I've been handling power infrastructure orders for our facilities team for eight years. In that time, I've made five significant mistakes that I've documented—mostly because my boss asked me to, but also because I got tired of watching other people make the same ones.
The biggest one cost us $3,200 in emergency replacements and a three-day scramble. And the frustrating part? The problem wasn't the batteries. Everyone checks the batteries.
The problem was everything else.
The Problem You Think You Have
If you've ever had a UPS fail, you know the drill: the unit starts beeping, the battery indicator goes red, and you order a replacement battery. Maybe you upgrade from a Back-UPS to a Smart-UPS because "the bigger one must be better." Then you plug it in, confirm the lights are green, and forget about it until the next beep.
For years, that was my entire maintenance strategy. Check the battery light. Replace what's broken. Move on.
Then came September 2022. I walked into our server room and noticed a smell—not quite burning, but electrical. Three APC units were running hot. Two had battery casings swollen so badly they were bulging out of the chassis. The third wasn't swollen, but its fan was grinding like it had swallowed gravel.
Here's what my battery-light method missed: a swollen battery doesn't happen overnight. It builds over months. And the fan—that's not a battery issue. That's an airflow issue. Which means something upstream was clogging things up.
That was the moment I realized I'd been treating one symptom and ignoring the system.
What's Actually Eating Your UPS
When I started digging into why those batteries failed early—one was only 18 months old—I found three causes that had nothing to do with battery quality.
Your Environment Is Doing More Damage Than You Think
Everyone focuses on battery chemistry. But heat and dust kill batteries faster than cycling ever will.
Our server room had decent HVAC, but the UPS units were tucked into a corner where airflow was restricted. Dust accumulated on intake filters. And here's the thing I didn't know: the same logic that applies to how often should you replace air filter in home HVAC systems applies to your UPS. If your intake filters are clogged, the unit runs hotter, and every 10°C increase above 25°C roughly halves battery life. That's not a rule of thumb—that's IEEE 1188 data on VRLA battery life.
We were changing HVAC filters quarterly. We never touched the UPS intake filters. Three years of neglect, and I was wondering why batteries were dying early.
Now I check UPS filters every time I check HVAC filters—same schedule, same logic. If the environment is dusty (warehouse, workshop, or just a home office with pets), you probably need to check every 1–3 months. Clean or replace as needed.
The Sine Wave Problem Nobody Warns You About
Around the same time, we started having issues with a small ups apc es 700 that was connected to a piece of networking equipment. The unit tested fine with a multimeter—voltage looked good—but the equipment kept rebooting randomly. Not often. Maybe once a week. Just enough to be annoying.
Turns out the ES 700 outputs a stepped approximation of a sine wave (APC calls it "stepped approximation to a sine wave" on the spec sheet; most people just call it simulated sine wave). For basic electronics—routers, chargers, simple power supplies—that's fine. But for equipment with active PFC power supplies or sensitive components, it can cause random shutdowns.
I didn't connect the dots until I read the spec sheet properly. The fix was switching to an apc sine wave ups (specifically a pure sine wave model like the Smart-UPS line). The random reboots stopped immediately.
So no, not all APC UPS units output pure sine wave. The Back-UPS and ES series are simulated. That's fine for most residential and basic commercial use. But if you're running medical equipment, active PFC servers, or anything with a motor, check the spec sheet before you assume "UPS is UPS."
Your Transfer Switch Is Lying to You
Here's one that still bothers me. We had a zenith zts transfer switch on one of our backup power circuits. I assumed it worked because the indicator light was green. Why wouldn't it work? It was installed by a licensed electrician. It passed inspection. It had a green light.
Then we had a real outage—not a test, an actual grid failure—and the switch didn't transfer. By the time we manually engaged the bypass, we'd already lost power to a critical circuit for 47 minutes.
What I learned: a green light on a ZTS doesn't mean the switch will transfer under load. It means the switch has power. That's it. The actual transfer mechanism—the contacts, the relay, the control circuit—could be stuck, corroded, or misconfigured, and you'd never know until you need it.
Now we test the ZTS quarterly under load. That means simulating a utility failure and watching the transfer actually happen. It takes about 15 minutes. The first time we did it, we discovered the switch had a 3-second delay configured where it should have been 0.5 seconds. That's not a life-safety issue for our application, but for someone with medical equipment, three seconds of downtime could be catastrophic.
The Relay You Never Tested
During the ZTS investigation, our electrician mentioned something that stuck with me: "When's the last time you tested the relay?"
I had no idea. I didn't even know which relay he meant. But here's what I picked up: you don't need a multimeter to do a basic how to test a relay without a multimeter check. Three things will catch most failures:
- Listen for the click. When you energize the relay (or simulate the condition that should trigger it), you should hear a distinct mechanical click. No click = no action. That's a free test.
- Visual inspection. Disconnect power and look at the contacts. Pitting, burn marks, or discoloration means the relay has been arcing—either it's failing or the load is too high for the contact rating.
- Load test. Bypass the relay manually and see if the load works. If it does, the relay is the problem. If it doesn't, the relay isn't the (only) problem.
That's it. No meter required. I'm not saying a multimeter isn't useful—it is, for voltage and continuity checks. But for "is this relay going to do its job when I need it," the click, the visual, and the load test catch most issues.
We found two failing relays in our transfer switch panel that year. Neither had set off any alarm. Both would have failed under real load.
What That Mistake Actually Cost
Let's talk numbers.
The two swollen batteries: $480 each for emergency replacement (standard pricing would have been $320 if we'd planned ahead). The grinding fan in the third unit: $180 for the replacement assembly plus $150 in labor because we had to take the unit offline during production hours.
But the real cost was the downtime. That 47-minute outage on the critical circuit? It took down a client-facing application that our account manager had promised would have "99.9% uptime." We missed the SLA. The apology email was embarrassing. The service credit was $2,100.
Total: $3,200 and change. Plus a very uncomfortable meeting with our operations director.
Here's the thing that still kills me: none of this was a battery problem. The batteries failed because the units overheated because the filters were clogged. The uptime failure happened because the transfer switch wasn't tested. The random reboots came from an ES 700 on equipment that needed a pure sine wave.
Every single one of those failure points was preventable with about 15 minutes of quarterly maintenance.
What I Do Now (The Short Version)
I'm not going to pretend this is some revolutionary checklist. It's boring. But it works.
Quarterly:
- Check and clean (or replace) UPS intake filters. Same schedule as HVAC filters. If the environment is dusty, check monthly.
- Test the transfer switch under load. Not a voltage check—an actual transfer simulation.
- Listen to every relay in the power path. If you don't hear a click within 2 seconds, investigate.
- Review battery health via the UPS management software (we use APC's PowerChute for the Smart-UPS units). Look for impedance trends, not just "pass/fail."
Annually:
- Full load test on all UPS units. Confirm runtime meets spec.
- Thermal imaging of terminal blocks and relay contacts to catch hot spots before they fail.
- Review the sine wave compatibility of every connected load. Anything with an active PFC power supply should be on a pure sine wave unit.
That's it. No complicated software. No expensive monitoring system. Just a filter check, a load test, and a click test.
And yes, I know how this sounds. "Facilities guy discovers basic maintenance—news at 11." But here's what I've learned the hard way: everyone checks the batteries. Nobody checks the filters, the transfer switch, or the relays. And those are the things that actually cause the failures you remember.
Bottom line: if your UPS maintenance plan starts and ends with "check the battery light," you're one dusty intake filter away from a $3,200 mistake.
Leave a Reply