MOSFET (Power Switching Transistor)

A MOSFET (metal-oxide-semiconductor field-effect transistor) is a voltage-controlled switch and the most common power-handling device on a mining ASIC's hashboard and inside its PSU. A small voltage applied to the insulated gate opens or closes a conduction channel between the source and drain, letting the part switch large currents on and off millions of times per second while drawing almost no gate current of its own. That combination of high speed and low control power is exactly what a switch-mode converter needs, which is why a single hashboard can carry dozens of them.

It helps to picture where these parts sit in the power chain. Wall power enters the PSU, is rectified and switched down to the low DC voltage the boards run on, and then each hashboard's own converters step that voltage down again to the fraction of a volt the ASICs actually need. MOSFETs are the workhorses at every one of those switching stages, turning current on and off fast enough that a modest inductor and capacitor can smooth the resulting pulses into a clean rail. Because they carry the full load current at each stage, they are also where the heat and the electrical stress concentrate, which is exactly why they sit at the top of the suspect list the moment a board refuses to power up or trips its protection.

Why it matters for hashboard repair

Each voltage domain on a hashboard is fed by a buck or boost converter, and the switching MOSFETs are among the parts most often found dead after a board failure. When a MOSFET fails short it usually drags its rail down and can blow the upstream fuse or trip the PSU's over-current protection — the classic "board draws current but produces zero hash" symptom. A failed boost stage often shows up as no domain voltages at all, with the MOSFET, inductor, or diode as the prime suspects. This is a per-domain power structure, not a per-chip one: several chips share each converter, so a single blown transistor can silently take a whole cluster of ASICs offline while the rest of the board looks fine.

Testing one on the bench

Technicians test the device in diode mode on a multimeter and measure resistance across gate, source, and drain. A dead short between any two of those pins condemns the part, and a leaky gate oxide — a low but non-zero gate-to-source resistance — is a common silent killer that leaves a device half-on and cooking itself. Because the transistors on a rail often sit in parallel, one shorted part can make its neighbours read faulty too, so the honest procedure is to lift or remove a suspect and retest rather than trust an in-circuit reading. A thermal camera or even a fingertip on a briefly powered board finds the hot part fast.

N-channel vs P-channel

N-channel devices conduct when the gate is driven positive relative to the source and carry the bulk of the load current; P-channel parts conduct on a negative gate drive and are often used for high-side switching or reverse-polarity protection. The key spec for power work is R_DS(on), the on-state resistance — lower values mean less heat and higher efficiency, which is critical when a single domain may draw tens of amps. Switching losses matter too: the faster and cleaner the gate transition, the less energy the device wastes each cycle heating itself rather than the load.

Replacing one correctly

Swapping a MOSFET requires matching voltage rating, current rating, package, and gate-threshold characteristics. A mismatch — say a part whose threshold the gate driver cannot fully overcome — can leave the device permanently half-on, overheating within seconds and taking neighbouring components with it. Always check the filter capacitors and the gate-drive circuit before assuming the transistor alone was at fault, since a failed driver will simply kill the replacement too. Understanding the buck converter stage the MOSFET lives in is the first step to diagnosing why a board no longer powers up; when the fault is beyond hand tools, our repair intake is built for exactly this class of work.