I Thought Our UPS Was Fine. A $12 Multimeter Told Me Otherwise.

2026-11-10 · Elaine Zhou · Engineering note

The lights flickered at 7:40 AM on a Tuesday in November 2023. Then my Slack lit up.

"Server room UPS is alarming." From our facilities manager.

We didn't have a full power outage—just a sag the utility later called a "momentary interruption." But our small server room (eight racks running the front-of-house systems for about 60 users) went down for 11 minutes. The lobby display read SYSTEM OFFLINE. Three executives walked past it coming in.

I'm the office administrator at a 400-person company. I handle around 60–80 orders a year across 8 vendors—roughly $600K in annual spend. When something breaks in the building, I'm the one who has to figure out whether we buy, fix, or ignore it.

That Tuesday convinced me we couldn't ignore this one.

Starting with the batteries

Our UPS had been installed in 2018. The battery string was due for replacement in May 2023—but our service vendor had signed off with "looks fine." I didn't know enough to push back at the time.

So I did what any non-engineer does: I Googled how to test a battery with a multimeter. Turned out it wasn't complicated. Our facilities tech had a Fluke sitting in a drawer, and I had the afternoon.

The basics: for a 12V sealed lead-acid battery in a string, you test each one individually. A healthy cell reads 13.5–13.8V at rest. Anything under 12V is a red flag. Under load—which requires disconnecting the string and running a small load test—you want to see the voltage sag stay above 11.5V.

Out of 32 batteries, we found 9 reading below 12V. One cell was sitting at 10.7V. The string had been running on about 70% capacity for who knows how long. Our vendor never load-tested anything—just eyeballed the terminals and called it a day.

The tech also pointed out something I would have never caught: the battery charger plug on the backup string was hard-wired with an exposed connector that had been arcing internally. That's a fire hazard sitting inside a closet we walk past every day.

I should mention: I found most of this through the Schneider Smart-UPS community. I posted a question about string testing and got two replies within 48 hours—one from a data center tech, one from a facilities guy in Ohio who'd been running Smart-UPS since 2011. That forum saved me from overthinking the whole thing.

"Bottom line: if your battery string is more than four years old and you haven't done a load test, you don't actually know what capacity you have. Voltage at rest tells you almost nothing."

Now we're comparing modular UPS systems?

Replacing the batteries felt like putting a band-aid on a bigger problem. The UPS itself was 5 years old, the charger plug was compromised, and we'd had two smaller outages in the previous 14 months that I hadn't connected the dots on until now.

So I pulled two quotes: Schneider Electric (Galaxy VS, 40kVA modular) and Siemens (SITOP modular). I'd read comparisons of these two brands before, but seeing the actual numbers side by side on my desk was a different story.

Here's what jumped out: Siemens was about $1,800 cheaper. Not a huge gap, but my procurement instinct said cheaper wins if the specs are similar. I was on the fence for about a week.

Then a colleague—a former IT ops guy who now works on our internal tools team—asked me one question over lunch: "If a module goes down, are you sending the whole unit somewhere, or are they swapping just the part?"

I had no idea. So I called both vendors back.

  • Siemens: module-level repair, but the unit retrofitting process required shipping the power module back to their facility for exchange. Lead time on a replacement: 3–4 weeks.
  • Schneider: hot-swappable modules. They had two certified techs in our metro area, and standard response time was next business day on a module swap.

That was the deal-breaker for us. We're a business that can't afford 3 weeks of degraded capacity. And honestly, that's the kind of detail I wouldn't have known to ask about if I hadn't been nudged.

I later found out from one of the Schneider support engineers that the Galaxy series uses a standard modular architecture with external battery charger plug configurations—so replacing that hard-wired connector on the old string was basically a plug-and-play fix once the new unit was in.

The thing nobody warns you about: shared circuits

While we were at it, our facilities team started tracing every circuit in the building. Turns out our break room had an electric brewery control panel—company brewing club, long story—that had been sharing a 20A circuit with the server room for at least two years.

That explained why the controller had been resetting every time the lights flickered. We moved it to a dedicated circuit and added a small surge protector. The whole brewing club was very relieved, though mostly they were upset their amber ale batch had been interrupted.

The larger lesson for me: the UPS was the emergency problem. The shared circuit had been a slow-motion problem nobody had noticed.

The number that actually mattered

We went with the Schneider Galaxy VS. Final cost: about $37,400 including installation, new battery string, and the charger plug replacement.

The Siemens quote would have been closer to $35,600. So we spent roughly $1,800 more for faster module replacement, a local tech pool, and a documented upgrade path that doesn't involve shipping anything back to a factory.

That's not a huge premium for the kind of redundancy we need. But when I ran the numbers for my VP, I framed it simply: "The cost difference is one month of downtime minutes avoided." She signed off without a second conversation.

What I actually learned

Six months later, our tech whispered something to me in the hallway that stuck. "We haven't had a single alarm since the install."

That should be the goal, right? Not "we handled the outage well," but "there was never an outage."

Here's my honest takeaway, and it's probably not the one you'd expect from a procurement person.

The quality of your backup power is invisible until it isn't. Nobody notices a healthy UPS. They notice a dead server. They notice the lobby screen that says SYSTEM OFFLINE. They notice the client who scheduled a call at 8 AM and has to reschedule because your front desk can't pull up their account.

My job isn't just to buy equipment. It's to make sure the building presents the way our company actually is—reliable, professional, ready. When I skipped the load test in 2022 because the vendor said "looks fine," I was saving about $400 in service costs. That 11-minute outage probably cost more than that in lost billable time.

So, where does that leave me? A few things I'd tell anyone in my seat now:

  1. Test the batteries, not the paperwork. A $12 multimeter and an afternoon will tell you more than any service report.
  2. Ask how modules get replaced, not just how they perform. Downtime logistics matter more than spec sheet numbers.
  3. Check your circuits. You probably have something sharing a line with something important, and it's been fine for three years until it isn't.
  4. Find a community. The Schneider Smart-UPS forum didn't sell me anything. It just told me what to ask the vendors.

And one final thing: the multimeter was $38.99 from Amazon. The battery test took about two hours. That's the cheapest insurance we've ever bought.

Leave a technical question