Trend load vs MS/TP stability

Yesterday on a 38.4 kbps MS/TP trunk with 85 BACnet devices, 15‑sec BMS trends for DAT and valve position pushed token rotation past 5s and we started losing COVs and seeing spurious alarms; bumping trends to 60 sec calmed the bus but killed the resolution I need to tune a hunting AHU cooling valve loop. Do you offload high‑freq trends to an edge logger (MQTT/OPC) or bite the bullet and segment the trunk/increase baud to keep analytics local?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌‍‍​‌⁠‌‍‌‍​‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​​​⁠‍‌​⁠​‍​⁠​‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌​‌​​‍‌⁠‌‍‌‍‌‍‌‍‍​‌‌​‍‌‌‌‌​⁠‍​‌‍⁠‍‌⁠‌‍‌‌⁠⁠‌‍‌‍‌​‍‌‌‌⁠⁠‌​‌⁠‌​⁠​​‍​‍‌⁠⁠‌

I log 5‑sec points in-controller and backfill nightly; MS/TP trends stay 60s and no lost COVs.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌‍‍​‌⁠‌‍‌‍​‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​​​⁠‍‌​⁠​‍​⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌‍‌⁠​⁠​‌‌‍‍​‌‍​‌‌‌​‍‌​⁠‍‌​‍‍‌‌‌⁠‌⁠‌​‌​‌‌​‍⁠‌‌‍⁠​‌‌‍​‌‍​‍​⁠‌‌​‍​‍‌⁠⁠‌​

Before offloading, cap Max Master to your highest MAC and set the supervisor’s Max Info Frames to 1; that alone dropped my rotation from about 5s to about 1.8s on a 38.4 kbps, about 80‑node trunk… For the hunting loop, I do a 10–15 minute “burst” trend at 2–5s in‑controller and let it backfill, then flip back to 60s so the bus can breathe. If the fleet supports it, bump to 76.8 kbps or split the segment, but I’d try the Max Master/Info Frames tweak first, @OP.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌‍‍​‌⁠‌‍‌‍​‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​​​⁠‍‌​⁠​‍​⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌⁠⁠‌‌‍‍‌‍⁠‌‌‌‌‍‌‌​⁠‌⁠‌‍‌​‌​‌‌‌‍‌‍‌‍​⁠‌⁠‌‍​⁠‌‍​‍​‍⁠‌‌‍​‌‌​‌‌‌‌​‍​‍​‍‌⁠⁠‌​