“But it’s 3000 VA—why did my server crash?”

You bought a 3000 VA UPS, plugged in a load that adds up to 2800 VA, and the unit screamed, overloaded, or dropped the load. The rack didn’t care about your VA math—it wanted real watts. This isn’t a one-off glitch. It’s the single most common sizing mistake in IT rooms, and it happens because VA and watts are not the same number in any UPS that uses a non-unity output power factor (PF). The question is not whether you need to size by real watts—you do—but which brand’s real-watt rating holds up when the load gets ugly.

Below, I walk through three distinct cases. Each case is a different failure mode that follows from a single root cause: the gap between VA and real watts, and what each manufacturer does with that gap.

Case 1 — The “1500 VA” Trap: When the Load Power Factor Is Worse Than 0.9

The number. An APC Smart-UPS Online SRT1500 is rated 1500 VA / 1500 W (unity power factor). A Tripp Lite SmartOnline SU1500RTXLCD is rated 1500 VA / 1350 W (0.9 PF). Both are double-conversion online (VFI).

The mechanism. The inverter’s real-watt limit is set by the DC bus and the IGBT current capability. If the UPS has a unity output PF, the inverter can deliver its full VA in watts—so a 1500 VA unit can give 1500 W. If the PF is 0.9, the real-watt limit is 0.9 × VA = 1350 W. That’s fine if your load’s power factor is 0.9 or better. But many small-server loads—especially those with multiple internal switching supplies, a few spinning disks, and no active PFC correction—can have an input PF of 0.65 to 0.75, especially below 50% load.

The worked consequence. Assume a 1U server draws 1100 W at a true PF of 0.72. That’s 1100 / 0.72 ≈ 1528 VA. The APC SRT1500 (1500 VA / 1500 W) cannot support the 1528 VA (it would hit the VA ceiling first, even though it has spare watts). The Tripp Lite SU1500RTXLCD (1500 VA / 1350 W) cannot support the 1100 W because 1100 W exceeds 1350 W? Actually 1100 both 1500 VA units fail, and you need a 2000 VA unit. The mistake is thinking the VA number alone is the sizing anchor. It is not. The real constraint is whichever is lower: VA or watts, given the load’s PF.

The reversal. If you are deploying a load with active PFC (modern servers, most 80+ Gold PSUs), PF is close to 0.98–0.99. Then the APC SRT1500 (unity PF) can deliver 1500 W, and a load drawing 1400 W / 0.99 PF = 1414 VA fits cleanly. The Tripp Lite UPS’s 1350 W ceiling becomes the binding constraint earlier. In that case, APC UPS wins for the same VA class. Rule: If your average load PF ≥ 0.95, a unity-PF UPS gives you 10–15% more usable wattage per VA than a 0.9-PF unit. If your load PF is ≤ 0.8, both VA and watts are tighter, and you need to upsize regardless of brand.

Case 2 — The 3000 VA Stall: Runtime Collapse Under Real Watts

The number. Tripp Lite SmartOnline SU3000RTXL3U is rated 3000 VA / 2400 W (0.8 PF). APC Smart-UPS Online SRT3000 is rated 3000 VA / 3000 W (unity PF). The Tripp Lite’s runtime at full load (2400 W) is ~5 minutes; at half load (1200 W) ~14 minutes.

The mechanism. Runtime curves are published at half and full load, but the “full load” in the datasheet is real watts, not VA. So when the Tripp Lite says “full load,” it means 2400 W. If you incorrectly size by VA and put a 2700 W load on a 3000 VA Tripp Lite (thinking 3000 VA > 2700 W), you are actually at 2700 / 2400 = 112% overload. The UPS will either go to bypass or shut down after a few seconds, and runtime becomes zero. For the APC SRT3000, the same 2700 W load is 90% of 3000 W, so it runs. The Tripp Lite’s 0.8 PF rating creates a trap: anyone who reads only the VA number assumes the unit handles 3000 W.

