PROBLEM SOLVED

As the problem manifested itself with loss of LAN (wired) communications, that is where I was focussing my attention.

In fact, the problem was the 9v/1A power supply included with the unit. I was using the included supply for my bench testing. It was never my intention to use the “wall wart” in the final installation.

Powering the A3 at 24 VDC from a POE splitter has resulted in the unit running for nearly and hour now without a hiccup. So I think the problem is solved.

My conclusion is that the tiny (1A) supply is simply inadequate and at the 11th minute the A3 was doing something that required an extra bit of power causing a momentary part-undervoltage condition.

Usually the micro can signal an undervoltage condition (typically in /var/log/syslog). A feature request could be the addition of such an alarm – usually many voltage “near misses” are seen before the one that actually impacts operation. That certainly would have saved me a *lot* of time.

Ken