Trace Current Capacity

Trace current capacity, or ampacity, is the maximum current a PCB trace can carry before it heats beyond an acceptable temperature rise. A trace has finite resistance, so current through it dissipates power as heat; the trace settles at whatever temperature lets it shed that heat to its surroundings. Capacity is therefore always quoted against a target temperature rise — such as 10°C or 20°C above ambient — never as a single hard limit. The same trace that is fine at a 20°C rise may be undersized in a hot enclosure where ambient is already elevated, which describes the inside of every running ASIC miner.

How it is calculated

The governing reference is IPC-2152, published in 2009, which replaced the older formulas of IPC-2221 with charts derived from extensive testing of real boards. It relates four main variables: trace cross-sectional area (width times copper thickness), current, temperature rise, and whether the trace is on an external or internal layer. Internal traces generally run hotter for the same current because laminate insulates them, while external traces shed heat to air. IPC-2152 also accounts for what the older standard ignored: nearby copper planes and vias that conduct heat away from the trace, board thickness, and even whether the board is in still air. Because real boards almost always have planes nearby, IPC-2152 often permits narrower traces than the conservative IPC-2221 numbers for the same current — but the reverse caution applies to isolated traces with no copper around them. Capacity scales directly with copper weight: doubling from 1 oz to 2 oz copper doubles the cross-section of a trace of the same width.

Why it matters on hashboards

Mining hardware is an extreme case of the problem. A modern hashboard feeds its ASIC chain from DC-DC conversion stages, and the currents involved are enormous: the low-voltage side of each hash domain runs at well under a volt while the board as a whole dissipates kilowatts, so the copper carrying that power is doing serious work. This is why hashboards use heavy copper, wide pours rather than discrete traces for power delivery, and stitched vias to spread current across layers. A power path sized too thin becomes a hot spot, and heat compounds the problem: copper's resistance rises with temperature, so a hot trace dissipates more, gets hotter, accelerates laminate degradation, and in the extreme lifts or vaporizes the copper — the burnt, delaminated power sections seen on boards that suffered a short or a failed regulation stage.

Repair implications

Two bench rules follow. First, a jumper that bypasses a damaged trace must match the original current capacity: a thin wire strapped across a blown power trace simply relocates the failure and adds a fire risk, so size the wire for the actual current — on power paths that often means heavy-gauge wire or a rebuilt copper section, not a strand of wrapping wire. Signal traces are forgiving; power traces are not, and the difference is knowing which one you are looking at. Second, respect what surrounding copper was doing: a repair that severs a trace's connection to the plane that was heatsinking it changes its thermal environment, and the follow-on failure can appear centimeters from the original fault. High-current return paths likewise depend on via stitching to move current between layers; damaged or corroded stitching concentrates current in the survivors, overheating them in turn.

Undersized copper is a slow failure with a long fuse — exactly the kind that shows up two months after a repair. When a board's power section is beyond a clean bench fix, start a repair with D-Central rather than gambling a jumper on a kilowatt path.