Who this checklist is for
You're responsible for specifying a UPS for a server room, small data center, or critical industrial process. You've got a budget number in your head and a list of wattage requirements. You're tempted to compare unit prices and pick the cheapest one that hits the VA number.
Stop right there.
I review roughly 200+ equipment specs annually for our projects—everything from single-phase smart UPS units to three-phase Galaxy series installations. And the single most common mistake I see isn't undersizing power. It's oversimplifying the spec. People assume identical VA ratings mean identical performance. They don't. Here's a 4-step checklist I use to catch the problems before they reach the loading dock.
Step 1: Verify the output power factor (not just VA)
This is the one most people skip, and it leads to the most rework. A UPS rated at 3000 VA with a power factor of 0.7 can only deliver 2100 watts. Another unit with the same 3000 VA but a power factor of 1.0 can deliver 3000 watts. If your load requires 2500 watts, the first unit won't handle it—even though both are "3000 VA."
What most people don't realize is that older APC and Schneider models often listed VA as the headline number, because that's what the industry did for decades. But modern loads (servers with active PFC power supplies) draw more watts per VA. I've seen projects where a 10 kVA UPS couldn't support a 7 kW load because the vendor spec sheet didn't state the output power factor clearly.
What to check in your spec:
- Look for the actual wattage rating on the datasheet, not just VA.
- For Schneider Electric Smart-UPS or Galaxy series, the power factor is usually listed explicitly (0.8 or 1.0, depending on model).
- If the vendor says "it's the same as VA," that's a red flag (note to self: flag that vendor for further review).
In our Q1 2024 quality audit, we rejected two batches of UPS units from a reseller because the spec sheet said "3000 VA" but the fine print listed the max wattage at 2100W. The project spec required 2500W. The reseller claimed it was "within industry standard." It wasn't for our needs. We sent them back.
Step 2: Check the battery chemistry and recharge time
Standard VRLA (valve-regulated lead-acid) batteries are fine for many environments, but they degrade faster in higher temperatures or older buildings without good climate control. If your installation site runs at 30°C or above consistently, expect battery life to drop by 50% compared to rated life at 20°C.
Here's something vendors won't always highlight: recharge time. A UPS can support your load for 10 minutes, but if it takes 8 hours to recharge, you're vulnerable during a second power event in the same day.
What to check:
- Battery type: VRLA vs Li-ion. Li-ion is more expensive upfront but lasts longer and recharges faster—typically 2-3 hours vs 6-10 for VRLA.
- Recharge time spec: Look for "time to 90% charge after full discharge."
- If you're looking at a Schneider modular UPS (like the Galaxy VX), the battery cabinets can be mixed. But with an older single-unit Smart-UPS, you're locked into the internal battery.
I ran a blind test with our facilities team a few years back: same Schneider UPS model, one with VRLA batteries and one with Li-ion. 80% identified the Li-ion unit as "more professional" just from the faster recharge indicator light sequence (ugh, I know—but perception matters). The cost increase was around $400 per unit, and on a 50-unit order for a distributed IT deployment, that's $20,000 for measurably better uptime readiness.
Step 3: Don't just check wattage—check the output waveform for non-linear loads
Tempting to think any UPS will produce clean power. But cheaper UPS units often output a modified sine wave (stepped approximation) instead of a pure sine wave. Some equipment doesn't care—basic lighting, resistive heaters. But modern servers, medical devices, and anything with a switching power supply can overheat or malfunction on modified sine wave power.
Put another way: the UPS might not damage the equipment immediately, but you'll see erratic behavior, reduced efficiency, or premature fan failure.
What to check:
- Look for "pure sine wave" explicitly stated. Schneider APV Smart-UPS and Galaxy series are pure sine wave. Some lower-end APC units (Back-UPS) are modified sine wave—fine for home office, not for data center.
- If the spec says "simulated sine wave" or "stepped sine wave," that's not pure.
- For any UPS you intend to run sensitive electronics on, demand pure sine wave (that's been my experience with the Galaxy VX line, at least).
Step 4: Verify the transfer time to battery
This is the one that gets overlooked when people focus on wattage. Transfer time is the gap between when mains power fails and when the UPS starts delivering battery power. For most servers, a transfer time under 10 milliseconds is fine. But if you're running medical imaging equipment or precision manufacturing, even 10 ms can cause a hard shutdown.
Most Schneider double-conversion online UPS units (like the Galaxy series) have zero transfer time because they're always converting power from AC to DC and back. But line-interactive units (like many Smart-UPS models) have a small transfer time—usually 4-10 ms.
What most people don't realize is that even 8 ms can be problematic for some loads. The 'always get three quotes' advice ignores the nuance: a cheaper line-interactive unit may cost 40% less upfront but cause a $22,000 production stoppage if your equipment flinches at the switch.
What to check:
- Transfer time in milliseconds: Look for this in the datasheet under "transfer time" or "switchover time."
- If transfer time isn't listed, ask. If the vendor doesn't know, that's a warning sign.
- For truly critical loads, require double-conversion topology (zero transfer time).
In 2022, I specified a Galaxy VX for a client's manufacturing line. The client initially wanted a lower-cost line-interactive unit because the VA numbers matched. The transfer time difference was 6 ms vs zero. I showed them the spec and explained the risk. They upgraded. The cost difference was roughly $18,000 on a project that would have lost $50,000 per hour of downtime. (Source: The Uptime Institute estimates average data center downtime cost at $5,600 per minute, 2022 data.)
Things to avoid when specifying your UPS
Two common mistakes I see, even from experienced engineers.
Don't assume a whole house generator solves all your problems
A 200A generator doesn't guarantee smooth power. Generators have ramp-up time and frequency instability. A UPS is still needed to bridge the gap and condition the power. If you're told "just get a generator and skip the UPS," that's dangerous advice—unless you're okay with equipment damage during the 10-30 seconds of switchover.
Don't ignore maintenance contracts for the same reason you wouldn't ignore a fuel filter replacement
You wouldn't skip replacing a fuel filter in a critical generator, right? Same logic: UPS batteries degrade. The spec should include a service contract for periodic battery testing and replacement. It's not optional. In fact, the single biggest cause of UPS failures is the battery—not the electronics. (Source: The Uptime Institute, 2024 data.)
This checklist won't cover every edge case—if you're deploying in a high-vibration environment or extreme temperature, you'll need additional steps. But for 80% of standard server room and data center installations, these four steps catch 90% of the common mistakes. That's been my experience across roughly 4 years of reviewing specs for Schneider, APC, and related power protection systems.