How to read a UPS datasheet: VFI, kVA, PF, THD, MTBF and more explained

Put three UPS datasheets side by side and the comparison falls apart quickly.

One says VFI. Another quotes THDi below 3%. One advertises 99% efficiency. Another gives an MTBF above 100,000 hours. Then come kVA, kW, PF, THDv, overload percentages, battery voltage and bypass ranges.

None of those specifications is meaningless. The problem is the conditions behind them.

A 99% efficiency figure may be measured in ECO mode rather than in normal double-conversion operation. A wide input-voltage range may only apply at partial load. And "150% overload" says nothing until the manufacturer also states how long the inverter holds it.

So read a datasheet in a fixed order: topology and usable power first, then what the UPS draws from the upstream supply, what it delivers to the load, how it behaves during overloads and input failures, and finally how the battery and bypass are built.

Comparisons only mean something once the class and the kW ceiling match. Every other line on the page depends on those two.

1. Start with VFI, VI and VFD

Before comparing efficiency or battery runtime, identify the classification.

IEC 62040-3 classifies a UPS by how its output behaves when the input supply moves. The code has three parts — VFI-SS-111 is the one you see most often — and the first part decides what the UPS can protect at all.

VFD — voltage and frequency dependent

Output voltage and frequency both follow the input supply during normal operation. Standby and offline designs sit here: the load runs on utility, and the inverter takes over when the input fails.

Because that transfer has to happen, there is a short break. A few milliseconds is fine for a desktop PC or a POS terminal. It is not fine for a PLC that has to ride through the gap.

VI — voltage independent

The UPS regulates output voltage across its specified operating range, while output frequency still tracks the input.

Line-interactive designs sit here. Automatic voltage regulation absorbs a band of input-voltage variation without touching the battery, so the battery gets called on less often. VI suits network equipment, telecom cabinets and smaller IT rooms where full double conversion is not justified.

VFI — voltage and frequency independent

Output is independent of input voltage and frequency within the stated operating conditions. Online double-conversion designs sit here: incoming AC is rectified to DC, and the inverter rebuilds controlled AC and feeds the load continuously.

Data centres, medical imaging, industrial control and process equipment normally specify VFI.

The rest of the code

The letters and digits after the dash are where most datasheets stop short:

• SS — output waveform. S means sinusoidal, in normal operation and in stored-energy operation; the standard's threshold is total harmonic distortion below 8% under linear and reference non-linear load. X means sinusoidal under linear load only. Y means not sinusoidal at all.

• 111 — dynamic output performance, three digits: behaviour on change of operating mode, on a linear load step, and on a non-linear load step. 1 is the most stringent class, 3 the least.

Ask for the full code. "VFI" tells you the output is independent of the input. It does not tell you how clean that output is, or well the UPS holds voltage when the load steps.

Do not open with the question of which UPS has the better efficiency or the larger kVA number. Make sure the class matches the load first.

2. kVA, kW and output power factor

A 100 kVA UPS does not automatically deliver 100 kW. Check this before you compare prices.

kVA — apparent power

Single-phase:

kVA = V × A ÷ 1000

Three-phase:

kVA = √3 × V × A ÷ 1000

kVA describes the voltage-and-current capability of the UPS. On its own it does not give you the active power the load can draw.

kW — active power

kW is what the load actually uses. Two datasheets, both 100 kVA:

• 100 kVA / 90 kW

• 100 kVA / 100 kW

Same kVA. Ten kilowatts of difference in what you can connect.

Output power factor

For the UPS output rating:

kW = kVA × output power factor

A 100 kVA unit at PF 0.9 is a 90 kW machine. Read output PF as a ceiling on what the UPS can supply, not as a number your load has to match — a load running at PF 0.95 does not need a PF 0.95 UPS, it needs a UPS whose kW ceiling clears its kW demand.

That ceiling is also no guide to how the UPS handles motors, transformers or other high-inrush loads. Those depend on overload capability, transient response and the starting characteristic of the load itself.

