APW3++ No 12V Output and Fan Not Spinning

Verify the wall outlet is alive: plug a known-good appliance (lamp, laptop charger) into the same outlet. Test the C13/C14 cord by powering a different appliance with it. APW3++ inlets fatigue and miner-grade cords corrode at the contacts — this rules out 50% of 'dead PSU' tickets in 30 seconds. If the outlet is dead, fix that first. If the cord shows continuity above `1 Ω` from end to end, replace it with a 14-AWG-or-better cord rated for 15 A. Try a different outlet on a different breaker if you suspect line issues.

Check ambient temperature at the PSU intake. If your meter reads under `10 °C`, this is the documented APW3-series cold-start lockout — a thermal characteristic, not a defect. A bitcointalk operator running 80+ units reported less than a third would boot at near-0 °C and most worked at 10 °C. Move the PSU into a heated room above `15 °C` for 30 minutes and retest at the same outlet. If it boots warm, the fix going forward is enclosure heat retention, exhaust recirculation, or relocating the rig — not component repair.

Power-cycle properly to clear any latched protect state. Unplug AC, wait until the fan has fully stopped if it was spinning at all, wait an additional `60 seconds` for bulk-capacitor discharge through internal bleeder resistors, then plug back in. Some APW3++ revisions latch into protect after an over-current event and need this full discharge cycle to clear. Bitmain documents a 10-minute discharge wait for stuck-protect mode — observe it on the second cycle if the first does not clear.

Try a different miner load. If you have access to a second S9-class miner or a known-good hashboard, plug the suspect PSU into that. Some failure modes only appear under load, but a totally dead APW3++ stays dead with anything attached or detached. This step rules out a subtle mismatch with the original miner's PSU sense circuit. If the suspect PSU also fails to start with a different load, the fault is internal and Tier 2 begins.

Bench-test the PSU with no load attached. Disconnect every PCIe lead from the hashboards and the controller. Apply AC and flip the switch. A healthy APW3++ spins its fan within `1-2 seconds` of AC inrush — it is self-starting, no PS_ON signal needed. With a multimeter on DC, probe a 6-pin output bullet (red to black depending on the connector convention): healthy = `12.0-12.6 V` no-load. Dead = no voltage at all. This isolates 'PSU is broken' from 'miner's PSU sense circuit is broken' definitively.

Inspect the PSU casing for visible damage. Look for burn marks around the IEC inlet, scorching on the heat-sink fins, swollen or vented top-side electrolytic caps visible through the vent slots, and melted plastic at any output bullet. Photograph everything before opening for warranty/return purposes. Any of these findings = component-level repair territory; if you are not equipped, package and ship to D-Central. Test the C13 cord with a multimeter: continuity from each prong on the IEC connector to the corresponding prong on the wall plug should read under `1 Ω`. Higher = corroded contacts or broken strands.

Verify input voltage at the inlet under live conditions. With a clamp meter or multimeter on the live wire feeding the PSU, confirm `108-126 V` on a 120 V circuit or `205-250 V` on a 240 V circuit. Anything outside that range and the PSU may refuse to start or trip into protect. Dedicated 240 V circuits are the right answer for an APW3++ — you get the full 1600 W output, half the current draw, lower cable heating, and longer connector life. The 110 V mode is a fallback, not a target.

Pull the cooling fan and connect a known-good 12 V fan to the same 4-pin connector with AC applied. If the substitute fan spins, your original fan was dead — a $5 swap fixes the unit (some APW3++ revisions stop with a dead-fan condition because the housekeeping micro reads zero RPM). If neither fan spins, the auxiliary 12 V supply on the primary side is dead, and you need to proceed to Tier 3 cover-off work.

Remove the four casing screws and lift the top cover. Discharge the bulk capacitor before touching anything: place a `10 kΩ 5W` resistor across the bulk cap terminals for 30 seconds, then verify with a DMM that the cap reads under `10 V` DC before proceeding. The APW3++ bulk cap holds approximately `380 V DC` after a normal startup — a probe slip on a charged cap is a hospital trip. Inspect the primary PCB for burn marks, swollen caps, scorching around `Q4`, `D7`, `U2`, and discoloration on the heatsink-mounted devices. Photograph the layout for reassembly reference.

Continuity-check the AC inlet fuse. Located on the primary PCB near the IEC inlet — usually a slow-blow ceramic body in a clip or solder pads, typically 15-20 A. Closed (continuity) = good fuse, the fault is downstream. Open = blown fuse, the fault is upstream of or at the bridge rectifier. Critical: do not replace the fuse and re-power without first checking Steps 11 and 12. A blown fuse is a symptom; the cause is almost always a shorted bridge or shorted PFC MOSFET, and a fresh fuse will pop instantly under that condition.

Diode-test the bridge rectifier (`U2`) AC pins both directions with the unit unplugged and the bulk cap discharged. Healthy = ~`0.5 V` drop one way, open the other. Short either direction = replace with a `GBU808` or `KBU806` (8 A, 800 V) equivalent through-hole package. Diode-test PFC MOSFET `Q4` drain-to-source: should be open both directions on a healthy unit (a small reading is normal for body-diode breakdown). Short D-S = replace with the original 800 V N-channel or equivalent (`IPW60R125P6`, `STW45N60M2-4`, `STF12N60M2`). When `Q4` shorts, it almost always took the fuse with it — replace both.

Diode-test `D1`, `D2`, `D21`, `D22` and PFC boost diode `D7`. The Zeus Mining APW3 repair PDF lists these as the high-failure-rate diodes on the primary side — secondary causes for dead-unit symptoms when `Q4` and `U2` test fine. Any reading short or fully open out-of-spec = replace. Use the original part number where possible. `D7` is typically a fast-recovery boost diode such as `STTH8L06` or `SCS108AG`. Reflow with a hot-air station at `350 °C` for surface-mount; through-hole replacements on heat-sinked devices need fresh thermal paste on reassembly.

After replacing failed components, install a fresh fuse of the original rating (typically `15 A` slow-blow). First power-up: apply AC through a current-limiting bulb tester — a 60 W incandescent in series with the AC live conductor. If the bulb glows steady-bright, you have a remaining short and must continue diagnosis. If the bulb flashes once and dims to nearly off, the unit is charging the bulk cap normally and the repair held. Only after the bulb test passes should you apply unrestricted AC. Then verify `12.0-12.6 V` on the output bullets no-load and again under load with a known-good miner attached.

Stop and ship to D-Central ASIC Repair if any of the following: multiple primary components tested bad simultaneously (a transient took out the entire input stage), PCB shows scorching or lifted traces, you do not have an isolation transformer for extended primary-side work, you are not 100% confident discharging the 380 V bulk cap, or the unit is older than 5 years with visible secondary-side cap bulge (a planned full re-cap is faster as a shop job). Mainboard-level rework on a 380 V boost converter requires bench gear most home miners don't have. Book a repair slot at https://d-central.tech/services/asic-repair-services/ — flat-rate evaluation, parts at cost, return shipping included for North American customers.

Decide repair-vs-replace based on parts-and-labor math. If parts cost (typically `$15-30 CAD` for fuses + diodes + MOSFETs at distributor pricing) plus your labor exceeds 60% of replacement cost (`~$150 CAD` for a tested second-hand APW3++), buy a replacement. D-Central sells refurbished, bench-tested APW3++ units at https://d-central.tech/product/apw3-psu/ — every unit verified under load before it ships. For miners running S19-class loads who are upgrading anyway, consider a Delta DPS-2000 or DPS-2500 server PSU with a breakout board — battle-tested at 24/7 240 V continuous duty.