Uninterruptible Power Supply (UPS)

An Uninterruptible Power Supply (UPS) is a device that provides instantaneous backup power when the incoming mains fails, sags, or spikes. Unlike a standby generator that needs seconds to start, a UPS draws on stored energy — traditionally sealed lead-acid batteries, increasingly lithium packs or a full battery energy storage system — to keep critical equipment running with zero perceptible interruption. In a sovereign infrastructure stack, the UPS is the difference between a power blip and a corrupted node database.

The three topologies

UPS designs fall into three families. Standby (offline) units pass mains straight through and switch to battery only when power fails, with a transfer gap of a few milliseconds — fine for most PCs. Line-interactive units add an autotransformer that corrects sags and swells without touching the battery, extending battery life in areas with dirty power. Double-conversion (online) units rectify incoming AC to DC, keep the battery charged from that DC bus, and continuously re-invert to clean AC for the load. Because the load is always fed from the inverter output, a grid disturbance never reaches the equipment and there is no transfer time at all. When utility power drops, the UPS rides through on battery for seconds to minutes — long enough for an automatic transfer switch to bring a generator online or for the grid to recover.

What a UPS is actually for in mining

ASIC miners themselves are rarely worth UPS protection — a multi-kilowatt hashing load would flatten any reasonably priced battery in moments, and miners simply restart and rejoin the pool after an outage with nothing lost but a few minutes of work. The control plane is a different story. Network switches, routers, pool-routing gear, monitoring servers, and any sovereign AI inference hardware benefit from clean, uninterrupted power: a farm whose miners survive a blip but whose switch reboots for two minutes still loses the hashrate. Putting the modest-wattage network and management layer on a small UPS is one of the cheapest uptime upgrades a mining operation can make.

Protecting the sovereignty stack

For self-hosted Bitcoin and sovereignty infrastructure — a full node, a Lightning routing server, a Nostr relay, a home server — a modest UPS prevents the data corruption and downtime that an abrupt power loss can cause. Databases mid-write, filesystems mid-journal, and Lightning channel states are all more fragile than the hardware they run on; a UPS buys the seconds needed for a clean shutdown, which most units can trigger automatically over USB. A Lightning node in particular should never lose power mid-operation if you can help it.

Sizing in two dimensions

Size a UPS on both power and energy: the VA/watt rating must exceed the load's draw (with headroom, since ratings assume ideal power factor), and the battery capacity sets runtime at that load. A node-plus-networking load of under 100 W runs a long time on even a small unit; the same unit asked to carry a 1,500 W miner will not last a minute. Decide what genuinely must not lose power, protect exactly that, and let the ASICs — and their PSUs — ride through outages the way they were designed to: by restarting.

A UPS is also a maintenance item, not an appliance. Sealed lead-acid batteries fade in three to five years — faster when kept warm — and a UPS with a dead battery is just a surge strip with delusions, so schedule an annual runtime test under real load and replace batteries on evidence, not on failure. Two bench rules save grief: never plug a miner, laser printer, or other high-surge load into the battery-backed outlets of a small unit, and confirm the automatic-shutdown USB link to your node actually triggers a clean halt before you need it in anger.

D-Central treats UPS sizing as part of designing resilient, self-hosted power.