The kVA number is identical in all three rows. What you can connect to it is not.

Put kVA and kW next to each other in every quotation, and if a supplier quotes kVA only, ask for the rated kW before you size anything.

3. Harmonics: THDi upstream, THDv downstream

Two different distortions, in two different directions, quoted under two different conditions.

Input PF and THDi — what the UPS draws

Input power factor is the ratio of real input power to the apparent power the UPS draws. Input current THD describes how far that current waveform departs from a sine wave. Both load the upstream infrastructure: cables, switchgear, transformers and the generator.

Rectifier topology drives both numbers. Older thyristor-based rectifiers draw more reactive and harmonic current, 12-pulse designs reduce the characteristic input harmonics, and modern IGBT rectifiers with active PFC reach input PF close to unity with much lower THDi.

The actual figures depend on source impedance, input filtering, load level and the rectifier design itself. A PF quoted at 100% load is not the PF at 20% load, and a THDi figure with no load condition cannot be ranked against one that has it.

THDv — what reaches the load

THDv is distortion of the output voltage the load sees. Manufacturers quote it separately for linear and non-linear loads, and that distinction is the whole point: a switched-mode power supply draws a non-sinusoidal current, and that current distorts the voltage of whatever feeds it.

So THDv <2% on one datasheet and <5% on another do not rank the two units until you know which load each was measured on. It is also why the IEC 62040-3 waveform letter matters — the difference between S and X is exactly whether the output stays sinusoidal under the reference non-linear load.

The number matters. The condition behind the number matters just as much.

4. Efficiency: check the mode and the load

Efficiency: 99% looks excellent. The first question is: 99% in which mode?

Manufacturers publish separate figures for double-conversion mode and ECO mode, usually at 25%, 50%, 75% and 100% load.

ECO mode is not a more efficient version of the same protection. In ECO mode the load runs on the filtered utility supply and the inverter idles until the input goes out of tolerance — so that 99% describes a UPS which, for that period, is not conditioning the power. Comparing it against another unit's online efficiency compares a bypass path with a double-conversion path.

Load matters as much as mode. A UPS bought for 100 kW of future capacity may spend its first three years at 30–50% load, and efficiency at that operating point — not the full-load peak — is what shows up on the electricity bill.

Ask for the efficiency curve, or at minimum the double-conversion efficiency at the load you will actually run.

5. Overload: the percentage means nothing without the duration

"150% overload capability" is half a specification.

You need two numbers: overload percentage, plus how long the inverter holds it.

Datasheets normally give a ladder:

• 105% — continuous, or a long limit

• 125% — typically minutes

• 150% — typically seconds to a minute

• above 150% — transfer to bypass

When the time expires, the UPS transfers the load to bypass. That is fine if a bypass supply exists and sits within tolerance. It is not fine on a site that was relying on the inverter to carry the peak.

For loads with high starting or transient demand — motors, transformers, compressors, medical imaging — the useful question is not "can it handle 150%?" but:

How many seconds does the inverter hold 150%, and what happens in the second after that?

6. Transfer time: which two states is "0 ms" between?

Online double-conversion units are specified at 0 ms from normal operation to battery operation, because the inverter is already supplying the load. Lose the utility input and the DC source changes; the inverter does not have to start up and take over.

"0 ms" does not mean every transition inside the UPS is instantaneous. Transfers involving static bypass, ECO mode and fault conditions are separate events with their own figures.

On VI and VFD designs, the move from utility to battery or inverter involves a real interruption, typically a few milliseconds. Most IT power supplies ride through it; some industrial loads do not.

When a datasheet says "transfer time", check which two operating states it is comparing.

7. Steady-state regulation and transient response are different tests

They sit next to each other on the page and measure different things.

Output voltage regulation: ±1% describes how tightly the UPS holds voltage under steady conditions.

Transient response describes what happens when the load changes fast. Step from 20% to 100% load and the output voltage moves outside its steady-state band before it recovers.

