The Most Expensive Part of Power Backup Isn't the Hardware
The most expensive line item in a power backup plan is almost never the unit on the purchase order. It's the shortcut you take when the spec sheet looks close enough, the deadline is Friday, and nobody has carved out an hour to check the load profile properly.
I handle IT operations and facility power procurement — nine years now, mostly mid-sized deployments. Server rooms, a couple of small data closets, the generator that keeps the building's life-safety gear alive, and everything in between. Over that stretch I've made (and actually wrote down) six power-related mistakes that cost roughly $16,000 — or rather, closer to $18,000 once I count the two weekends of unpaid labor I spent fixing them. Reorders, dead batteries, one very awkward conversation with a facilities director. The pattern in all six was identical: I optimized for how fast I could place the order, not for whether the specification was right.
Which is a strange place to end up, because for years I told myself the opposite. I thought I was being efficient by not overthinking small purchases. I was being fast in the one category where fast is expensive.
Where Close Enough Actually Costs You
1. Sizing a Tripp-Lite UPS on VA alone
In 2019 I ordered three Tripp-Lite UPS units for a server refresh. I looked at one number, the VA rating, matched it against the sum of the nameplate labels on the equipment, and called it done. Ordered, received, installed. It looked fine on the dashboard for about six weeks.
Then a UPS started throwing overload warnings during RAID rebuilds — a load condition that is not the steady-state load, which is exactly the point. The problem was not the Tripp-Lite UPS hardware. It was that I sized on apparent power (VA) and ignored real power (watts), and the power factor on that particular mix of gear was nowhere near unity. A unit rated 1500VA does not automatically deliver 1500W. If you're cutting it close on VA and the load power factor is low, you can hit the watt ceiling long before the VA ceiling.
What I should have done takes twenty minutes: pull the actual watt draw from equipment spec sheets or a meter, add margin for startup inrush, then run the vendor's selection tool. Tripp-Lite's online UPS selector does that math for you (their support site has it, and honestly, searching tripp-lite-ups gets you to the right page faster than their navigation does). I had apparently decided I didn't need it. That decision cost $1,900 in replacement units plus a Saturday.
The lesson was not buy bigger. The lesson was that apparent-power numbers are the easiest thing to grab and the least useful thing to plan around.
UL 1778 is the standard governing uninterruptible power supply equipment in the US, and it covers construction and safety — not whether your load math is right. Nobody is going to catch a sizing error for you. That's on the buyer.
2. The wrong charger for the wrong chemistry
Second mistake, and this one took longer to notice because the damage was quiet. We run a standby generator for life-safety circuits. Like most standby units, it has a small battery maintainer — the thing vendors loosely call a generac battery charger when it ships bundled with the generator package. Ours failed in spring 2022, and I replaced it with a general-purpose charger I had on the shelf. Physically compatible. Electrically wrong.
Charging is chemistry-specific. A flooded lead-acid 12V battery wants a float around 13.5 to 13.8 volts at 25°C, with temperature compensation — that figure shows up in pretty much every battery datasheet I've read. A lithium iron phosphate pack has a completely different profile: constant current, then constant voltage, and it does not appreciate being held on a permanent trickle. Use a battery charger for lithium ion battery packs on a lead-acid starting battery, or the reverse, and you don't get an immediate failure. You get a battery that quietly loses capacity, sulfates, dries out, or in the lithium case, ages years ahead of schedule.
Ours took about eleven months to die. It picked the morning of a scheduled generator load test to do it, which I suppose is the good version of that story (ugh, but lucky).
The replacement battery was $310. The generator service call to figure out why the battery wasn't holding was $260. The near-miss on a real outage isn't billable, but it's the number I actually remember.
3. Treating an ongoing problem like an emergency
This one is an analogy, but it's the cleanest way I can explain the third mistake. Ask anyone who works around vehicles about battery charger vs jump starter and they'll separate the two immediately: a jump starter delivers a burst of cranking current for a few seconds, a charger restores charge over hours. Different problems. One is for emergencies, one is for maintenance.
I made the power-infrastructure equivalent of leaving a jump starter wired to the battery. For two years we treated recurring UPS battery alarms as emergencies — swap the battery, close the ticket, move on. Nobody tracked battery age. Nobody asked why we were replacing the same strings every eighteen months. We were jump-starting a process problem and calling it responsiveness.
The fix was unglamorous: a battery replacement log, an age column, and a calendar reminder 60 days before the manufacturer's expected service life. When we finally moved to rack mount UPS Tripp-Lite units in the main closet, we deliberately picked a model with hot-swappable battery modules and front-accessible trays — mostly because the old rear-access setup was the real reason nobody wanted to do maintenance in the first place. Mounting depth and rail kits mattered more than I expected too. Measure the rack, not the room.
There's something satisfying about a maintenance schedule that runs itself. After two years of firefighting the same alarms, watching the replacement log fill itself in ahead of schedule was the best part of that whole project.
The Obvious Objection
Fair pushback: not every deployment deserves this level of rigor. If you're running a small office with one UPS under a desk, you don't need a specification review process. My experience is based on roughly 60 power-related purchases, and almost all of them sat on systems where downtime had a real cost attached. If you're buying for a home lab or a single workstation, a lot of what I just described is overkill, and I'll say so plainly.
But there's a version of we're too small to bother that gets aimed at the wrong purchases. The generator battery was not a data center asset. It was a $310 part on a system whose entire job was to work on the one day nobody was watching. The sizing error wasn't on the biggest UPS in the building — it was on three mid-range units. Scale didn't protect me. The checklist did.
And I'd push back on the idea that rigor is slow. The pre-order check I run now takes about fifteen minutes. Fifteen minutes against $18,000 is not a close call.
What I Actually Do Now
Before any power-related purchase goes out, four things get checked: real watt draw from measured or datasheet values rather than nameplate VA; chemistry match between every battery and every charger, written down; rack dimensions and access direction for anything rack-mounted; and a service-life date entered in the log at the moment of installation, not later.
That's it. It isn't sophisticated. It's the list I wish someone had handed me in 2017, and it exists because nobody did.
Tripp-Lite UPS hardware was never the weak link in any of these failures. My ordering process was. Efficiency isn't ordering faster — it's ordering once.
