How to Test a UPS: Charge Is Not Health

A UPS's charge reading is not a health reading. When a UPS shows full charge, it can still drop the load the moment it transfers to battery power. The display may show everything is fine, but it is actually answering a different question. The gap between charged and working is important. Sizing the unit is settled territory, but this is about proving the UPS does its job on schedule.

The built-in self-test is not a runtime test and covers less than the name suggests. Most units test themselves when powered on and every two weeks, running on battery for approximately ten seconds. This test measures the transfer relay closing and inverter coming up. A battery that has lost most of its capacity will hold ten seconds without complaint.

Charge is a percentage, and capacity is minutes. An untested UPS is a surge protector with a battery. Test on schedule and make it a real test by connecting the load and pulling the input cord. Watch what the load does during transfer and time the battery runtime before the configured shutdown fires. A ten-second self-test skips the load and a runtime figure from a spec sheet skips your gear.

The UPS runtime calculator can model the curve, but a modelled runtime and a measured runtime are not the same. If you prefer not to pull a cord on live equipment, you can start a deep battery test using NUT. A UPS can fail in a way that takes the load down with it, which is a real failure mode. Pure sine wave power is recommended for servers with active PFC power supplies, as simulated sine waves can cause interruptions during transfer.

USB and NUT are cheaper alternatives to network management cards for monitoring a UPS. USB plus NUT provides the necessary functionality for most homelabs. Testing the UPS ensures it provides a clean shutdown and avoids expensive unclean power-offs.