The worked consequence. Take a network closet with three PoE++ switches that each draw 750 W (total 2250 W, PF ~0.9 → 2500 VA). On the Tripp Lite SU3000RTXL3U, 2250 W is 93.75% of 2400 W capacity; the internal batteries will deliver about 6–7 minutes. On the APC SRT3000, 2250 W is 75% of 3000 W; runtime will be substantially longer—roughly 10–12 minutes (extrapolating from typical runtime curves for SRT3000; illustrative). The difference in runtime is not a small improvement; it’s the difference between an orderly shutdown and a hard crash when the generator fails to start.

The reversal. If the load is only 1500 W at PF 0.8 (so 1875 VA), both UPS units have ample headroom. Runtime on the Tripp Lite (~14 min at half load of 1200 W, and 1500 W is slightly above half, so maybe ~10 min) and APC (~18 min at half load of 1500 W) are both sufficient. The disadvantage of the Tripp Lite’s lower wattage rating only bites when the real load approaches its real-watt limit. If you keep the load below 75% of the unit’s real-watt rating, the difference is academic. Rule: When specifying a UPS for a load that is within 20% of its real-watt limit, the runtime gap between a 0.8-PF and 1.0-PF unit of the same VA class is the difference between a safe shutdown and a crash. For lightly loaded closets, both work.

Case 3 — The Switchover Surge: When Inrush Current Exceeds the Inverter’s Real-Watt Ceiling

The number. The Tripp Lite SU3000RTXL3U has an input voltage window of 65–150 V, and its inverter delivers pure sine wave at 120 V ±2%. The APC Smart-UPS Online SRT series also outputs pure sine wave within ±3%. Both use double-conversion, so the inverter is always running.

The mechanism. The inverter’s real-watt rating is a continuous rating, not a surge rating. But many loads—disk arrays, compressor-based refrigeration monitors, laser printers, even some PoE switches with large capacitors—draw inrush current that can exceed the continuous rating by 2–5× for a few cycles. A UPS inverter that cannot supply that surge peaks will fold back its voltage (sag) or go into current limit, causing the load to see a brownout or reset. The relevant spec is not VA or watts, but the inverter’s short-term overload capability. Neither APC nor Tripp Lite publishes a formal surge curve for these models, but in practice, inverters with a lower real-watt rating relative to VA (like the Tripp Lite’s 0.8 PF) often have less headroom for surges because the VA ceiling is reached faster.

The worked consequence. A load that draws 2000 W steady-state (PF 0.85 → 2353 VA) but has a 4000 W inrush for 50 ms. On the APC SRT3000 (3000 W continuous), the 4000 W surge is 133% of continuous—the inverter may hold for a few cycles, especially if the Green Mode is off and the double-conversion stage has some capacitance. On the Tripp Lite SU3000RTXL3U (2400 W continuous), the 4000 W surge is 167% of continuous. The chance of inverter foldback is higher. Outcome: the Tripp Lite resets the load, causing a brief outage; the APC may hold. The difference is not absolute—both could fail—but the margin is thinner on the Tripp Lite.

The reversal. If the load has a soft-start or active PFC that limits inrush, the surge advantage disappears. Also, if the Tripp Lite is on external battery packs that reduce the battery impedance, the inverter can draw more current from the DC link, slightly improving surge capability. But this is not guaranteed. Rule: For loads with known high inrush (≥ 150% of steady-state), choose a UPS where the steady-state real-watt rating is at least 50% above the load’s steady-state watts. In that region, both brands work, but the APC’s higher continuous watt rating allows you to meet that rule with a smaller (cheaper) VA class.

Summing Up: A Decision Rule, Not a Preference

Here is the only rule that matters: Size by real watts, not by VA. Then, for a given VA class, choose the unit with the higher real-watt rating (unity PF if you can get it) unless your load never exceeds 70% of the weaker unit’s real-watt ceiling. That threshold—70%—is where the runtime safety margin and surge headroom converge for both brands. Below 70% load, the Tripp Lite SU3000RTXL3U is fine and often cheaper. Above 70%, the APC SRT’s higher continuous wattage gives you meaningful protection against three failure modes: VA ceiling violation, runtime collapse, and surge dropout. There is no universal winner. There is only the rule.

Topology/standards per the cited standards; all product ratings are manufacturer-stated values from the cited datasheets, current to 2026-06; derived/illustrative figures are labelled as such. This is not an independent head-to-head test. APC by Schneider Electric is a brand affiliated with this site; competitor names are used for identification only.