate my Casio digital hydrometer is; so yea may be they are all as bad as each other)
If you are looking to monitor the temperature of a room to a specific range for a baby for example, this error might be enough to knock the system out of its usefulness; however if you are trying to see the trend of temperature change across time and not critical about single values, this will work well.
All in all I am quite happy with it, I think its nice piece of kit - albeit with limitations - that I can deploy around the house without strong worded complaints from the interior designer in chief aka wife. I managed to cook up a little animation that calculates the time of day relative to sunrise/sunset and shows indicatively where the sun is including elevation (i.e. higher in summer, lower in winter). Demo video (with random values) below. And there goes my easter holiday...”
