Voltage Sag

A voltage sag (also called a dip) is a short-duration decrease in RMS voltage on an AC supply. Under IEEE 1159, a sag is a reduction to between 0.1 and 0.9 per unit of nominal voltage lasting from half a cycle up to one minute — deeper and it is classed as an interruption, longer and it becomes a sustained undervoltage. It is the single most common power-quality disturbance on real-world electrical systems, and for a home miner it is the disturbance most likely to knock hardware offline without leaving any obvious fault to find afterward.

What causes a sag

Sags come from two directions. Locally, large loads starting up pull the voltage down across shared wiring: a well pump, an HVAC compressor, or another miner energizing draws heavy inrush current, and the resistance of your branch circuit and service conductors converts that current spike into a momentary voltage dip for everything on the same run. Remotely, faults elsewhere on the utility's distribution network — a short circuit two feeders over, a lightning strike, a tree branch — depress voltage across a wide area for a few cycles until protective devices isolate the fault. You cannot prevent the remote kind; you can absolutely engineer around the local kind.

How mining PSUs respond

An ASIC power supply rides through shallow, brief sags using energy stored in its bulk input capacitors — that stored energy is the PSU's entire holdup budget, typically enough for a cycle or two at full load. Deeper or longer sags push the input below the supply's undervoltage-lockout threshold, at which point the PSU drops the load to protect itself and the miner reboots. APW-class supplies on a proper 240V circuit have more margin than the same hardware limping on 120V: at 240V the same percentage sag leaves more absolute voltage above the lockout threshold, and the halved current means less additional drop across your own wiring. This is one of several reasons a dedicated 240V circuit is the correct way to run a full-size PSU at home. Repeated sag-induced restarts are not free even when the hardware survives: each reboot adds a thermal cycle to the hashboards — the same thermal fatigue mechanism that eventually cracks BGA joints — and discards in-progress work.

Diagnosing and fixing sag problems

A miner that reboots "randomly" but shows clean temperatures and a clean kernel log right up to each restart is a power-quality suspect, not a firmware one. Correlate restart timestamps against household events: does it die when the dryer starts, when the AC kicks in, when a second miner powers up? A logging voltage monitor on the miner's circuit settles the question definitively. Fixes escalate in cost: move the miner to a dedicated circuit, upsize or shorten conductors to cut resistance, stagger startup of multiple machines so their inrush currents do not stack, and only then consider ride-through equipment. Voltage sags are distinct from a voltage swell, the opposite event, and from sustained undervoltage; all belong to the broader discipline of power quality.

Logging sag frequency and depth before adding hashrate tells you whether the site needs conductor upgrades or sequencing first. Electricity is the one input a home miner fully controls — treating the supply side with the same rigor as the firmware side is what separates a stable operation from a mysterious one.

Sag tolerance is also a specification worth reading before buying: PSU datasheets state an input voltage range and a holdup time, and hardware run near the bottom of its input range has already spent its margin before the first disturbance arrives. Keep mains-side wiring short, tight, and generously sized on mining circuits — loose terminations add resistance that deepens every local sag, and they heat under load, which loosens them further.