Bench Power Supply

A bench power supply is an adjustable, regulated DC source used on a workbench to power a board under test. Its defining feature for repair work is independent control of both voltage and current. In constant-voltage (CV) mode it holds the set voltage and lets the load draw whatever current it needs; when the draw reaches the set current ceiling, the supply switches to constant-current (CC) mode, holding current steady and letting voltage sag instead. Most modern units light a CC indicator or change display colour when this happens — and learning to read that transition is half the skill of using one.

Why the current limit is everything

The current limit is the single most important safety feature when bringing up a repaired hashboard. By setting a conservative cap, a technician can apply power to a board with an unknown fault without instantly destroying it, the supply, or the fresh repair. If the board harbours a short circuit, the supply simply slams into CC mode and the current pegs at the limit rather than dumping unlimited amps into the fault. The first seconds of power-up become a diagnostic reading instead of a gamble: a board that should idle at a fraction of an amp but immediately hits the limit is announcing a short before any further damage is done, while near-zero draw on a board that should wake up points to an open enable line or a dead rail upstream.

Using it to diagnose a hashboard

The bench supply is the starting point of structured hashboard diagnosis. Power the board through the limited supply, watch the current, and interpret: expected idle draw means you can proceed to probing each voltage domain with a multimeter to localise a fault; a pegged limit means stop and find the short first. A classic bench technique for locating a shorted component is to feed the shorted rail a low voltage at a controlled current and let the fault reveal itself by heat — a thermal camera, freeze spray, or a careful fingertip finds the component cooking itself. Because current is capped, the board tolerates this interrogation. It is, simply, the safest way to wake a board up — which is why the repair-bench rule is: unknown board, current-limited supply, every time, and never straight onto a multi-kilowatt miner PSU that will happily deliver hundreds of amps into a fault.

Sizing one for ASIC work

For control boards, fans, and rail-level diagnosis, a modest adjustable supply covering the low-voltage rails with a few amps of headroom does the job. Full hashboard testing is another matter: a modern board hashing under load draws enormous current, and test procedures that actually exercise the chips call for a supply in the 50 A class at board voltage. Many shops run a two-tier bench — a small precision supply for diagnosis at low current, and a high-current unit (or the miner's own APW12-class PSU, once the board has proven safe) for load testing. Whatever the size, the discipline is identical: set the voltage, set the current limit deliberately, and treat the CC light as information, not annoyance.

Bench habits that save boards

A few rituals make the tool trustworthy. Set voltage and current limit before connecting the board — with outputs off, dial the voltage, short the leads or use the preset mode to set the current ceiling, then connect and enable. Verify the actual voltage at the board with a multimeter rather than trusting the front panel, since thin test leads drop measurable voltage under load. Keep polarity idiot-proof with color discipline and labeled harnesses, because reversed leads kill boards faster than any short. After power-down, remember large bulk capacitors hold charge; give them a moment or discharge them deliberately before probing or reworking. And log what you saw — the current draw at first power-up is a fingerprint worth writing on the repair ticket, because the same board on the bench next month tells a story only if you recorded this chapter.

The bench supply works hand in hand with the voltage domain map, the multimeter, and a methodical approach to the short circuit in any repair workflow.