All Sky cam

Software: https://github.com/AllskyTeam/allsky
Hardware design features:

1 ) Computer: RPi compute module 4 with 32GB eMMC and 8GB RAM (no SD card). The motherboard is a Waveshare CM4 POE board (B) version.

2 ) Camera is RPi HQ with 2.5mm f1.6 fisheye 4K lens

3 ) The body is made of an industrial PVC 125mm connector. Wall is 8mm thick. Very sturdy!

4 ) The internal structure is made of 3D printed PLA, while the exposed parts (to the Sun and elements) are made of PETG.

5 ) Airflow: the CM4 is completely enclosed and a 40mm fan can cool it blowing the air into the dome thru an air pipe. the OS runs a service that keeps the CPU temp at 45C by controlling the fan with a PID algorithm.

6 ) The lens has a focusing system made with herringbone gears and a 12V 28BYJ-48 (converted to bipolar) and driven by a DRV8825 current limited accordingly. Two end stop microswitches are also protecting the lens focus ring and allow the system to find the focus range center. An interactive python script controls that.

7 ) Sensors: Dome has a BMP280 for temp, hum, atm pressure. Focusing motor has a DS18B22 temp sensor. The on board fan controller is 4 wires with tachometer hence allowing precise control.

8 ) 6 inch acrylic dome (treated with K2 - car headlight UV hazing protector)

9 ) Most fasteners are made of steel 316 to help with rust resistance. Screws into the PVC tube are not piercing thru maintaining the enclosure sealed.

10 ) The dome floor around he lens is made of concentric ridges to avoid internal reflections

11 ) Power and network connectivity rely on IP67 RJ45 connectors.

12 ) Ancillary software: OS Service for PID fan control, starts at boot. python focusing aid runs on the RPi and controls the focusing motor, with user inputs, multiple step sizes, and backlash correction. A Python script can run on a laptop to analyze the images and provides 4 sharpness metrics on a user defined crop of the image to help with focus tuning.

Grafana dashboard to monitor camera environmental parameters.
The camera pushes the data to na MQTT server. Node-Red collects the data into an InfluxDB database and grafana plots it. The image is the latest available to the local webpage.