It was 11:40 on a Tuesday morning in February 2023 when the office went dark. Not a flicker—four hours of nothing. Our switches powered down, the VoIP phones went silent, and the forecast model I'd been building since 6 a.m. never saved.
The next morning, our CEO forwarded a one-line email from the operations director: "We need backup power. Give me options by Friday."
That's how I became the reluctant owner of our office's power resilience project. I'm a procurement manager—six years tracking a $180K annual facilities budget, 60+ vendors, a TCO spreadsheet with 2,400 rows. Power systems were not my category. But when the CEO says Friday, you learn a category fast.
What I got wrong (and what the manual would have told me)
My first call was the obvious one: a UPS for the network closet. I compared three brands over a week. APC was the safe default, Eaton was the enterprise option, and Tripp-Lite was the value pick. I'd used Tripp-Lite rack units at a previous job—they worked fine. Their site has a selector tool, a runtime calculator, and the kind of spec sheet a cost guy appreciates.
I picked the Tripp-Lite Internet Office 900VA UPS. Price: $240 from one reseller, $255 from another. Our network closet draws roughly 450W at steady state and about 480W when the PoE switch is fully loaded. The spec page said "900VA"—and I filed that as "covers 900 watts."
It doesn't.
According to Tripp-Lite's own product page for the Internet Office series (tripplite.com, accessed January 2024), the 900VA model is rated 900VA / 460W. VA is apparent power; watts are real power. Power factor is the conversion, and for office IT gear it's usually between 0.5 and 0.6. That's not a Tripp-Lite quirk. APC, Eaton, CyberPower—they all quote the same way. It's a standard you learn once and then never forget.
I learned it the expensive way. The UPS arrived, I plugged everything in, powered up the closet, and the LCD flashed OVERLOAD within forty seconds. The alarm is roughly the pitch of a smoke detector and about as popular with the IT director.
Here's the part that stings: the answer was in the manual. After the overdue alarm incident, I downloaded the Tripp Lite Internet Office UPS manual and read the specifications section. Page 34—give or take, I don't have it open right now—spells out VA vs. watts, load-level calculations, and runtime at 50%/75%/100% load. The data was all there. I just hadn't asked the right question when I was shopping.
We returned the 900VA, ate $38 in return shipping (the reseller's policy), and bought the Tripp-Lite SmartPro 1500VA (1500VA / 900W) for $469. Net cost of that mistake: about $267 including the shipping. That's a lot of money to pay for a lesson on power factor.
Extending the timeline: from a UPS to a dual-fuel generator
The UPS solves a ten-minute problem. Our CEO wanted to solve a twelve-hour problem. A UPS buys you time to shut down cleanly or ride out a blink—it does not run a switch closet for half a day. Trust me on that one too.
So we moved into generators. I'd never spec'd one. The two options the facilities team brought forward were a standby generator hardwired into the building (pricier, needed landlord approval) and a portable dual-fuel generator with a manual transfer switch. We also got pitched on a backup generator with solar panels—a combination of a battery bank, PV panels, and a generator as a tertiary backup.
Let me be honest about that last one: for a 25-person office in a leased building, a backup generator with solar panels didn't pencil out. Quoted cost was $14,000–$18,000 for a meaningful system, and our lease prohibited rooftop modifications. Payback was in the fifteen-plus-year range before we even got to landlord negotiations. If you own your building and have a longer horizon, the math changes. For us it didn't.
We ended up with a 7,500W dual-fuel generator (gasoline/propane) and a manual transfer switch. The generator was $1,150, the transfer switch was $680, and the electrician quoted $1,620 for install and inspection—total $3,450. Per NEC Article 702, optional standby systems require proper transfer equipment and disconnecting means, so I wasn't going to DIY this. The electrician's quote was line-itemed, which I appreciated, since I'd been burned on bundled quotes before.
The multimeter I almost didn't buy
During the install walkthrough, the electrician told me something I've since repeated to three other procurement managers: "Before you trust a transfer switch, verify the phases. Don't guess." Our maintenance lead didn't have a meter that could handle both AC and DC safely. I didn't either.
So I did what I do—I went shopping. Where to buy a Fluke multimeter wasn't obvious to me. Grainger had the Fluke 117 in stock at $169. Home Depot's site had industrial meters but not the Fluke line. Amazon listed it around $145 from third-party sellers, and I didn't want to risk gray-market stock on a calibration-sensitive tool. I bought the Fluke 117 from Grainger the same day.
