Sizing a UPS Isn't One Question: Four Scenarios and How to Tell Yours Apart

I get some version of this question about four times a month: what UPS should I buy? And I've stopped giving a single answer, because there isn't one.

The question hides three or four different questions. A Smart-UPS 1000XL sitting in a wiring closet with one switch and one NAS isn't solving the same problem as a 3000VA unit on a dedicated 30A circuit holding up a production line. And neither of those has anything to do with the person calling me about an RV inverter generator.

For context on where I'm coming from: I'm the quality and compliance manager at an electrical distributor. I review incoming spec sheets and field install reports before they reach a customer — roughly 300 items a year. (280, maybe. I'd have to pull the system.) I've rejected about 12% of first deliveries this year, mostly for label and spec mismatches rather than failed hardware.

So here's how I break it down now:

• Scenario A — one closet, a handful of loads
• Scenario B — a room that has to stay up
• Scenario C — mobile or off-grid power
• Scenario D — panel-level, generator-backed

I'm not a licensed electrician, so I can't speak to code compliance on your specific install — that's your AHJ and your electrician. What I can tell you is where installs fail review, and what I'd specify.

Scenario A: One Closet, a Handful of Loads

Switch, firewall, a small NAS, maybe a couple of PoE injectors. This is Back-UPS territory, or the low end of the Smart-UPS line. Honestly, the model barely matters here. What matters is that you size for the load you'll have in three years, not the load you have today.

The thing people underestimate is the battery. The 1000XL designation means extended runtime, which means external battery packs racked or stacked alongside the main unit. Each of those packs has its own sealed cartridge, and they do not age identically.

I've walked into closets where the internal cartridge was replaced in 2022 and the external pack was the original from 2016, still reporting healthy on the display. The label lies. A runtime test doesn't.

Two things I'd put in any replacement procedure:

1. Write the install date on the cartridge in Sharpie before it goes in. Date codes on the housing are fine, but nobody reads them.
2. Replace on schedule, not on failure. Lead-acid in a 90°F closet does not get you the five years on the datasheet — three is more realistic. Battery capacity roughly halves for every 15°F above 77°F, which is a well-established rule of thumb in the industry.

Also budget the calibration. Some Smart-UPS models recalibrate runtime on their own after a battery swap; others will keep reporting the old runtime until you run a deliberate discharge test through the management software. Skip that and your monitoring dashboard becomes decorative.

Budget: a 1000XL runs roughly $700–900 street, and a replacement cartridge around $150–250 (based on distributor quotes, mid-2025 — verify, these move).

Scenario B: A Room That Has to Stay Up

This is where the Smart-UPS 3000VA lives, and this is where I'd push back on the most common piece of advice I hear: just add up your watts and round up.

It's tempting to think 3000VA means 3000 watts. It doesn't. Depending on the model, the watt rating on a 3000VA unit lands somewhere between 2,100 and 2,700W — read the nameplate, don't do arithmetic on the VA number.

Then there's the part that actually kills installs: 3000VA at 120V is 25A. That means a 30A circuit and, on most tower units, an L5-30P plug. If you bought it to sit where a 20A circuit already exists, you're pulling wire. In my experience, more than half of the 'the UPS doesn't work' tickets we get are circuit problems that got blamed on the UPS.

If you're deciding between line-interactive and online double-conversion at this size, the question isn't reliability — it's transfer time. Anything with a transfer gap measured in milliseconds is fine for switch-mode power supplies. It's not fine for certain lab equipment, older medical devices, or anything where a 4–6ms gap trips a relay. That's a downstream-load question, not a UPS question.

One physical note: the rack versions with battery packs are heavy, and the packs are usually the heaviest single piece. Plan two people. I've seen a single tech try to seat a battery pack solo and drop it — the pack survived, the flooring didn't.

Scenario C: RV Inverter Generator (This One Surprises People)

Here's where the common advice is backwards. The standard line is that inverter generators produce clean power, so they're safe for electronics and you don't need to worry. The waveform usually isn't the problem.

