On this page
People often use "battery test" to mean two very different operations. One takes ten seconds and is safe to run any time. The other drains your battery and leaves your equipment unprotected for hours afterward. Knowing which you are starting matters.
Two tests, two jobs
| Self-test | Runtime calibration | |
|---|---|---|
| Duration | Typically a few seconds to about a minute | Minutes to an hour or more, depending on load and battery |
| What it checks | Battery and inverter can carry the present load right now | How long the battery actually lasts at the present load |
| Battery afterward | Nearly full | Near empty; needs hours to recharge |
| Risk to load | Low (load drops only if battery or inverter is failing) | Moderate: little or no backup until recharged, and the load may drop if the UPS misjudges the end |
| Battery wear | Negligible | One deep cycle, which counts against lead-acid life |
| How often | Every few weeks to every 6 months (often automatic) | After battery replacement or when estimates look wrong; a few times a year at most |
How a self-test works
The UPS opens its input relay or otherwise switches the load to the inverter, then watches battery voltage under that load. If voltage drops too far, too fast, or the inverter reports a problem, the test fails. Most units then show a replace battery indicator, an alarm or a software event; see UPS replace battery light.
Many units run a self-test automatically when powered on and periodically afterward. The interval and whether you can change it vary by brand and model, so check the manual or software settings.
Running a self-test by hand
- Make sure the battery is charged. A self-test on a partly charged battery can fail falsely. After a new install or outage, wait for a full charge.
- Start the test. On many consumer models you hold the power or test button for a few seconds until it beeps; on others it is a menu option on the display, or a button in the vendor software, web interface or a command such as
upscmdin NUT. Check your manual, because holding the same button longer may turn the unit off. - Watch and listen. You will usually hear a relay click and see the on-battery indicator for a moment.
- Read the result. Passed tests typically return quietly to normal. A failure shows as a replace battery warning or an event in the log.
- Note it. Record the date, result and estimated runtime at your current load.
How a runtime calibration works
The UPS's runtime estimate comes from a model of battery capacity. As batteries age, the real capacity falls and the model drifts. A calibration runs the load on battery until the battery reaches a low threshold, measures how long that took at that load, and updates the model. Afterward the UPS recharges normally.
On some APC Smart-UPS and similar business units, calibration is a dedicated menu or software function and the unit manages the end point. On some consumer models, calibration is not offered at all. Behavior and requirements (minimum load, full charge first, whether the test can be canceled) differ by model, so treat the steps below as a general pattern.
Calibration leaves you unprotected
After a calibration, the battery is near empty and recharging usually takes several hours. An outage during that window may give you little or no runtime. Calibrate at a time when an outage would not matter, and never with life-safety or medical equipment on the UPS.
Running a calibration
- Charge fully. Leave the UPS on line power until it reports 100%, ideally several hours beyond that.
- Arrange a representative, non-critical load. Many manufacturers want a reasonably steady load, often cited as at least about 30% of capacity; follow your manual. A spare PC or lamp can stand in for critical equipment.
- Save work and disable automatic shutdowns if needed. Software-triggered shutdowns at low battery will interrupt the test. Some guides suggest temporarily changing shutdown thresholds; restore them afterward.
- Start the calibration from the panel, vendor software or the network card. Avoid changing the load during the test.
- Let it finish. The UPS returns to line power when done. Do not cut input power to "help" it.
- Recharge and verify. Allow a full recharge, then compare the new runtime estimate with the old one.
The manual alternative: a timed runtime test
If your unit cannot calibrate, a timed test answers the important question: will it last long enough?
- Fully charge the UPS.
- Connect a known, steady, non-critical load and note the watts.
- Unplug the UPS from the wall (or switch off its input breaker) and start a timer.
- Stop at your target time (say, the 5 to 10 minutes you need for a shutdown) or when the low-battery alarm sounds. Plug back in.
Stopping at the target time instead of running to empty gives you the answer without a full deep discharge. Compare the result with the runtime calculator estimate for that battery and load.
