UPS Capacity Math: Real Runtime Across 5 Models from 850VA to 1500VA

What actually makes an outage painful is not that the lights go out. It is that your NAS is still flushing to disk, the database is mid-transaction, and all you can do is watch it get hard-cut before spending a weekend repairing data. The most common buying mistake here is to simply match the "1500VA" on the box. What you end up with is a machine that cannot actually deliver 1500 watts, and — worse — one whose runtime has almost nothing to do with that number.

I have bought and installed UPS units for my own gear and for a few NAS setups around me, and the recurring mistake is treating VA as runtime. On CyberPower's official spec pages, two machines both labeled "1500 VA" ship 1500VA / 900W at 3 minutes of full-load runtime and 1500VA / 1000W at 2.5 minutes. A smaller 850VA unit from the same maker also manages 3 minutes at full load. In other words, a bigger VA number does not mean longer runtime.

All five UPS units discussed here are documented from vendor spec pages and the official user manual, retrieved 2026-10-19. Prices are vendor MSRP plus the listing state I observed at time of writing, and both change; treat the checkout page as authoritative. This post contains affiliate links: the Amazon links carry tag=techpassive-20, and I earn a commission if you buy through them at no extra cost to you. That does not change a single flaw I disclose below — if anything, because there is commission involved, the flaws matter more.

TL;DR：Four points. First, runtime is governed by battery watt-hours multiplied by conversion efficiency and divided by your actual load; the VA rating only sets the ceiling on how large a load it can carry, not how long it lasts. Second, when reading a spec sheet, look at the battery configuration first (how many cells, 12V/9Ah or something else), because that sets your watt-hour ceiling. Third, across four models I could verify from official sources, the real conversion efficiency at full load lands in a narrow 19% to 24% band, so the manufacturer runtime figures are trustworthy — do not "recalculate" them from raw battery capacity. Fourth, units like the CP1500AVRLCD3 do not support external battery packs: the words expansion and external battery appear zero times in the entire manual, so do not plan on adding runtime later.

Three Numbers That Are Not Interchangeable

The VA, the watt, and the battery watt-hour get mixed up constantly. They have different units and different jobs, and treating them as one number produces a completely wrong answer.

VA (volt-amperes) is the output-side capacity ceiling — the largest apparent power the unit can safely pass through. Think of it as what a wall circuit can handle. The CyberPower CP1500AVRLCD3 is rated 1500 VA, meaning its inverter output path is built to that apparent-power limit.

W (watts) is the real continuous power it can deliver, and it equals VA multiplied by the power factor. These two 1500VA units have different power factors, which is why their watt ratings differ. The CP1500AVRLCD3 official spec table gives an Output Power Factor of 0.6, so 1500 × 0.6 = 900 W. The sine-wave CP1500PFCLCD is listed at 0.7, so 1500 × 0.7 = 1000 W. That is the entire explanation for "same 1500VA, one is 900W and one is 1000W" — it is the power factor, not marketing spin.

Battery watt-hours (Wh) is the number that actually decides runtime. A cell marked "12V / 9 Ah" stores 12 × 9 = 108 Wh. Two cells in series, the usual UPS arrangement, store 216 Wh. This is the real ceiling on available energy.

Once those are separated, the logic is straightforward. Conservation of energy requires battery Wh = output watts × time × conversion efficiency. Battery watt-hours is fixed, and efficiency sits in the neighborhood of 20% (I verify this below against official data), so lower output power buys more time and doubling the power roughly halves the time. That matches the intuition that a bigger battery-backed unit can still have a short runtime.

The key point: VA does not enter this calculation at all. The 1500VA figure only matters when the power draw of what you plug in approaches that limit.

Reverse-Engineering Conversion Efficiency From Official Data

That "roughly 20%" is not my guess. It falls out of the manufacturers' own runtime figures, and it can be cross-checked against each other across four models.

The method is simple. Multiply the stated full-load runtime by the rated wattage to get the energy actually delivered during that interval. Divide that by the battery watt-hours, and you have the fraction of stored energy that became usable output.

