A quick overview of the revised plan for the system is required. Here are the changes since the first one:

1. Wireless support has become ubiquitous due to the low cost ESP8266 and ESP32 boards.
2. Raspberry Pi boards are now capable of routing
3. Field installation realities have been identified to require remote mounting of some sensors, e.g., wind sensors (10 m or 33 ft high) and soil sensors (multiple may be desired).
4. Microprocessor connectivity to wired Ethernet has become less common than WiFi.
5.  Automatic data logging and processing can easily be accomplished by a Raspberry Pi or similar Single Board Computer (SBC).

For this reason, we have begun to re-tool to use wireless in a restricted WLAN to connect various sensors to a base hub. This will also allow the hub to be removed from the sensor cluster(s) to avoid self heating effects. The new approach will now be:

1. Raspberry Pi 3+ or 4 (Rpi) hub hosting Docker containers and bare-metal apps to include at least: hostapd, python, mqtt broker, influxdb, grafana, and node-red.
2. Remote sensor interfaces running on ESP based modules with WiFi links to the hub and MQTT data outputs. Coding will be Arduino IDE compatible.
3. Fixed sensors with complex post-processing will be interfaced directly to the hub by cable.
4. Power for the system will still remain solar, but all power equipment will be handled and housed separately near the base of the main sensor pedestal.
5. At this point we are looking at making a Stevenson Box optional, but our system will continue to use one due to local environmental and pest conditions.

At this time we have one temperature/humidity module operating remotely to the RPi. It has been running with no issues continuously for 7 months. We did have one outage where the unprotected ESP board developed corrosion and a resulting leakage between pins on the PC board. Cleaning thoroughly with IPA resolved the problem. In a permanent deployment we now plan to conformally coat the PC boards.

The load on the RPi 3+ has been averaging about 25% during this time. It seems to have plenty of processing capability for this project.

Our test system is also interfaced to Home Assistant software which allows remote viewing of the data, and various automations to be performed.