DIY weather station
I have created a monster using various sensors attached to numerous rpis.
ingredients
- raspberry pi zero w (these used to be abundant and $5 before 2020)
- lots and lots of python
- bme280 sensor (reasonably accurate for temperature, humidity, and pressure)
- 2 weatherproof electrical conduit / housings
- Some kind of power adaptor that works well outdoors and isn't too bulky
- Dallas DS18B20 temperature sensor (for the ground)
- RG11 rain sensor - RG11
- Nick discovered this impressive device
- the GPIO needed a stronger pull-down resistor to reduce false clicks, jwz - my pay phone runs linux was the right breadcrumb path.
- The cycle time is 50ms for each 'click' on the rg_11 bucket mode
- LaCrosse-TX141THBv2 intercepted with
rtl_433and a rtl-sdr - PM 2.5 sensor - Adafruit PMSA003I
- Can easily detect when I cook on the stove or the neighbors use their fireplace
- Wired up to turn on an air filter when the value is over 5
- Verified that using tap water in an ultrasonic humidifier is bad!
- CO₂ - Adafruit SCD-30
- adafruit e-ink bonnet
- BME688 (not in use yet)
challenges
temperature accuracy in sunlight
- various shades and enclosures tried
- none good enough yet
- changed to LaCrosse sensor to help positioning
remote thermometer housing - I expect the metal venting to radiate more heat
barometric logging
One of the more interesting pieces of home technology in the later 1800s was the barograph. When Kraktau expoded in 1883, these were able to track the blast wave's progress across the world more than 4 times.
Winchester, Simon (2003). Krakatoa: The Day the World Exploded, August 27, 1883 - Excellent reading for anyone with an interest in volcanoes.
- my 60s barometric logging was resampled in the rrd before I had a chance to check for the Tonga volcano blast wave. Damn you Kraktau, I was almost setup to measure this!
Getting a peak value from a digital barometer is a slightly bigger challenge. The relaxation phase of a huge blast may take several minutes so detection is relatively easy. Smaller blasts, like those from house or warehouse explosions have been strong enough to wake me up but well under the measurable size for 60s sampling.
possible enhancements
- Modify one housing for shade and airflow (temperature and humidity accuracy)[possible Stevenson screen]
- VEML6075 UVA UVB UV Index Sensor
- TTC-103 Thermistor
- add live data via MQTT websockets
- Argent wind-rain sensors
meteorological reference materials
Some calculations and definitions were needed
- Marine forecasts glossary for the pesky barometer rules.
- Arden Buck equation - for computing dewpoint from humidity and temperature.
motivation
Why do things the easy way? Build your own weather station guide
i2c issues
Some more info on working on the reliability of i2c.