Model	Battery config	Battery Wh	Rated watts	Full-load runtime	Back-solved efficiency
CP1500AVRLCD3	2 × 12V/9Ah	216 Wh	900 W	3 min	~20.8%
CP1500PFCLCD	2 × 12V/9Ah	216 Wh	1000 W	2.5 min	~19.3%
CP850AVRLCD	1 × 12V/9Ah	108 Wh	510 W	3 min	~23.6%
BRG1000AVRLCD	1 × 12V/9Ah	108 Wh	600 W	1 min	~9.3%

The first three give 20.8%, 19.3%, and 23.6% — all inside a narrow 19% to 24% band. These three differ in battery count (one cell vs two), power factor (0.6 vs 0.7), and waveform (simulated vs pure sine), yet land on nearly the same coefficient. That convergence is what makes the band credible rather than coincidental.

The fourth unit is an outlier, and the outlier is itself a buying signal.The BRG1000AVRLCD official spec table states "Runtime at Full Load: 1 min." Running 600 W for one minute delivers just 10 Wh, which back-solves to 9.3% efficiency — less than half the other three. I cross-checked that figure against CyberPower's own spec table plus B&H Photo Video, Newegg Business, and CDW; all four state 1 minute. So it is not a vendor typo and not an arithmetic error on my part. The machine genuinely cannot reach a full minute at full load.

What that means for you: if your equipment draws close to 600 W, this unit's usable time is shorter than the 850VA model's (510 W at 3 minutes). Its 1000VA is nominal. A 0.6 power factor means it can only output 600 W, and it is doing that with essentially no headroom.

Why Your Runtime Comes In Below the Published Number

The math above uses full load. In practice you rarely run at full load, so the half-load column matters too. The half-load figures back-solve just as cleanly: the CP1500AVRLCD3 at half load (450 W) for 12 minutes gives 41.7%, and the CP850AVRLCD at half load (255 W) for 11 minutes gives 43.3%. Efficiency is higher at half load, which matches lead-acid discharge behavior — higher current draw means more internal losses and less extractable energy.

Three real-world factors shrink your usable time further:

First, published runtime is measured with new, fully charged batteries at normal temperature.CyberPower states this directly next to the runtime chart: runtimes are based on testing fully charged, new batteries at normal operating conditions, the curve is approximate, and results vary with battery age, level of charge at test, environment, and other variables. That sentence is the vendor's, not a disclaimer I added.

Second, lead-acid batteries age.The CP1500AVRLCD3 manual lists Typical Battery Life as "3 to 6 years, depending on number of discharge/recharge cycles." Note the range rather than a single number — on a machine several years old, internal resistance has risen noticeably and full-load runtime will be shorter than a new unit's.

Third, lead-acid has a defined cutoff voltage.The manual's low-battery warning sounds every half second to tell you the battery is nearly spent and you should save files and shut down, or let software do it. So your truly usable time is whatever you calculated minus the shutdown sequence itself. If your NAS takes 30 seconds to shut down cleanly, those 30 seconds come out of your budget.

Worth noting: in the section titled DETERMINE THE POWER REQUIREMENTS OF YOUR EQUIPMENT, CyberPower's manual recommends keeping the load on the battery outlets below 80% of the unit's capacity. The advice is sound — headroom keeps the inverter in a gentler operating region and reduces how often AVR has to intervene.

Official Specs, Five Models Side by Side

Every value below comes from the manufacturers' own spec pages, retrieved 2026-10-19:

Model	Topology	VA	Watts	PF	Waveform	Battery	Wh	Full load	Half load	Transfer	Warranty
CP1500AVRLCD3	Line Interactive	1500	900	0.6	Simulated sine	2×12V/9Ah	216	3 min	12 min	4 ms	3 yr
CP1500PFCLCD	Line Interactive	1500	1000	0.7	Sine wave	2×12V/9Ah	216	2.5 min	10 min	4 ms	3 yr
CP850AVRLCD	Line Interactive	850	510	0.6	Simulated sine	1×12V/9Ah	108	3 min	11 min	4 ms	3 yr
BRG1000AVRLCD	Line Interactive	1000	600	0.6	Simulated sine	1×12V/9Ah	108	1 min	9 min	4 ms	5 yr
BR1500MS2 (APC)	Line Interactive	1500	900	—	Stepped sine	—	—	—	—	—	—

A few numbers deserve their own note.

Transfer time is 4 ms across the CyberPower units.All three spec tables list Typical Transfer Time as 4 ms. This is how quickly the unit switches over. For virtually any PC or NAS, utility and battery output share the same 50/60Hz frequency, so the switch is effectively imperceptible — I would not spend much attention on this spec.

