This is the one-device band. 7 units in stock, delivering 330-600 W of maker-stated real power - enough for a small NAS or a single desktop and its network gear, and not enough for a rack. 2 state pure sine; 5 do not say.
Prices updated · 7 units in this band · maker-stated watts run 330-600 W · from $94.97
2 in-stock listings · VA, watts, waveform and outlet counts are the maker’s claims, read from the listing · refreshed every few hours
| Unit | VA | Watts | Waveform | Form | Comms | Price | |
|---|---|---|---|---|---|---|---|
APC SMT750C APC Smart-UPS, SMT750C Tower Back-UPS, 750 VA Backup Battery Power Supply, 6 x | 750 VA | 500 W | pure sine | Tower | USB, Serial, SNMP, LCD | $529.60 | Buy Now |
APC SMX750C APC Network UPS, 750VA Smart-UPS Sine Wave UPS with Extended Run Option, SMX75 | 750 VA | 600 W | pure sine | Rackmount | not stated | $499.99 | Buy Now |
5 in-stock listings · unknown is not the same as safe - check the maker's datasheet · VA, watts, waveform and outlet counts are the maker’s claims, read from the listing · refreshed every few hours
| Unit | VA | Watts | Waveform | Form | Comms | Price | |
|---|---|---|---|---|---|---|---|
APC BE550G APC UPS 550VA/330W Backup Battery Power Supply for PC, Electronics, BE550G | 550 VA | 330 W | not stated | not stated | not stated | $101.33 | Buy Now |
APC BE650G1 APC UPS 650VA/390W Battery Backup & Surge Protector for PC, NAS | BE650G1 | 650 VA | 390 W | not stated | not stated | not stated | $94.97 | Buy Now |
APC SMT750 APC Smart-UPS 750 VA Tower UPS - 750VA/500W - 4.6 Minute Full Load - 6 x NEMA | 750 VA | 500 W | not stated | Tower | not stated | $449.00 | Buy Now |
APC APC-750VA APC SMART-UPS 750VA LCD 230V with SMARTCONNECT | 750 VA | 500 W | not stated | Tower | USB, SNMP, LCD | $499.99 | Buy Now |
Eaton Eaton-850VA Eaton Tripp Lite ECO850LCD 850VA UPS 425W Battery Backup Surge Protector | 850 VA | 425 W | not stated | Tower | not stated | $130.10 | Buy Now |
Of the 7 in-stock units we track, 2 state an output waveform - 29%. Of those, 2 say pure sine and 0 say simulated. The remaining 5 say nothing at all.
The asymmetry is the finding. Pure sine is a selling point and gets printed; a stepped approximation is not and does not. So a silent listing is more likely to be the cheaper waveform than the better one, and we still do not claim to know which - we count it as unknown and tell you to check the datasheet.
600 to 900VA is the smallest size worth buying for anything holding data. Below it you are into desk-side units designed to keep a monitor alive for long enough to save a document, which is a real product with a real job but not this one.
Across the units we have in stock in this band, the maker-stated real power runs 330-600 W. That is the number to size against, and the gap between it and the VA rating on the box is the whole reason the arithmetic below matters: a 900VA unit is not a 900W unit and no manufacturer ever claimed it was.
The realistic load for this band is one machine plus the network hardware it depends on. A two-bay or four-bay NAS, the switch it hangs off, and the router. Add a desktop and you have probably left the band. The specific figures are on your devices' own spec sheets - we publish no per-device wattage on this site, because we have not fetched and logged a datasheet for every machine and an approximate figure attributed to a typical device is not a citation.
One thing that catches people at this size in particular: drives spinning up all at once at power-on draw considerably more than they do at rest. A unit sized precisely to the idle figure has nothing left for the restart, and the restart is exactly when it is being asked to work.
The temptation in the cheapest band is to treat waveform as a premium feature. It is the opposite: this is the band where stepped-wave units are most common, because it is the band where the price difference is proportionally largest.
Of the 7 units we track here, 2 state pure sine output in their own listing and 5 say nothing at all. Nothing in our catalogue, at any size, states a simulated wave outright - sellers advertise the good waveform and stay quiet about the cheap one, so silence in this band is worth more suspicion than silence higher up.
If the machine you are protecting has an active PFC power supply - and a NAS almost certainly does - then a unit that does not state its waveform needs looking up on the manufacturer's own datasheet before you spend anything. It takes two minutes and it is the difference between a UPS that transfers cleanly and one that drops the machine at the moment of transfer.
At this price point the battery is a large fraction of the unit cost, which has a consequence people miss: replacing it in three to five years is not obviously cheaper than replacing the whole unit. APC's own guidance puts the expected service life of the sealed lead-acid cell at three to five years, shortened by heat.
Cartridges for common small units are widely reported around $30-50 - reported, not measured by us, because we do not track UPS battery prices. Against a unit that cost well under $200 new, that is a meaningful fraction rather than a trivial top-up, and it is worth checking the price of the specific cartridge your model takes before assuming the cheap-repair path exists.
The failure is silent either way. A unit with a dead battery keeps filtering, keeps passing mains through and keeps looking healthy on its front panel. Pull the plug deliberately once a year and watch what happens - Synology's own compatibility documentation gives the same instruction, in the same words, about simulating an outage to confirm the behaviour.
Sellers lead with VA because it is the bigger number. VA is apparent power; watts is real power, and watts is what decides whether your load fits. The ratio between them is the unit’s power factor and it is not a constant - Synology’s own guidance is that “Different manufacturers have different volt-ampere (VA) ratings based on the performance of their UPS. To determine the total power (W) that the UPS can support, we recommend consulting the manufacturer’s product manual.”
Their published method, which is the one these pages use rather than one of our own:
Synology’s worked example takes a 650 W load and a UPS with a 70% output power factor and arrives at 928.5 W - so a “1000VA” unit rated at 600 W would not have carried it, despite the reassuring four-digit number on the box. Add your own devices’ figures from their spec sheets; we publish no wattage on your behalf.
The UPS hub carries the waveform rule, the sizing method and the battery economics · 1000-1500VA · 1500VA and up · For a NAS · For a rack · tower vs rackmount (4 tower and 1 rackmount units sit in this band).
This page exists because the catalogue holds at least 4 units in the band. Bands below that floor are not published rather than published thin.
For a small NAS and its switch, usually yes - the units in this band state 330-600 W of real power. Size it properly rather than by feel: take your NAS's own power consumption figure from its spec sheet, add the switch, then divide by the UPS's output power factor to get the output rating you need. And leave headroom for spin-up, which draws considerably more than idle.
Some headroom is sensible, mostly for drive spin-up at power-on. Buying two sizes up to get more runtime is usually the wrong purchase for a storage machine, because the job is a clean shutdown in a couple of minutes rather than riding out the outage. Spend the difference on a unit that states pure sine output instead.
Because it is the specification that decides whether the unit works with your hardware. An active power factor correction power supply - most servers, desktops and NAS units - can shut down unexpectedly when fed a stepped approximation of a sine wave, which CyberPower's own documentation names as the failure sine wave output exists to prevent. A listing that does not state its waveform is marked not stated rather than assumed safe.
Specification claims on this page were checked against the pages below on 4 August 2026. Where a widely-repeated figure did not survive that check, we report what the source says now and say so in the text rather than repeating the familiar number.