For loads that step — lifts, welders, imaging equipment, compressor starts — ask for three numbers:

• maximum transient voltage deviation

• recovery time

• the load step the figures were measured at

A datasheet that only gives ±1% steady-state regulation has not told you how the UPS behaves during the event you actually care about.

8. Battery voltage, Ah and runtime are three different numbers

DC voltage tells you the configuration

A nominal 384 VDC bus usually means 32 × 12 V blocks in series:

32 × 12 V = 384 V

That is the string layout. It says nothing about backup time.

Ah is rated at a discharge rate, not in the abstract

A 12 V 100 Ah block carries that rating at one discharge rate, one temperature and one end-of-discharge voltage. Change any of the three and the capacity you can actually draw changes with it.

That is why nameplate Ah cannot be turned into a 10- or 30-minute runtime by arithmetic. UPS applications discharge far faster than the rate the rating was set at, and a fast discharge meets the cut-off voltage before the nameplate capacity is used up.

For sizing, use the battery maker's constant-current or constant-power discharge table, read at the runtime and end-of-discharge voltage the project needs. Battery ageing, temperature and design margin are then applied on top, according to the project requirements and the applicable battery-sizing practice. Runtime calculated from nameplate Ah is a guess with a decimal point after it.

Runtime needs the load

Runtime depends on the actual load in kW, the number of blocks, the Ah and the rate it is rated at, the DC bus configuration, discharge characteristics, inverter efficiency, battery temperature and condition, and the end-of-discharge voltage.

"32 × 100 Ah" does not tell you whether that is ten minutes or an hour. The load does.

Ask for runtime at your load in kW, and ask which discharge table and cut-off voltage it was calculated from, rather than asking for a bigger Ah number.

9. Static bypass and maintenance bypass do different jobs

"Bypass included" is worth one more question.

Static bypass is an automatic alternative path. Overload, inverter fault and overtemperature each end in one of two outcomes: a transfer to static bypass, or a shutdown. The static bypass is what keeps the load alive through the first one.

Maintenance bypass is a manual isolation path that lets a technician de-energise the UPS power electronics for service while the load stays supplied. It is a maintenance function, not a second name for static bypass.

On critical sites the two are specified separately, because maintenance planning is part of the power architecture rather than an accessory. If a quotation says "bypass included", ask which one — or whether it means both.

10. MTBF: a reliability statistic, not a service life

MTBF >100,000 hours does not mean the UPS runs for more than eleven years without failing.

MTBF is a statistical reliability metric calculated under a defined method and assumed conditions. Before it means anything, you need five answers:

• Which standard or calculation method was used?

• Does it cover the complete UPS, or only the power module?

• Are batteries included?

• What ambient temperature and load were assumed?

• Is the matching MTTR published, so availability can be worked out?

Two manufacturers using different methods produce two numbers that cannot be ranked. Fans, capacitors and batteries have their own replacement intervals whatever the MTBF says.

11. Input voltage range: the full-load number is the one that counts

A wide input range keeps the UPS on utility through voltage excursions instead of burning battery cycles.

Input voltage range: 110–300 V still needs one qualifier: at what load?

Many units accept a wider range at reduced load, and the range can also shift with output rating, input frequency, battery charging state and bypass limits.

So ask:

What input-voltage range does the UPS accept at 100% rated load, without derating and without transferring to battery?

At sites where utility voltage sits high or low for hours at a time, that single answer decides how often the battery gets used.

12. Derating, environment and the numbers in the fine print

Five figures rarely make the front page and always turn up in the installation manual. Read them before you commit to a layout.

• Temperature derating. The kVA rating holds up to a stated ambient, commonly 40 °C. Above it, capacity follows a published curve downwards.

• Altitude derating. At height, reduced air density affects cooling and clearances. Makers set their own altitude limit and publish a derating curve above it — use that curve rather than a rule of thumb, because the percentage is not the same across designs.