Warranty length and runtime are unrelated.The BRG1000AVRLCD is the only five-year unit (CyberPower notes units purchased before April 1st, 2016 get three years, and units purchased on or after that date get five). The other three CyberPower units are three years. Yet the BRG is the shortest at full load, at one minute. A longer warranty does not mean longer runtime; they are independent dimensions.

The APC BR1500MS2 official figures are blank in this table, and that is deliberate.APC's US product pages returned 403 anti-bot responses to my requests, so the official wording I could obtain comes from APC's other regional storefronts for the same product: 1500VA / 900W, stepped approximation of a sine wave, six battery-backed outlets out of ten with four surge-only, an input range of 88–147V, a typical recharge time of 16 hours, expected battery life of 3 to 5 years, 1080 J surge suppression, and 8 ms typical / 10 ms maximum transfer time. I did not obtain the full US spec sheet, so I am not quoting runtime minutes for this unit. A gap stays a gap; I will not fill it with reseller numbers. If you need the complete figures, check APC's official page or the manual in the box.

One more comparison point: APC documents a 16-hour recharge against 8 hours for all three CyberPower units. If your supply voltage swings often enough to keep AVR engaged, double the recharge time means slower battery recovery.

Model by Model: What Each One Does Well and Where It Falls Short

CP1500AVRLCD3 (CyberPower, 1500VA/900W)

The product page is CyberPower CP1500AVRLCD3 Intelligent LCD UPS.

The strengths are concrete. Six of its twelve outlets are battery-backed, it offers both USB and serial communication, it includes data-line protection for RJ45 and coax, official MSRP is $219.95, the three-year warranty covers the battery, and the connected equipment guarantee is $500,000. The LCD shows estimated runtime, battery charge percentage, and load percentage, making it the most informative panel among the five here.

Three real flaws. First and most important: it does not support external battery expansion. I extracted the full official manual and searched it — expansion appears zero times, external appears zero times. Whatever runtime you buy is the runtime you keep, and extending it means replacing the unit. Second, full load lasts only three minutes. That figure is trustworthy, since the other three units' efficiency coefficients corroborate each other, but if you read "1500VA" and expect a long ride, your expectations will be badly calibrated — the 1500VA describes a capacity ceiling, not endurance. Third, simulated sine wave output. CyberPower describes simulated sine as economical battery backup for equipment that does not require sine wave output, which is a way of saying it is unfriendly to active PFC power supplies. Modern PC power supplies and some network gear use active PFC, and powering those from a simulated sine output can produce odd fan speeds, faint buzzing, or in some cases repeated restarts.

CP1500PFCLCD (CyberPower, 1500VA/1000W sine wave)

The product page is CyberPower CP1500PFCLCD PFC Sinewave UPS.

If any of your equipment uses an active PFC supply, this is the one to look at. CyberPower positions it explicitly for equipment requiring active PFC power source compatibility. Its power factor is 0.7, so 1500 VA yields 1000 W — 100 W more than the same-capacity simulated sine unit. It supports USB, relay, and SNMP, ships with PowerPanel Business desktop software, and can take an optional RMCARD205 card for remote management. Official MSRP is $274.95, connected equipment guarantee $500,000, twelve outlets.

The real flaw: it is the most expensive of the five on official MSRP, and it does not have the best runtime. At 2.5 minutes full load and 10 minutes at half load, both figures trail the CP1500AVRLCD3. The reason is not mysterious: a higher power factor means the same VA delivers more watts, so the battery discharges at a heavier working point and drops faster. If sine wave output is what you need, the premium is justified; if you only care about runtime, this is not the optimal pick. It also lacks external battery expansion — neither the manual nor the spec table lists an expansion port.

CP850AVRLCD (CyberPower, 850VA/510W)

The product page is CyberPower CP850AVRLCD Intelligent LCD UPS.

I consider this the best value of the five, because its runtime has almost no relationship to its VA number. At only 850VA / 510W with a single 108 Wh cell, it holds 3 minutes at full load, matching the 1500VA unit, and 11 minutes at half load, close to that unit's 12 minutes. Its back-solved efficiency of 23.6% is the highest of the four. Official MSRP is $169.95, the cheapest here. Nine outlets split five battery-backed and four surge-only, USB plus serial, data-line protection, 1080 J surge suppression.

