The network is running again and it's tied to enough horsepower that I would need to do a cause & effects analysis on the DCS system and the power distribution system before I'd be allowed to apply for a hot work permit just to put my scope back on the terminals. Could take days.
It's against my nature to detect a problem or anomaly then walk away without solving it, but my competition is being paid for 5 years of 24/7 support on it. If they want to tell me to stop investigating, I'll stop investigating.
Dan, thank you for that practical look at RS485 biasing, and for a look at the possibility of an "all 1's" packet. The absence of Start bits rules that out, unless my resolution was really too low to see them.
Biasing of termination of RS485 one of the most poorly understood topics in industrial data communications. The guys at B&B, RESmith, and RobustDC certainly understand it, but ordinary field techs are left to the recommendations of manufacturers representative and tech support engineers (including me !) who are just repeating, sometimes with misplaced enthusiasm, what they've been told.