I bought the most expensive cable I could—and it still died. Welcome to RS485 vs 1000V DC.

At a solar farm, an EKO MS-80S pyranometer on RS485/Modbus RTU worked at first but then kept timing out every 1–2 hours.

Swapping the sensor and cable did not help; the real issue was EMI from a nearby 1000V DC line and power equipment coupling noise into the signal entry.

The RS485 transceiver may have latched up until the connector was reset, pointing to a severe real-world noise problem despite shielded twisted-pair cabling.

The system was eventually converted to 4-20mA, which proved more reliable in the high-voltage environment.