Per Fluke's product page (fluke.com, accessed January 2024), the 117 is a True-RMS meter rated CAT III 600V—enough for the 240V circuits we needed to verify. That $169 felt expensive for about ten minutes. It paid for itself the first time we checked our transfer switch and found the neutral wasn't bonded correctly.
That's the thing about cost control: I hate spending money on tools we'll rarely use. But I hate paying electricians to come back more.
What actually happened during the July storm
In July 2023, a derecho knocked out our grid power for about five hours. Here's the sequence, because it's the part that convinced our CEO the budget was worth it:
- Grid dropped at 14:12. The SmartPro 1500VA carried the network closet for 14 minutes—maybe 13, the runtime curve flattens out at that load level—right in line with Tripp-Lite's published numbers at ~60% load.
- Our maintenance lead walked out, started the dual-fuel generator on propane (we keep two 40lb tanks), and flipped the manual transfer switch.
- Load transferred at 14:29. Nobody noticed. The phones kept ringing, the switches kept switching.
But here's the honest part: the story didn't end cleanly. When the grid came back at 19:24 and we transferred back, the UPS threw a "battery fault" warning. I pulled out the Fluke 117 and checked the battery string. Two of the four 12V blocks were reading 10.6V under load—end-of-life territory. We replaced the battery pack for $95. Lead-acid batteries age. That's not a Tripp-Lite defect, that's chemistry. But nobody tells you the UPS is a maintenance item when they sell it to you.
So our total spend on this project ended up at about $4,400 after batteries and shipping. Not $240. Not even close. But every dollar after the first $240 was tied to something we actually needed and would have discovered anyway, just later and at a worse time.
Where this approach doesn't make sense
I've recommended this playbook to two other procurement managers since. I've also told two others not to bother. Here's the split:
If you just need to protect a couple of switches and a router (under 200W total), the 900VA Internet Office would have been perfectly fine. My problem was load, not the product. Don't overspend on capacity you'll never use.
If your building has a landlord-managed generator and you're on a floor with a central UPS room, skip the portable generator. You're adding complexity for a power problem someone else already solved. Just size a rack UPS for the transition time between utility drop and building generator start—usually 10–30 seconds.
If you're planning a backup generator with solar panels, be honest about your building situation and your payback window. The combination is fantastic for owned facilities with high outage frequency. It's a poor fit for leased space and low outage counts. I'd have wasted $14K on a solution our landlord would never have approved.
If your load is genuinely business-critical (data center, medical, 24/7 operations), a portable dual-fuel generator is the wrong tool. You want a fixed standby system, an automatic transfer switch, and probably N+1 UPS redundancy. Different budget. Different process. Don't let a $4,400 project set expectations for a $40,000 one.
The TCO math I should have run from day one
If I could redo the first two weeks, I'd have built the spreadsheet before opening the first vendor site:
- Load audit. Actual watts, not nameplate. Measured at the breaker with a clamp meter, not estimated from datasheets.
- Runtime requirement. How long do you need to bridge? 10 minutes (shutdown) or 10 hours (full operations)? The answer changes everything downstream.
- Power factor math. Convert VA to watts at the actual PF of your gear, not the round number that sounds friendly on a product page.
- Battery replacement schedule. UPS batteries are a consumable. Budget $80–$150 every 3–5 years per unit, whether you remember to or not.
- Install, inspection, and permit costs. A transfer switch isn't plug-and-play. My quote was $1,620 for a $680 part.
Also: check that the UPS itself is UL 1778 listed. This is basic safety equipment, not a premium feature. If a vendor can't confirm it, keep shopping.
None of that is glamorous. All of it is cheaper than a Saturday at the office waiting for the grid.
The CEO's "keep receipts" email has a happy ending: we spent about $4,400 total, saved roughly $1,200 per avoided outage (we've had two since July), and I've got a repeatable process for the next facility. But the $267 lesson on VA vs. watts is the one I'd put at the top of any buyer's guide.
Prices in this article are based on vendor quotes from January–July 2023 and Tripp-Lite, Fluke, and Grainger product pages accessed in January 2024. Verify current pricing and specifications before you buy—the numbers move.