Power & UPS
UPS load planning
A UPS is for safe shutdown and short continuity, not for powering every device on a desk during an outage.
Who this is for
Home operators in 2026 deciding what belongs on UPS battery outlets, how much runtime is enough for NAS/router/desktop shutdown, and whether to stay on lead-acid AGM or migrate to LiFePO4 — against a grid backdrop where NERC projects winter peak demand rising 65% by 2035 and Australia's coal exit is structurally tightening evening capacity.
Outcome
A measured load-watt budget separated into critical (battery) and convenient (surge-only) outlets, a 2026-realistic runtime estimate accounting for battery age and Peukert's law, a wired-up NAS auto-shutdown integration (Synology DSM 7.3, TrueNAS Scale 25.10 Goldeye, or Unraid 7.3.3 NUT), and a pure-sine waveform check that rules out the Active-PFC-PSU silent failure mode.
Required inputs
- A Kill-A-Watt or equivalent plug-in wattmeter (or a UPS LCD that reports actual load watts — APC BR1500MS2 and CyberPower CP1500PFCLCD both do).
- Device list grouped by criticality: ONT/modem, router, mesh nodes, switch, NAS, primary desktop, monitors, laser printer, speakers, phone chargers. Note 80 Plus rating and Active-PFC status of each PSU.
- Current UPS make/model/VA/W rating + battery age + last self-test result. (UPS LCDs typically show battery age in months or service hours.)
- Local electricity context: prone to PG&E PSPS, ERCOT cold-snap risk, Australian post-coal evening peaks — sets runtime target (5-10 min for graceful shutdown vs hours for ride-through).
Step-by-step procedure
Separate critical from convenient at the outlet level
Do: Map every device to battery-backed vs surge-only outlets. On battery: ONT/modem, router, primary switch, NAS, UPS-connected desktop for graceful shutdown. Surge-only: laser printer (fuser pulls 400-1000W in sub-second spikes — trips overload on consumer UPSes), monitors (3-5 min battery time is rarely useful), powered speakers, phone chargers, lighting. A common 2026 mistake is putting the laser printer on a battery outlet 'just in case'.
Expected result: Battery-backed load list is short, ideally network + NAS only at ~70-110W combined. Convenience load is on surge-only outlets.
If not: If everything is on battery, the UPS will trip overload during a real event before any useful runtime is delivered.
Measure actual load with a wattmeter, not nameplate ratings
Do: Plug each device through a Kill-A-Watt for 5 minutes and record steady-state watts. Total the battery-backed set. Compare to UPS LCD load reading — they should match within ~5W. Typical 2026 home stack: ONT 8-12W, router (Wi-Fi 7 mesh main node) 18-25W, switch 5-15W, NAS at idle with spun-down drives 15-25W, NAS with drives spinning 35-60W. Combined: 70-120W realistic.
Expected result: Measured load matches UPS LCD reading. Your runtime calculation now uses a real number, not the UPS marketing 'up to X minutes at full load' figure (which is a worst-case 100% load number).
If not: If LCD reading is wildly off from wattmeter total, suspect a phantom load (PoE switch with unannounced devices, NAS background scan) — investigate before sizing.
Calculate realistic runtime with battery-age derate
Do: Use vendor runtime curves: at ~80W on a fresh 900W lead-acid UPS (APC BR1500MS2 class), expect 25-35 min year 0. Apply derate: year 2 ~20% loss, year 3 ~30% loss, year 4-5 risk of sudden-death failure (lead-acid). For LiFePO4 (EcoFlow DELTA 2, Bluetti AC180), expect 8-15 hours at the same load and ~5-10% loss after 5 years. Year-3 lead-acid: a 30-min nameplate becomes ~20 min, then often <10 min by year 4.
Expected result: Specific runtime number for the actual load + battery age. Marketing minutes are now scaled to reality.
If not: If you don't know battery age, run the UPS self-test (every model has one) and note the result. APC 'replace battery' chirp = 1 beep/min, repeating every 5 hours.
Wire up NAS auto-shutdown integration
Do: Synology DSM 7.3: Control Panel > Hardware & Power > UPS. Connect USB cable to NAS. Enable 'Network UPS server' if other devices share the same UPS. Set 'Time before NAS enters Safe Mode' to ~3-5 min (must be less than measured runtime at typical load). Leave 'Shut down UPS when system enters Safe Mode' OFF (DSM 7.x only guarantees Omron for kill-power). TrueNAS Scale 25.x: System Settings > Services > UPS. Driver `usbhid-ups` for APC, `blazer_usb` for many CyberPower OEM clones. Set Shutdown Mode=Battery + Shutdown Timer=30s for short-runtime UPSes, or LowBatt to wait for UPS's own low-battery signal. Critical gotcha: vi edits to upsmon.conf revert on reboot — use the UI's 'Auxiliary Parameters' field. Unraid 7.3.3: Settings > UPS Settings (built-in apcupsd) for APC, or install NUT v2 plugin for non-APC. Set Shutdown battery level=10-15%, Shutdown runtime left=3-5 min, Shutdown UPS timer=0 (leave UPS alive so Docker containers can flush). APC BGM1500 has a known scheduled-self-test bug under NUT — disable scheduled self-test or set OFFDELAY correctly.
