Solder Joint Fatigue

Solder joint fatigue is the gradual mechanical failure of a soldered connection under repeated cyclic strain. Whereas a brittle fracture happens in one event, fatigue accumulates damage over thousands of cycles: micro-cracks nucleate at a stress concentration, grow a little with each cycle, and eventually sever enough of the joint that resistance climbs and the connection becomes intermittent or fully open. It is one of the most common reasons aged boards develop flaky, hard-to-trace faults — and on Bitcoin mining hardware, with its heavy chips, high temperatures, and constant fan vibration, it is a leading cause of the classic dead or degraded hashboard.

Thermal versus mechanical cycling

The cyclic strain has two main sources. Thermal cycling drives strain through coefficient-of-thermal-expansion mismatch: the silicon chip, its substrate, and the FR-4 board all expand at different rates, so every heat-up and cool-down shears the solder joints between them, with the outermost balls of a large BGA package taking the worst of it. Mechanical cycling comes from fan-induced vibration and resonance, board flex during handling and shipping, and repeated connector mating. In both cases the solder — a soft, creep-prone alloy — deforms plastically each cycle, and that accumulated plastic work is what consumes the joint's life. This is why a miner that is power-cycled daily ages its joints far faster than one that runs continuously at a steady temperature: the damage is counted in cycles and cycle depth, not hours.

Diagnosis on the bench

Fatigued joints often look dull, ringed, or visibly cracked under magnification, and a fault that comes and goes with temperature or with light finger pressure on a package is the classic symptom. A board that hashes when cold and drops chips as it warms — or the reverse — should put joint fatigue near the top of the suspect list. Techniques range from thermal probing and gentle mechanical stress while watching the kernel log, to powering the board from a bench power supply and mapping which chips disappear. Distinguish fatigue from its assembly-time cousins: a cold joint was weak from day one, while a fatigued joint was healthy and wore out.

Repair and prevention

Repair means renewing the joint: fresh solder and flux on discrete parts, or full hot-air rework and reballing for BGA-mounted ASICs whose cracked balls cannot be reached with an iron. Done properly with controlled reflow profiles, a reworked joint is as strong as the original. The better defense is reducing the strain in the first place: keep operating temperature stable rather than merely low, avoid unnecessary power cycling, dampen vibration, mechanically support heavy heatsinks and connectors, and fix failing fans promptly since a wobbling fan is a vibration generator bolted directly to the board. For fleets, steady 24/7 operation at moderate power is the longevity sweet spot.

For the underlying driver of thermal-cycle damage, see thermal fatigue; for a defect present from assembly, see dry joint. If a board has you beaten, our repair service handles exactly this class of fault daily.

Fleet-level implications

Fatigue thinking changes how you operate a fleet, not just how you fix one board. Shipping is a fatigue event — boards travel best with boards supported and machines packed so nothing flexes; a unit that "worked before shipping" and arrives flaky very often suffered its final crack in transit. Firmware policy matters too: aggressive auto-restart loops that hard-cycle a faulted machine dozens of times a day are quietly consuming joint life across every package on the board, which argues for diagnosing root causes over letting a watchdog thrash. And when planning heat-reuse setups that modulate miner power with demand, favour gentle ramps over on-off cycling — the joints count every deep thermal swing, and smooth operation is free reliability.

Trace intermittent faults with the diode & voltage reference.