I don't think the cheapest APC Back-UPS units can be monitored except in Windows

Original Article ↗ Hacker News Discussion ↗

Cheap APC Back-UPS models are reported to behave poorly under Linux monitoring tools, emitting rapid false power and battery events that Windows software likely masks with debouncing and vendor-specific handling. Commenters broaden this into a critique of low-end UPS gear in general: from twitchy power detection, proprietary USB protocols, and weak Linux support to concerns about fire risk, poor surge protection, and short-lived lead-acid batteries. Many recommend spending more on higher-end or enterprise-grade units (often pure sine-wave, from brands like Eaton or APC’s pro lines), repurposing used UPS hardware with new batteries, or exploring LiFePO₄-based and “portable power station” alternatives.

Monitoring issues with cheap APC Back‑UPS units

  • Several reports that newer, low‑end APC Back‑UPS models spam rapid battery attach/detach and power fail/restore events under Linux.
  • Consensus that APC’s Windows tools likely debounce or filter these “twitchy” signals; open‑source tools (apcupsd, NUT) expose the raw noise.
  • Some users report specific models working fine on Linux/OpenWrt when:
    • Using the vendor’s oddball USB/serial cables.
    • Disabling USB autosuspend via udev rules.
  • Others see similar flapping even on higher‑end APC units, so it’s not strictly “cheap only”.

Linux compatibility and who’s responsible

  • Debate over expectations: some say “don’t buy consumer UPS if you want monitoring”; others argue vendors should follow USB HID power standards.
  • Critique that Linux “compatibility” claims (by vendors and projects) can be stale or misleading; hardware can change internally without SKU change.
  • Split view on blame: one side says users must research Linux support; the other says closed protocols and Windows‑only tools are the real problem.

UPS quality, safety, and reliability

  • Multiple anecdotes of cheap units: frequent relay clicking, false power events, early failures, and some reports of smoke/fire.
  • CyberPower consumer units called out for problematic yellow glue that can carbonize and arc; APC criticized for odd proprietary cabling and perceived post‑acquisition quality decline.
  • Counterpoint: some long‑lived APC/Eaton units still running after decades with only batteries and capacitors replaced.
  • Strong disagreement on fire risk: some say UPS batteries are a major datacenter fire source; others argue fires are statistically very rare versus installed base.

Power quality: sine wave vs stepped output

  • Disagreement on need for pure sine wave:
    • Some see it as overkill for consumer gear.
    • Others report active‑PFC PSUs, some desktops, and some 3D printers failing or behaving oddly on stepped/square‑wave UPSes.
  • For datacenters and “serious” gear, sine‑wave output is widely preferred and used as a proxy for overall build quality.

Batteries: cost, chemistry, and lifespan

  • Replacement SLA batteries often cost nearly as much as a new low‑end UPS; shipping heavy lead adds significantly.
  • Users commonly replace OEM packs with generic SLA of the same size/voltage, reporting good results and lower cost.
  • Complaints that stock batteries, especially in consumer CyberPower/APC units, die in 2–3 years; third‑party replacements sometimes last longer.
  • Lithium‑ion/LiFePO₄ UPSes discussed:
    • Seen as ideal for cycle life and safety (LiFePO₄), but still much more expensive and rare in low‑end gear.
    • DIY retrofits into SLA‑designed UPSes raise concerns about charge profiles, cutoff voltages, and runtime miscalculation.

Alternatives and design strategies

  • Some argue a good surge protector or whole‑house surge device is more important than a cheap UPS; abrupt power loss is often acceptable for home systems.
  • Others oversize UPSes for long runtimes for network gear, pairing with cellular backup when ISP plant batteries die.
  • Alternatives mentioned:
    • “Solar generator” / portable power stations (questionable switchover times).
    • Hybrid inverters and off‑grid gear acting as whole‑house UPS.
    • DC‑first approaches (12 V battery + DC‑DC + picoPSU) to avoid double conversion losses.
    • Network‑management‑card UPSes (SNMP/web) to avoid USB quirks.

Is UPS monitoring even necessary at home?

  • One camp: modern OSes, filesystems, and many databases handle power loss well; monitoring/shutdown adds complexity and attack surface. Let UPS run flat.
  • Other camp: sudden loss can still brick some SSDs or upset stateful services; graceful, timed shutdown on low battery is worth the integration pain, especially for homelabs and storage servers.