Short Circuit

A short circuit is an unintended low-resistance connection that lets current bypass its intended path and flow in an excessive, uncontrolled amount. On an ASIC hashboard, a short is one of the most common and most diagnosable faults: a shorted switching MOSFET, a failed ceramic capacitor, a shorted LDO regulator, or an internally damaged chip creates a near-zero-resistance path that drags a voltage rail toward ground and draws enormous current. Because a hashboard is built as a series string of voltage domains, a short in any one domain disturbs the whole chain — which is also what makes it findable.

How a short presents

Powered from a current-limited bench power supply, a board with a short immediately pegs the current limit instead of idling at a fraction of an amp — the unmistakable signature. With the board unpowered, a multimeter in resistance mode reads near zero between a rail and ground where it should read high; measuring across each domain in sequence localizes the fault, since a healthy domain shows a consistent voltage share while a shorted one reads at or near zero. A partial short is subtler: a domain measuring 100 mV or more below the average of its neighbors typically hides a failed chip or a broken-down filter capacitor that leaks current without being a dead short. In firmware terms, the same board often reports zero chips detected on that chain, because a shorted chip can pull shared signal lines low and silence everything behind it.

Finding the exact component

Once the faulty domain is identified, the hunt narrows to its power parts: the buck converter MOSFETs (a resistance check across the transistor pins reveals a blown part — a short between any two pins means MOSFET failure), the domain's LDO regulators (each domain carries LDOs for the 1.2 V and 0.8 V auxiliary rails, and a shorted LDO output is a classic failure), the ceramic filter capacitors, and finally the ASICs themselves. Thermal imaging is the great shortcut: raise voltage slowly on the current-limited supply and the shorted component self-identifies by heating first, showing as a bright hot spot on a thermal camera before anything else on the board warms. Under magnification, also check for foreign material — solder balls, tin bridges, flux residue — that can short adjacent pads without any component having failed at all.

Fixing it safely

Prevention is the quieter half of the discipline. Most shorts have mundane origins: conductive dust and metal debris settling across pads, condensation from moving a cold miner into warm humid air, corrosion in salty or damp environments, thermal-cycling fatigue cracking solder joints and capacitors, and voltage surges from failing PSUs taking MOSFETs with them. Clean intake air, letting hardware acclimate before power-up, stable power, and periodic inspection under magnification prevent more board deaths than any amount of skilled rework repairs. A repair bench sees the same handful of stories on repeat — and nearly every one of them was cheaper to prevent than to diagnose.

The repair is component-level: remove or replace the failed part with hot-air rework, clean the site, and re-verify resistance to ground before reapplying power. Two bench disciplines keep short-hunting safe. First, always bring a repaired board up on a current-limited bench supply, never straight on a 300 A-class mining PSU — the current limit caps the damage a remaining fault can do, while a full PSU will happily vaporize traces. Second, clip the meter's ground probe to true board ground, not to a heatsink, since probing heatsinks risks creating the very short you are hunting. Methodical, domain-by-domain measurement turns "dead board, draws huge current" from a mystery into a twenty-minute diagnosis — which is why the short circuit is the first fault every repair tech learns to chase, tying together the bench supply, the multimeter, the thermal camera, and the voltage-domain map of the board.