Our rule of thumb: test to your need, not to empty
Most people need a UPS to cover a specific window: ride-through of short blips, or enough time for a clean shutdown (see how much runtime you need). A yearly timed test that stops at twice that window answers the practical question with only a partial discharge. If the UPS cannot reach twice your required window on a fully charged battery, plan the replacement now, because capacity loss in aging lead-acid batteries tends to accelerate.
Worked example
A homelab UPS rated 1500 VA / 1000 W carries a measured 300 W (30% load). The vendor runtime chart suggests roughly 15 minutes for a new battery at that load. The NAS needs 4 minutes to shut down cleanly.
- Target for a timed test: 2 x 4 = 8 minutes.
- Year 1: unit reaches 8 minutes and still shows 50% charge. Healthy.
- Year 3: the low-battery alarm sounds at 7 minutes. The battery has lost a large share of its capacity, and on the next outage the NAS shutdown margin will be thin. Replace the battery, then calibrate so the estimate matches the new battery.
Why runtime is not proportional to load or battery size is covered in UPS runtime explained.
Reading the results
| Result | Likely meaning | Next step |
|---|---|---|
| Self-test passes, runtime normal | Healthy | Keep schedule |
| Self-test passes, runtime far short | Capacity loss in an aging battery | Plan replacement; see short runtime |
| Self-test fails once on a full charge | Battery near end of life | Retest after 24 hours of charging; replace if it fails again |
| Self-test fails with a brand new battery | Battery not fully charged, wrong battery, loose connector, or UPS fault | Check connections and charge time; contact the manufacturer |
| Calibration aborts early | Load changed, input returned, or battery too weak | Retry with a steady load after a full charge |
Software notes
Most vendor tools (PowerChute and PowerPanel) expose self-test and, on supported models, calibration. In Network UPS Tools, available commands depend on the driver and model; upscmd -l lists them, and common names include test.battery.start.quick and calibrate.start where supported. apcupsd offers self-test and calibration through its apctest utility on supported APC models, which requires stopping the daemon first.
# NUT: list commands your UPS supports, then start a quick battery test
upscmd -l myups
upscmd -u admin myups test.battery.start.quickTreat these as examples; names, users and support vary by setup.
Frequently asked questions
Why did my UPS pass the self-test but die quickly in a real outage?
A self-test only proves the battery can carry the load for a few seconds. A battery can pass that easily while having lost most of its capacity, so runtime collapses after a minute or two. A timed runtime test or calibration reveals this; a self-test cannot. Battery age over about three years makes it more likely.
Can a self-test drop my equipment?
It can if the battery is badly worn or the unit is overloaded, since the load moves to battery during the test. That is rare but is exactly why some admins schedule self-tests during low-impact hours. Equipment with dual power supplies fed from separate sources is protected; single-corded servers are not.
Should I run a calibration on a brand new UPS?
Let the battery fully charge first, typically for 8 to 24 hours depending on the model, then run a self-test. A calibration is optional on a new unit; some manufacturers suggest one after installing a new battery so estimates match. Follow the manual's instructions.
Why does my runtime estimate jump around?
The UPS calculates runtime from battery voltage, the current load, and an internal model of battery capacity. When load changes, the estimate changes instantly. Temperature, recent discharges and an uncalibrated model also cause drift, which a calibration corrects if the battery itself is healthy.
Can I calibrate a UPS without software?
On some models you can start a calibration from the front panel menu; on others it is only available through manufacturer software, a network card or a tool like apcupsd or NUT. Some consumer units have no calibration function at all, in which case a manual timed runtime test is the alternative.
Sources and further reading
- IEEE Std 1188, Recommended Practice for Maintenance, Testing, and Replacement of VRLA Batteries for Stationary Applications
- IEC 62040-3, Uninterruptible power systems: method of specifying the performance and test requirements
- Network UPS Tools (NUT) project documentation
- UPS manufacturer manuals and software guides (APC PowerChute, CyberPower PowerPanel, Eaton, Tripp Lite)