A Class A solar pyranometer, the original shielded cable, 10 meters. Dies every 1-2 hours. In the end, I switched to 4-20mA. A few years ago at a solar farm site, I learned something the hard way: EMI doesn't care how expensive your cable is. Here's what happened. We needed to bring data from an EKO MS-80S pyranometer (Class A — the kind academic labs use) into our system. The device had RS485 / Modbus RTU. The manufacturer supplied a shielded twisted pair cable. 10 meters distance, standard industrial environment. By textbook standards, this should have been a plug-and-play setup. Reality: The first 1-2 hours worked perfectly. Then comms died. Completely died. Every poll from the software side returned timeout — not corrupted values, not CRC errors, just no response. The weirdest part was the recovery: just unplug the RS485 connector and plug it back in, and comms returned to normal. But 1-2 hours later, it died again . Periodic, reproducible, like it was mocking me.…