• IP rating and audible noise. IP20 is fine in a clean electrical room and wrong in a wash-down area or an open production hall. Noise matters once the cabinet stands near people.

• Recharge time. After a full discharge, how many hours back to 90% capacity? That figure sets how soon a second outage can be survived.

• Frame capacity versus installed capacity. On modular units, a 300 kW frame with three 50 kW modules fitted is a 150 kW UPS. Quotations that show the frame number without the installed number are not comparable.

Ask for these as a list. A supplier that publishes the derating curve, the recharge time and the installed-module count is straightforward to compare against another one.

If those figures are missing from the datasheet, ask for them before you compare quotations. The derating curve and the recharge time are exactly what a headline kVA number hides.

13. UPS datasheet comparison checklist

Put the specifications from three quotations into one table instead of comparing brochures side by side.

14. What to send a supplier before you ask for a price

Do not open with a kVA number. Open with the load.

1. Load type — servers, CT scanner, motor, process line, telecom rack

1. Actual and peak load — in kW, and in A where you have it

1. Input and output voltage — including phase configuration

1. Required runtime — 10, 30 or 60 minutes, at what load

1. Starting or peak behaviour — inrush current for motors, transformers and imaging equipment

1. Redundancy architecture — single unit, parallel capacity, N+1 or 2N

1. Site conditions — utility-voltage range, generator supply, ambient temperature, altitude, IP requirement

From that, capacity, battery configuration and architecture get calculated against the application instead of picked from a catalogue page.

Don't compare the biggest number — compare the condition behind it

Four checks settle most comparisons: the classification code, the kW behind the kVA, the double-conversion efficiency at your real operating load, and the overload duration in seconds. Until those four are equal across two quotations, every other line on the datasheet is noise.

Comparing UPS systems for a hospital, a server room or an industrial line? A supplier's applications team can only size from what you send them, and load data beats a kVA number every time. TAFENG engineers work from the load list, the peak behaviour and the runtime requirement, then document which conditions each published figure applies to.

FAQ

What does VFI-SS-111 mean on a UPS datasheet? It is the IEC 62040-3 classification code. VFI means the output is independent of input voltage and frequency. SS means the output waveform is sinusoidal in both normal and stored-energy operation, with total harmonic distortion below 8% under linear and reference non-linear load. The three digits rate dynamic performance on change of operating mode, on a linear load step and on a non-linear load step, where 1 is the most stringent class. A VFI unit is not automatically VFI-SS-111: the waveform letters and the dynamic digits are graded separately, so two VFI datasheets can carry different codes.

Why does a 100 kVA UPS not deliver 100 kW? Because kVA is apparent power and kW is active power, linked by the output power factor: kW = kVA × PF. A 100 kVA unit rated at PF 0.9 is a 90 kW machine. Always compare the kW rating, not only the kVA.

What does "150% overload" mean without a time limit? Very little. Overload capability only means something as a pair — percentage and duration. A UPS rated 150% for 30 seconds behaves very differently from one rated 150% for 10 minutes, and the datasheet should also state what happens when the time expires, usually a transfer to bypass.

Is 99% efficiency in ECO mode the same as 99% online? No. In ECO mode the load is supplied from the filtered utility and the inverter idles, so losses are low but the load is not being conditioned until the input goes out of tolerance. Compare ECO with ECO and online with online.

How do I work out runtime from a battery Ah rating? You cannot, from Ah alone. Ah is specified at one discharge rate, one temperature and one end-of-discharge voltage, and UPS runtimes are far shorter than the rated rate. Use the battery maker's constant-current or constant-power discharge table at the runtime and end voltage you need, then apply ageing, temperature and design margin. Ask the supplier for runtime at your load in kW.

Does a higher MTBF mean the UPS will last longer? Not necessarily. MTBF is a calculated reliability statistic that depends on the method used, the scope of what is covered and the assumed operating conditions. Two manufacturers using different methods produce figures that cannot be ranked, and fans, capacitors and batteries have their own replacement intervals whatever the MTBF says.