Expected result: Test by pulling the UPS mains plug while logged into the NAS console. NAS receives the on-battery signal within 10s, schedules shutdown when threshold hit, completes a clean shutdown before battery dies.
If not: If the NAS doesn't see the UPS, suspect the USB HID compatibility list — Synology and TrueNAS list specific tested models. CyberPower OEMs sometimes need `blazer_usb` instead of the default driver. Don't rely on scheduled overnight backups without testing auto-shutdown — an outage during a 2 AM rebuild without clean shutdown can corrupt the pool.
Verify pure-sine output against Active-PFC PSUs
Do: Check the UPS spec sheet for 'pure sine wave' (not 'simulated sine' or 'stepped approximation'). Active-PFC PSUs in modern ATX builds, Synology Plus/XS+ models, and any server-grade PSU see the zero-crossing notch in stepped-sine output as a brownout and drop out the instant grid power fails — the very moment the UPS is meant to protect you. All four 2026-grade picks (APC BR1500MS2, CyberPower CP1500PFCLCD, Eaton 5P 1500, EcoFlow DELTA 2) are pure-sine. Older APC Back-UPS BE-series and budget LiFePO4 portables are not.
Expected result: UPS spec sheet explicitly states 'pure sine wave' (not 'simulated'). On a test, pull mains while monitoring NAS/PC — no PSU dropout.
If not: If your UPS is simulated-sine and your NAS/desktop has Active-PFC, replace the UPS before relying on it. The cheap fix you bought is silently doing nothing.
Recharge + post-outage review
Do: After each outage, record: time on battery, UPS LCD remaining capacity at outage end, recharge time to full (lead-acid: ~16 hr typical for BR1500MS2; LiFePO4: ~2-4 hr typical), any NAS dirty-shutdown flags, any backup-job interruptions. Lead-acid UPSes lose ~20% capacity per year after year 2 — track LCD runtime estimates over time as the leading indicator. IEEE replacement threshold: 80% of original capacity.
Expected result: Trend data over 12-24 months shows when battery replacement is due. LiFePO4 UPSes track 3000-6000 cycles to 80% — usually still healthy at year 5.
If not: Don't wait for the audible 'replace battery' alarm — that's a self-test failure, often years after the battery already failed its job during a real event.
Commands and settings paths
Measure actual load watts
Kill-A-Watt P4400 (or P4480) > plug device, wait 5 min, read W column.
Where: Physical access to each device's power cord.
Expected: Steady-state watts recorded per device; sum matches UPS LCD load reading within ~5W.
Failure means: Without measurement, sizing is guesswork. UPS box ratings are full-load minutes — not your minutes.
Safe next step: Use UPS LCD load reading as substitute if no wattmeter (BR1500MS2 and CP1500PFCLCD both display real-time W).
Synology UPS configuration
DSM Control Panel > Hardware & Power > UPS
Where: Synology DSM 7.3 admin UI.
Expected: UPS detected via USB. 'Time before Safe Mode' set to less than measured runtime at typical load (e.g., 3-5 min for a ~25-min-runtime stack).
Failure means: UPS protects against brownouts but not against running the battery flat mid-rebuild.
Safe next step: Test by pulling UPS mains plug while NAS console is open. Confirm Safe Mode triggers and clean shutdown completes.
TrueNAS Scale UPS service
System Settings > Services > UPS > Configure
Where: TrueNAS Scale 25.x web UI.
Expected: Driver=usbhid-ups (APC) or blazer_usb (CyberPower OEM); Shutdown Mode=Battery; Shutdown Timer=30s; auxiliary upsmon.conf parameters entered via UI not vi.
Failure means: Manual config-file edits revert on reboot. NUT driver mismatch = UPS detected as 'unknown' or wrong runtime reported.
Safe next step: Restart UPS service after config change. Check `journalctl -u nut-server` for driver errors.
Unraid NUT plugin or apcupsd
Settings > UPS Settings (apcupsd) OR Community Applications > NUT plugin
Where: Unraid 7.3.3 web UI.
