APC Smart-UPS Online vs. Eaton 9PX: Sizing by Real Watts—Where the Number Hits the Load

You read the VA number and the watt number. But the ratio between them—the output power factor—isn't just a spec. It decides how hard you can actually load a UPS before the inverter clips, and that changes what you buy. The APC Smart-UPS Online (SRT) and Eaton 9PX both claim 0.9 PF on many models, but one of them shifts to unity PF in the sizes that matter most for edge racks. Here is how that breaks down, in four dimensions, using real numbers from the respective datasheets.

1. Power Factor: The Magnitude of the Derating Trap

APC UPS's 6–10 kVA SRT models are rated at Unity output power factor (1.0 PF), meaning a 10 kVA unit delivers a full 10,000 W of continuous wattage . Eaton UPS's 9PX across the same range (e.g., 10 kVA) is rated at 0.9 PF, so the same nominal 10 kVA gives you 9,000 W . That is a 1,000 W difference—10% more usable power from the APC in the 6–10 kVA band. The mechanism is not a software trick: the APC's inverter stage is designed with a higher crest-factor capability and output switching that can source current at unity phase angle without de-rating. The consequence is that for the same VA space (same rack-U footprint, same input breaker sizing), you can run a larger real load on the APC. The inverter does not clip until 10 kW. On the Eaton at the same VA, you must cap at 9 kW, or you get nuisance overload transfers to battery .

When this flips: For loads that are highly reactive (e.g., older linear PSUs with very low PF, say 0.6–0.7), the APC's unity PF rating confers no advantage—the reactive component still loads the inverter kVA. In that case, both units behave similarly, but that load type is rare in modern IT equipment, which typically presents PF 0.95–0.99.

2. Density: Watts per Rack-U—5400 W in 3U vs. the Same?

Eaton quotes the 9PX at up to 5400 W in a 3U form factor . APC's SRT in the 5 kVA band (5,000 W at unity PF) fits in 2U for the slim-line models, but the 10 kVA unit (10,000 W) occupies 4U . The ratio: Eaton puts ~1,800 W/U (5400/3), APC puts ~2,500 W/U (10,000/4). That is a 39% higher power density in APC's largest single-chassis unit. This matters because the limiting resource in an IT rack is vertical space. If you are stacking UPS in a 42U rack, every U saved on the UPS goes to compute or storage. Under the same load of 9,500 W, you could use one APC 10 kVA in 4U, versus one Eaton 9PX 10 kVA (still 9,000 W usable) in 3U—but the Eaton would not carry the load; you'd need the next size up (11 kVA), which uses more space. The magnitude of the density advantage can change the rack count.

When this flips: For installations where the UPS sits on the floor outside the rack (e.g., tower config in a wiring closet), density doesn't matter at all. The comparative U-count is irrelevant.

3. Efficiency at Partial Load: The Hidden Mass of Losses

APC's SRT in double-conversion mode is rated at ~94–96% typical; in Green Mode (line-interactive with bypass), it claims up to 98% efficiency . Eaton's 9PX is ENERGY STAR qualified and quotes up to ~96% in double-conversion mode, but does not publicly list a bypass or eco-mode with a separate efficiency curve . The worked consequence: at a typical 30% load (e.g., 3,000 W on a 10 kVA unit), the APC's double-conversion loss is roughly 150–180 W (5–6%), while Eaton at the same load is similar. But if you run the APC in Green Mode, loss drops to about 60 W (2% of 3,000 W)—a saving of ~120 W per UPS. Over a year (8,760 hours) at $0.12/kWh, that's about $126 per UPS per year. For a rack with two UPS, that's $252/year. The magnitude is not huge in absolute dollars, but in terms of heat removed from the room, 120 W less heat per UPS means less cooling tonnage. For a 10-rack row, it's measurable.

When this flips: Green Mode is not recommended for all loads—some medical devices or sensitive network gear don't tolerate the ~2–4 ms transfer when the inverter re-engages. Eaton's 9PX does not advertise a bypass mode with a separate efficiency curve, so it essentially forces double-conversion operation for sensitive loads. If the load is truly critical (no-break), then the APC's Green Mode advantage is not applicable; you run double-conversion on both.

4. Runtime at Sizing Limits: The 9 kW Case Study

Assume a real load of 9,000 W. With APC SRT 10 kVA (unity PF), you are at 90% of the unit's capacity . With Eaton 9PX 10 kVA (9,000 W max), you are at 100%—no margin . To get margin on Eaton, you step to the 11 kVA model (9,900 W max), which adds cost and space. The runtime on internal batteries for a 9,000 W load is proportional to battery capacity. APC's internal battery in the 10 kVA SRT (model SRT10KXLI) provides about 5 minutes at full load. Eaton's 9PX 10 kVA internal battery provides roughly similar ~5 minutes at full load. But because the Eaton at 9 kW is at 100% loading, any battery aging or temperature rise that reduces capacity by 10% will cause the UPS to drop load sooner. The APC at 90% loading has a 10% headroom buffer. The magnitude proportion here is the margin—10% of 9 kW = 900 W of safety margin.

When this flips: If you already oversize (e.g., using a 15 kVA unit for a 9 kW load), both units will run at ~60% load, and the margin difference disappears. But that's a different purchase decision.

Comparison Table: Key Numbers at a Glance

Rule of Thumb for Sizing

If the real load is between 9,000 W and 10,000 W, the APC SRT at 10 kVA works. For Eaton, you must go to 11 kVA. If the load is under 9,000 W, both can size at 10 kVA, but the APC provides a 10% margin while Eaton provides 0%—meaning any battery aging or surge puts you at risk. The threshold: if the load exceeds 90% of the Eaton's 9,000 W limit, select APC or step up the Eaton VA rating. This is a decision rule, not a generality.