Antminer – APW PSU Fan Failure

Kill power at the wall breaker (not the miner button), wait 60 seconds, then shop-vac the PSU intake grille and output vents without opening the case. This clears ~40% of PSU_FAN_ERR tickets that come through D-Central's bench. Dust builds as a felt mat on the intake, starving the fan of airflow and letting lint wrap around the bearing shaft. Reconnect, power up, watch the fan for 30 seconds. If it spins steady and the error clears within 10 minutes, you are done.

Verify ambient at the PSU intake is at or below 35 °C using an IR thermometer or a cheap digital thermometer held at the grille. Not room-middle, not the hallway — at the intake. Move a circulation fan, open a door, or relocate the miner away from other heat-producing equipment. A PSU fan running full-tilt at 38 °C ambient is a fan counting down to death, and fixing ambient often resurrects a fan that was marginal.

Confirm the PSU sits with at least 10 cm of clearance on every side, with the intake (fan side) unobstructed by cables, walls, or rack shelving. Miners shoved against walls, stacked, or sandwiched in rack units routinely choke their own PSU intakes. Move the PSU, re-run for 30 minutes, and watch the fan RPM behaviour — if the fan ramps and holds without stalling, you are back in business.

Reboot the miner via the dashboard. A PSU_FAN_ERR latched in firmware can persist after the underlying condition clears. A clean reboot lets the control board re-read the fan's tach signal and clear stale state. If the error returns within 5 minutes of reboot, the fan is genuinely not spinning or below the tach threshold — proceed to Tier 2. If it clears, run a 30-minute burn-in to confirm.

Check firmware version at the Bitmain support portal (support.bitmain.com/downloads). A handful of old firmware builds misread APW tach signals on edge-case units. If you are on a known-buggy build for your hardware revision, roll one version back or forward. This is a 15-minute test that rules firmware in or out cheaply — and if you are running stock Bitmain firmware, now is also the right time to consider DCENT_OS for proper PSU fan RPM visibility going forward.

Power off at the breaker and wait 60 seconds. Open the PSU enclosure (4-6 screws, side panel first). Do NOT touch the primary-side bulk capacitors — those are the two large cylindrical caps on the mains-input side, holding 400 V+ of stored charge even with mains disconnected. Stay on the output (low-voltage) side of the isolation barrier, which is visually marked on the PCB by a long slot or a clear silkscreen line. If you are not comfortable identifying the isolation boundary, stop and go to Tier 4.

Inspect the fan cable and header on the PSU control PCB. Look for cracked solder joints on the PCB side of the header, oxidation on the header pins (green/white crust), a crimped or pinched wire, or a cable yanked out at the fan end. Reseat the connector firmly. If you spot a damaged wire, a strip-splice-heatshrink repair is acceptable; if the PCB joint is cracked, bench-reflow it with a soldering iron (low-voltage side only, remember).

Measure 12 V at the fan header pins with a multimeter while powered up (miner disconnected from PSU output for safety — you just need the PSU running, not loaded). Probe carefully with insulated leads. Voltage present plus fan not spinning equals fan is dead — go to Step 9. No voltage at header equals PSU's fan-drive circuit has failed — go to Tier 3 or Tier 4.

If the fan is confirmed dead, source a direct replacement: stock APW3++/APW7 uses an 80 x 80 x 25 mm 12 V DC fan in the 0.5-0.8 A class (common options: Delta AFB0812SH, Sunon PMD1208PMB, or equivalents rated ≥ 4,500 RPM and ≥ 50 CFM). APW9/APW12 uses the same form factor but slightly higher static-pressure spec. Match airflow (CFM) and voltage, not brand. Swap in, route the cable clear of the HV side, reassemble.

Reassemble the PSU. Before putting the cover back on, power up one last time outside the case (stand to the side, hands out of the enclosure). Verify: fan spins up at startup, stays on, and tachs correctly in firmware. If good, close it up, reinstall in the miner, and run a 30-minute burn-in before walking away. Watch the dashboard for PSU_FAN_ERR clearing and staying cleared.

Run an under-load 12 V output test. With the PSU back in the miner and the miner hashing at full power, probe the +12 V and GND pins on the PSU output connector with a multimeter. Expected: ≥ 12.2 V on APW3++/APW7, ≥ 13.6 V on APW9/APW12 sustained. Sag below those figures means the switcher and caps were stressed by the fan failure — you saved the fan but the PSU is tired. Budget for replacement within 30-90 days or go to Tier 3.

If the fan is good on a bench supply but no 12 V reaches the fan header on-PSU, the fan-drive FET or driver IC on the PSU's control PCB has failed. Locate it: on APW9/APW12 it's typically a small SOT-23 or SOT-89 MOSFET near the fan header, sometimes labelled 2N7002 or a generic small-signal FET. Probe with a multimeter diode-mode to check for shorted or open junction. Replace with an equivalent-spec part. Requires SMD soldering experience.

Inspect the output-side electrolytic capacitors. If the fan failed long enough for the PSU to cook itself, the low-voltage output caps are usually the first casualty. Look for dome-topped caps, leaked electrolyte residue, or discoloration. ESR-meter the caps if you have one. Replace any caps out of spec with 105 °C-rated low-ESR electrolytics of matching capacitance and voltage (typically 3300 µF / 25 V or similar — verify on the cap body before ordering).

Refresh thermal paste on the output-stage MOSFETs and the switching controller IC if they're thermally coupled to the PSU's internal heatsinks. Stock paste dries out after 3-5 years of continuous hot-running. Arctic MX-6 or Thermal Grizzly Kryonaut, thin uniform layer, reassemble. This is a worthwhile preventative step even if the fan was the only failed component — the PSU likely ran hot for days before the fan quit.

Before closing up, clean everything with compressed air and isopropyl alcohol (99% IPA, lint-free wipes). PSU internals accumulate conductive dust that reduces creepage distance on the high-voltage side — a fire-safety issue, not a performance one. Mining-operation fires traced back more often than not to undetected PSU dust accumulation. Never reassemble a PSU with visible dust on the HV side of the board.

Stop DIY: you don't have experience safely discharging 400 V+ bulk caps, the fan-drive circuit has failed and you can't identify the part, the PSU shows output-side damage (bulging caps, burnt components, char marks), or you've swapped the fan and the replacement died within a week (root cause is broader thermal failure, not the fan). Book a D-Central ASIC Repair slot and ship the PSU in.

What D-Central does at the bench: bench discharge of bulk caps; full fan + output-cap refresh as a matter of course on APW units over 3 years old; paste refresh on all heat-coupled semis; HV isolation test before and after; 12 V output stability check under programmable load up to the PSU's rated wattage; 24-hour burn-in at nameplate. Typical turnaround 5-10 business days, Canada-wide with US/international shipping welcomed.

Ship safely: pack the PSU in anti-static bubble wrap (aluminized foil works in a pinch), double-box with at least 5 cm of foam on every side, and include a note with symptoms observed, age of unit, miner model(s) it was running, and your contact. Context saves us diagnostic time, which saves you money. Insure for replacement value — APW12 units are not always in stock on the secondary market.