Expected: Shutdown battery level=10-15%, Shutdown runtime left=3-5 min, Shutdown UPS timer=0. APC BGM1500: disable scheduled self-test (known NUT compatibility bug).
Failure means: UPS timer=non-zero kills UPS power mid-shutdown — Docker containers don't get a clean flush.
Safe next step: Test scheduled self-test before relying on it. If UPS drops power during self-test, disable scheduled self-test entirely.
Pure-sine waveform verification
UPS spec sheet > Output Waveform field (must say 'Pure sine wave' or 'Sine wave').
Where: Manufacturer spec PDF or product detail page.
Expected: Output is pure sine. Simulated/stepped/approximated sine is NOT safe for Active-PFC PSUs (modern NAS, Synology Plus/XS+, gaming desktops).
Failure means: Stepped-sine UPS + Active-PFC PSU = PSU drops out the instant grid fails — the opposite of what the UPS is meant to do.
Safe next step: Pull mains plug during a low-stakes window and confirm NAS/desktop stays up. If they drop, replace the UPS with a pure-sine model.
Evidence to record
- Measured load watts (battery-backed total) + UPS LCD reading — they should match within 5W.
- Per-device breakdown: ONT, router, switch, NAS idle, NAS drives spinning, desktop idle, desktop under load.
- UPS make/model/VA/W/waveform (pure sine vs simulated) + battery age in months + last self-test date.
- Runtime estimate at measured load on current battery age (not nameplate).
- NAS auto-shutdown configured + tested via mains-pull (date of test + outcome).
- Outlet plan: which outlets are battery vs surge-only + why.
Common mistakes
- Putting a laser printer on a battery outlet — fuser pulls 400-1000W in sub-second surges, trips UPS overload, often damages cheaper UPSes. Always use surge-only for laser printers (and microwaves, hair dryers, space heaters).
- Confusing VA with watts — 1500 VA UPS is typically 900-1000 W. Sizing by VA undersizes by 30-40% on real Active-PFC loads.
- Gaming desktop on a 900W UPS — RTX 5080/5090 builds peak 600-700W; leaves zero headroom; trips overload on GPU transient spikes. Either separate UPS for gaming or move it off battery.
- Ignoring battery age until alarm fires — IEEE replacement threshold is 80% original capacity. By the time the 'replace battery' chirp triggers, you've been running degraded for 1-2 years.
- No temperature derating — equipment closet at 35°C cuts VRLA lead-acid life ~50% and derates rated VA ~10-20%. LiFePO4 handles heat better but still wants airflow.
- Trusting marketing runtime literally — the 'up to X minutes' figure is at 100% rated load. Your real load is typically 10-20% of rated, so actual runtime is 4-6x marketing number on a new battery — but collapses to marketing-number territory by year 3 on lead-acid.
- Stepped-sine UPS with Active-PFC PSU — works in mains passthrough; fails the moment grid drops. The single most common silent failure.
- No NAS auto-shutdown wired up — UPS gives you 30 min of runtime but the outage lasts 2 hours. NAS runs flat, dirty shutdown, pool needs scrub. Auto-shutdown is the entire point.
- UPS on carpet against a wall — vent blocked, thermal aging accelerates. Both lead-acid and LiFePO4 need airflow on at least two sides.
- No headroom for growth — sizing for current load only; add 25-40% for future drives, PoE switches, or a second NAS.
Stop points
- Stop immediately for heat, odor, swelling, leaking electrolyte, sparking, smoke, or an overload alarm that doesn't clear after removing load.
- Stop before working on house outlets, breakers, or wiring yourself — call a licensed electrician.
- Stop before opening a UPS chassis to replace battery on a model that isn't user-serviceable (most consumer UPSes are; enterprise online double-conversion often aren't).
- Stop before relying on overnight unattended backups before testing the auto-shutdown integration with a real mains-pull.
Last reviewed
2026-05-06
Source-backed checks
HomeTechOps turns official docs and conservative safety rules into a shorter runbook. These links are the source trail for the page direction.
Did this solve your problem?
Your answer helps us improve this page.
No account or personal details. We store the page, answer, optional reason and day with a 90-day retention period. A random reference in this browser prevents repeat votes. Feedback privacy
Planning a purchase?
We keep a source-backed, price-free comparison so you can buy once and right. No star ratings, every spec cited.
UPS for a NAS: APC vs CyberPower vs EcoFlow →Get the deal & firmware alerts
Home Stack Field Notes: NAS deals, firmware changes worth acting on, restore-test reminders, and new decision guides — plus the capacity & backup sizing cheatsheets from our calculators. Unsubscribe anytime.
Subscribe on beehiiv