The problem is the neutral-ground bond. A large share of portable inverter generators ship with a floating neutral — the generator's neutral isn't bonded to its frame or ground. Plug that into an RV with a surge protector or energy management system, and the EMS reads it as an open ground and refuses to pass power. Or it throws an error code and you spend an hour troubleshooting a generator that's working exactly as designed.

Second issue: a UPS downstream of a generator. The UPS doesn't know the difference between a generator and a brownout. Frequency wander and voltage sag on a small generator will make a UPS switch to battery, drain itself, then switch back — over and over. If you need a UPS behind a generator, look for a wide input window, test it under real load, and be prepared to lower the input sensitivity via the management interface. Sizing the generator at 2–3x the UPS load helps too.

And the thing I'll be blunt about: do not backfeed a panel through a dryer outlet. I've had two customers describe that setup to me this year as though it were clever. It isn't.

Scenario D: Panel-Level Switching — 3 Pole Contactor, 120V Coil

Different animal entirely. If you're doing generator-backed transfer or load shedding at the panel, you're specifying a contactor, and the specs that trip people up are boring ones.

Three poles means it switches three conductors. That could be three phases, or two hots plus a switched neutral, or three independent single-phase circuits. Those are not the same application, and the third pole isn't automatically a spare — check whether all three poles carry the same rating before you use one as a passthrough.

120V coil means the control circuit has to actually be 120V. This is where a surprising number of installs go wrong: somebody pairs a 24V control transformer with a 120V coil and then can't figure out why the contactor never pulls in, or why the coil gets hot and dies. Match the coil to the control voltage, every time, and label it inside the enclosure.

The other gotcha is coil burden. Sealed (hold-in) VA is small, but pull-in VA can be several times that. String three contactors off one control circuit and the first two may pull in fine while the third chatters. Add up the inrush, not the steady-state draw.

Before You Sign Off: What a Voltage Tester Will and Won't Tell You

Whatever scenario you're in, somebody has to verify the work. Here's how to use a voltage tester without getting lulled into a false sense of safety.

A non-contact tester is good for is this dead? It is not reliable for is this actually dead? Long runs pick up capacitive coupling — ghost voltage — and the tester chirps on a conductor that has no real source behind it. Test against a known live source before you trust it, and again after. That second check is the one people skip.

With a two-lead meter, high impedance is the enemy. A standard meter on a long unloaded run will read phantom voltage that collapses the moment you put a load on it. Use a low-impedance mode if your meter has one, or add a small load and re-measure.

Then check places people forget: between neutral and ground, not just hot to neutral. A few volts of potential difference there points at an open or shared neutral, which is a problem no UPS is going to fix.

And on three-phase work, a voltage tester tells you nothing useful about phase rotation. That needs a rotation meter. I've seen a motor spin backwards because somebody verified voltage and called it done.

How to Tell Which Scenario You're Actually In

Three questions, in this order:

1. How long does it have to run? Seconds to ride out a blip is a closet problem. A few minutes to shut down cleanly is a server room problem. Hours is a generator problem, and the UPS is only there to bridge the gap while the generator spins up.

2. Who maintains it? If the answer is nobody, buy fewer, larger units with hot-swap batteries and remote monitoring. Five small units nobody checks are worth less than one unit that emails you when a battery degrades.

3. What downstream can't tolerate a transfer? Motors with contactors, older medical equipment, RV energy management modules, anything with a compressor. If you can't name it, you probably don't have one — and you can stop over-specifying.

One last thing, and it's where I've changed my own practice. Five years ago I sized almost everything with the old divide-by-1.4 shortcut — figure your watts, multiply by 1.4, that's your VA. That shortcut assumed a power factor around 0.7, which was reasonable when everything had a conventional supply in it. Modern switch-mode supplies with active power factor correction run at 0.95–0.99. The math now over-provisions in the other direction, and I've watched people buy a size up for no reason. The fundamentals — runtime, load profile, maintenance — haven't changed. The arithmetic has.

If you're stuck between two scenarios, spec to the harder one. Upgrading a closet later is annoying. Explaining why the server room went down is worse.