The real flaw: 510 W is a hard ceiling on total output. Exceed it and the unit goes into overload protection and shuts itself down. It suits a NAS plus router plus monitor, typically 200 to 300 W, and it will not drive a tower workstation. Surge suppression of 1080 J is also below the CP1500AVRLCD3's 1500 J, and the connected equipment guarantee is only $250,000 against $500,000 on the 1500VA units. It is also simulated sine, so the same active PFC caveat applies.

BRG1000AVRLCD (CyberPower, 1000VA/600W)

The product page is CyberPower BRG1000AVRLCD Intelligent LCD UPS.

It is the only unit of the five with a five-year warranty, and it carries two USB charging ports at a shared 2.1 A, so phones can charge during an outage — an uncommon feature in this class. Ten outlets, split five and five, with a single 12V/9Ah cell.

The most glaring flaw is the one-minute full-load runtime.It is the only CyberPower unit whose back-solved efficiency falls outside the normal band, at 9.3% against 19% to 24% for the other three, and I confirmed that figure across four sources. Nine minutes at half load is unremarkable, but one minute at full load means essentially no peak buffer — the power spike during equipment startup could exhaust the battery on its own. The 1000VA / 600W spec is not honest in a full-load scenario, so read this one carefully before buying. Simulated sine, and no external battery expansion.

APC BR1500MS2 (APC, 1500VA/900W stepped sine)

The product page is APC BR1500MS2 Pure Sine Wave UPS.

This is the second waveform route among the five; APC's regional pages describe it as a stepped approximation of a sine wave. It is also rated 1500VA / 900W with six battery-backed outlets out of ten. APC's connected equipment guarantee is $250,000 lifetime, lower than CyberPower's $500,000 at the same capacity.

The real flaw, plus this post's sourcing limit: APC's US product pages returned 403 anti-bot responses to my requests, so the official parameters I used come from APC's other regional storefronts cross-checked against several reseller spec sheets. I could not obtain an official runtime figure, so I am not giving a number here. If you need it, check APC's official page or the manual. Separately, its 16-hour typical recharge time is double the CyberPower units', and its 88–147V input range is on the narrow side — if your mains voltage sits well away from 120V, AVR has a tighter band to work with.

Troubleshooting: Three Things That Will Actually Happen

Your calculated runtime is twice what you observe

Check whether you overestimated the load first. The LCD on these units shows a load percentage directly, so read that instead of estimating. A common error is treating the maximum power on a PSU nameplate as its consumption — nameplates list peak figures, and real sustained draw is often only a third to a half of that.

The other easy-to-miss cause: you did the math against new-battery runtime. The manual states a three-to-six-year battery life, so a two-year-old unit will naturally come up short.

The battery meter hits 20% and the unit shuts down immediately

CyberPower's manual explains that the battery capacity bar displays in 20% increments, and the low-battery alarm sounds every half second when the battery is nearly spent. If your equipment draws a lot, far less energy remains at the low-battery warning than you planned for, and the unit cuts out shortly after the alarm.

Two ways out: reduce the load, moving monitors and desk lamps to ordinary outlets, or move to a unit with more battery watt-hours. Note that means more watt-hours, not more VA. In the four-model table at the top, the 850VA unit carries a 108 Wh cell while the two-cell units carry 216 Wh, and that gap, not the VA number, is what separates a two-minute ride from a twelve-minute one.

Fans spin erratically or the device reboots itself when connected

This is almost always simulated sine meeting an active PFC supply. CyberPower positions its sine wave models as being for equipment requiring active PFC power source compatibility; read that backwards.

The fix is a sine wave model such as the CP1500PFCLCD or the APC BR1500MS2. Before spending money, verify: plug the suspect device into a UPS you already know works with it, and confirm that is the cause first.

A Decision Tree

Working backward from your equipment list is far more accurate than picking a VA number:

Start by computing your real draw.Read the nameplate of each device you plan to connect and take sustained consumption, not peak. A NAS, a router, and a 24-inch monitor usually lands between 150 and 300 W.

If total draw is under 200 W and any device uses an active PFC supply, take the sine wave route: the CP1500PFCLCD or the APC BR1500MS2. The trade in this tier is runtime for compatibility — the sine wave units actually show shorter runtime figures than the same-capacity simulated sine models.

If total draw is 200 to 350 W and the devices are all conventional, the CP1500AVRLCD3 is the best overall pick here: two cells give 216 Wh and the longest published runtime in this group at 12 minutes half load, with the most complete interfaces and a $500,000 guarantee.

