If your all-in-one Antminer is completely dead — fan won’t spin, no network-port LED, no green light on the control board — the fault is almost always in the power delivery chain, not the hashboards. This guide walks that chain from the wall outlet to the control board so you can isolate exactly where power stops: a bad cord, a tripped breaker, a failed PSU, or a dead control board. Work through it in order with a multimeter and you’ll find the break instead of swapping parts blind.
Before You Start: The Dual-Cord Rule
The integrated (“all-in-one”) power supplies on the 17-series (APW9/APW9+) and 19-series (APW12) machines are dual-input units: they take two AC cords and only produce output when both cords are powered simultaneously. If only one cord is live, the PSU stays dark and the miner appears dead — this is a requirement for two energized inputs, not a claim about a single outlet’s capacity. Input voltage differs by family: the APW9/APW9+ will run on 110/120V (at reduced output), while the APW12 is 200–240V input only. The dual-cord rule accounts for a large share of “dead miner” reports.
Safety: What Is and Isn’t Dangerous
The low-voltage DC side of the PSU (the 12–21V bus-bar / power lines feeding the hashboards and control board) is not a shock hazard — but it is a very high-current source that will arc, spark-weld a dropped tool, and cause burns if you short it. The real danger is inside the sealed PSU shell: the mains input and PFC capacitors hold roughly 410–420V DC and can stay charged after unplugging. Do not open the PSU casing unless you are qualified to discharge those caps and the unit has been unplugged long enough to bleed down. Every step below that involves opening the PSU assumes you understand that risk; if you don’t, stop at Step 5 and swap the whole PSU instead.
Fault Isolation, Step by Step
Step 1 — Reseat every connection
Power the unit down and unplug it. Re-seat both AC cords at the wall and at the PSU inlets, and confirm every plug clicks in tight. Loose or partially-inserted cords are the most common cause of a no-power miner. Re-power and re-test before going further.
Step 2 — Watch the PSU fan
With both cords live, look at the PSU’s own cooling fan. If the fan does not spin, the PSU is getting no usable AC input or has failed internally — continue at Step 3. If the fan spins but the miner still won’t boot, jump to Step 6.
Step 3 — Verify AC at the outlets
Set your meter to AC volts and measure both outlets feeding the two cords. You should read your line voltage (nominally 200–240V for these machines). If either outlet reads zero or low, the problem is upstream — a tripped breaker, a bad receptacle, or an over-current shutoff — and an electrician should check the circuit and breaker. No meter? Move the miner to a known-good outlet or plug a known-good load into the suspect one to confirm.
Step 4 — Verify AC at the cord ends
If both outlets read correct voltage, measure the AC at the far (PSU-side) end of each cord. No voltage at the cord end with good voltage at the outlet means the cord is broken internally — replace that cord. If you can’t measure, simply swap in known-good cords one at a time.
Step 5 — Replace the PSU
If both outlets and both cords deliver correct AC but the PSU fan still won’t spin and there’s no DC output, the power supply itself has failed. Replace it with the correct model for your machine (APW9/APW9+ for 17-series, APW12 for 19-series — they are not interchangeable across voltage families). This is the safe stopping point if you’re not equipped to open the shell.
Step 6 — Fan spins but miner won’t boot: check the control-board rail
If the PSU fan runs but the machine never comes up, the fault is likely the control-board power feed or the control board itself. Power down and unplug, open the miner and (if qualified — see Safety) the PSU cover, and unplug the control-board power lead. Re-power and measure that rail: on an APW12 the control board + fans run on a dedicated 12V output (separate from the higher-voltage hashboard output). No 12V on that rail means the PSU’s control-board output is dead — replace the PSU.
Bench-test caveat: when you enable a Bitmain PSU on the bench by grounding its EN (enable) pin with no I2C control board attached, it comes up at its default maximum voltage on the hashboard output — roughly 15.2V on an APW12, ~21.3V on an APW9 — and it is the miner’s firmware that lowers that output to the hashboard target at boot. So a healthy bench PSU reads high on the main output, not the running value; that is normal. The separate 12V control-board rail stays at 12V.
Step 7 — Rail is good, machine still dead
If the control-board rail reads a healthy 12V, power down, re-seat the control-board power cord and the voltage/sense (regulator) lead, and try again. If the miner still won’t boot with confirmed good power on every rail, the control board has failed and should be replaced.
Step 8 — Done
At this point you’ve traced power from the wall, through both cords, through the PSU, to the control board, and identified the failed link. Reassemble, restore both cords, and confirm the fan spins and the control-board LED and network light return.
Common Mistakes
- Testing with one cord. These PSUs need both cords live — a single-cord test always reads “dead.”
- Expecting the running voltage on the bench. A bench-enabled Bitmain PSU sits at default max; only firmware sets the working hashboard voltage.
- Opening a PSU you just unplugged. The primary-side caps hold ~410–420V; give them time and only open if you can discharge them.
- Mixing PSU families. An APW9 will not correctly power a 19-series board and vice-versa; match the PSU to the machine.
- Blaming the hashboards first. No fan and no LED is a power-chain fault, not a hashboard fault — start at the wall.
Related
Confirm your exact model’s wiring and PSU pairing in the miner manuals before you open anything, and if the fault turns out to be downstream on a board rather than the supply, run the ASIC fault finder to localize it. Need a replacement PSU, cord, or control board? See ASIC repair parts, or hand the whole job to a bench tech through start a repair.