If you want the cheapest unit and your draw stays under 300 W, the CP850AVRLCD is the value pick. Its 510 W ceiling is ample for that combination, $169.95 is the lowest MSRP here, and its runtime still approaches the 1500VA unit.

Treat the BRG1000AVRLCD with care.The five-year warranty and USB charging are genuine advantages, but a one-minute full-load runtime is anomalous in this class. If your draw approaches 600 W, such as a workstation with a discrete GPU, it will trip protection on the startup transient.

Whichever unit you pick, confirm one thing first: that you will not need an external battery pack.None of these five lists an expansion battery interface in the official manual or spec sheet. If you might need longer runtime later, none of them is the right choice and you should look at product lines with an expansion battery bay — that is a more important filter than the VA number.

FAQ

Q: Can a 1500VA unit actually handle my equipment?

A: Look at watts, not VA. Compare the manufacturer's watt rating, which equals VA times the power factor, against your total draw and leave 20% headroom. For the CP1500AVRLCD3, 1500 VA corresponds to 900 W of actual output, so it is only safe for loads under 900 W. CyberPower's manual also advises keeping loads below 80% of rated capacity.

Q: Why do two 1500VA units rate 900 W and 1000 W?

A: Different power factors. The official spec tables list Output Power Factor as 0.6 for the CP1500AVRLCD3 and 0.7 for the CP1500PFCLCD, so 1500 × 0.6 = 900 and 1500 × 0.7 = 1000. A higher power factor means more useful power from the same VA, which is one of the advantages of sine wave models like the CP1500PFCLCD.

Q: Does simulated sine versus sine wave matter for ordinary office gear?

A: For the vast majority of setups, no. The difference shows up mainly with active PFC power supplies, which includes most modern PC power supplies plus some NAS and network devices, and symptoms can include odd fan speeds, buzzing, or repeated restarts. If you are unsure which category your gear falls into, check the nameplate for PFC, or simply pick a sine wave model to be safe.

Q: Can I replace the battery myself, and how long does it last?

A: Yes, and the units here are all marked User Replaceable: Yes. The CP1500AVRLCD3 manual gives a typical battery life of three to six years depending on discharge and recharge cycles, while APC documents three to five years. Always disconnect mains power and unplug the unit before replacing; the manual has the full procedure.

Q: Should the UPS stay plugged in all the time?

A: Yes. In its care section, CyberPower's manual explicitly states that for optimal battery charge you should leave the UPS plugged into an AC outlet at all times, and that a unit stored for an extended period should be recharged every three months. Leaving lead-acid in a discharged state accelerates its decline, which matters far more than any standby power savings.

Wrapping Up

The most commonly wasted money in this category comes from deciding on the "1500VA" number alone. Once the official data for these five is laid out together, the pattern is clear: runtime is set by battery watt-hours, VA only governs peak headroom, sine wave costs you compatibility rather than endurance, and none of these units lets you add a battery later, which is a question worth answering during the selection stage rather than after.

The approach that saved me the most time across the setups I have configured is to invert the order: measure the real sustained draw of your devices first, then decide how many battery cells you need, rather than picking a VA figure and hunting for a matching box. Your equipment list is the only reliable input. The number on the packaging is a reference point.

If you are still choosing the NAS itself, start with our NAS drive showdown on CMR versus SMR — a UPS covers the instant of an outage, while drive choice covers reliability outside of that instant, and the two decisions belong together. It is also worth being clear that surge-only power strips and UPS units solve different problems; this smart power strip guide covers when the former is enough. And if you want the capacity-tier version of this topic, the 2026 programmer UPS buying guide covers which tier to buy, while this post covers how to read the numbers on it.

All specification data here comes from vendor spec pages and the official user manual, retrieved 2026-10-19. Prices and availability reflect the state at time of writing; defer to the live listing page. Amazon links in this post are affiliate links, and I may earn a commission if you buy through them, which does not change your price and does not affect any of the flaws disclosed above.

👉 Join Xiaomi MiMo Platform: Leading AI model platform with cost-effective inference

👉 Join Aliyun AI: Top AI products with exclusive coupons for business innovation

📌 This article was AI-assisted generated and human-reviewed | TechPassive — An AI-driven content testing site focused on real tool reviews