Bitaxe – BM1366 Not Responding After Power Surge

Unplug everything for a full 60 seconds before any recovery attempt. The TPS546-family regulator and the ESP32-S3 brownout watchdog can both latch into a fault state that a soft reboot will not clear. Main PSU, USB-C, fan connectors — all unplugged. Use the time to move to a different wall outlet on a different breaker from the one involved in the surge event. Empty the bulk input caps, force a true cold start, and change the upstream context before re-powering.

Try USB-C alone first — no main PSU connected. The ESP32-S3 is USB-bus-powered for bootloader access, so this isolates 'is the MCU alive' from 'is the rail alive.' Connect USB-C from PC to Bitaxe. If the PC enumerates any ESP32-S3 JTAG/serial device in Device Manager, `lsusb`, or `dmesg`, the MCU survived. If not, hold `BOOT` on the board, tap `RESET`, release `BOOT` — this forces the ROM bootloader, which is mask-ROM and cannot be damaged by firmware events. Still no enumeration means MCU-level damage and Tier 4.

Flash the factory image via the Bitaxe Web Flasher. If USB-C enumerated the device, open `bitaxeorg.github.io/bitaxe-web-flasher/` in Chrome or Edge (WebSerial required). Select the EXACT factory image for your board revision — Ultra 202/204/205/207, Supra 401, Gamma 601/602, Hex v303/v304 each have different images. Flashing the wrong one will not physically damage the board but will prevent correct boot until re-flashed correctly. Flash, wait ~30-60 seconds, disconnect USB-C, reconnect main PSU, watch for the Wi-Fi AP. Recovers surge-hit Bitaxes where damage was limited to a corrupted application flash partition.

Cold-boot from a known-good regulated PSU on a DIFFERENT outlet. Never power a surge-hit Bitaxe from the PSU that was attached during the surge until you have DMM-tested it — the PSU often eats the transient and fails subtly, passing slightly-wrong voltage afterwards that finishes the job on a marginal board. Swap to a regulated brick from Mean Well, CUI, or the D-Central Bitaxe bundle. Ultra/Supra/Gamma: `5 V / 3 A` minimum. Hex: `12 V / 5 A` minimum. Watch for Wi-Fi AP and OLED splash.

Reconfigure and monitor the board for 24 hours if it boots. A surge-recovered Bitaxe can look healthy for 5 minutes and then degrade as damaged components heat up or a marginal chip hits its thermal limit. Reconfigure Wi-Fi, pool, and any custom tuning, then monitor AxeOS status for: intermittent `Power Fault Detected`, HW% climbing above baseline (`<1%` on a healthy BM1366 Bitaxe), `Vin` sag under ramps, unexplained reboots, or chip temperature creeping up. Latent surge damage often shows up over days, not minutes. If anything looks marginal, reduce frequency/voltage and plan a bench assessment.

DMM the suspect PSU, open-circuit and under load. Unplug from the Bitaxe. DMM on DC volts, probe at the plug with no load. Ultra: expect `5.0-5.25 V`. Hex: expect `12.0-12.6 V`. Then apply a `2-5 A` load (12V halogen bulb for Hex, USB load bank for 5V, or an electronic load). Under load: Ultra `>= 4.8 V`, Hex `>= 11.5 V`. PSU reads `0 V`, oscillates, or sags materially = PSU is the primary casualty. Replace before powering the Bitaxe. Many 'surge killed everything' tickets resolve to 'the PSU saved the Bitaxe; buy a new PSU.'

Continuity-test the Bitaxe's input rail with the board fully disconnected from all power. DMM on continuity / diode mode. Probe tip-to-sleeve on the barrel jack (Hex) or VBUS-to-GND on USB-C (Ultra). Healthy: high-impedance, several kΩ to MΩ, with a brief charge-up spike. Shorted: continuous beep, resistance under `~10 Ω`. A shorted rail means the TPS546, the 5 V LDO, or a bulk cap failed shorted internally. Do not power a shorted board — the PSU's OCP will fire and may cascade further damage. Shorted = Tier 4.

Inspect the input connector and surrounding PCB under magnification (jeweller's loupe or USB microscope). A surge leaves subtle clues at its entry point: a vaporised trace section, a blackened TVS diode package, solder splash from a blown capacitor nearby, pitted contacts on the barrel jack or USB-C port, or black residue on the connector shell. Photograph anything suspicious for a D-Central repair note. Damaged connectors must be replaced before further testing — intermittent contact cascades into endless confusion.

Probe `Vin` live on a verified-good PSU. If the board boots, open AxeOS status and watch `Vin` across idle to full-hashing ramp. Expected: Ultra `5.0 V ± 0.1`, Hex `12.0 V ± 0.2`, stable across the ramp. A surge-damaged TPS546 or LDO will regulate badly — `Vin` may sag, oscillate, or spike during transients that a healthy board rides cleanly. Cross-check with a DMM on the rail; discrepancy between AxeOS and DMM suggests the ADC or a sense divider also took damage. This live probe distinguishes a fully-recovered board from one with latent damage.

Swap to a known-good PSU from another healthy Bitaxe to bisect PSU vs board. If you run multiple Bitaxes, borrow a proven-working PSU from a healthy unit, connect to the surge-hit board. Symptom moves with the PSU = PSU is the root cause. Symptom stays with the Bitaxe = board damage. Cheapest 5-minute diagnostic in the entire triage; also why D-Central's multi-Bitaxe bundles include a spare spec-matched PSU.

Scope the `Vin` rail and `Vcore` under load. A `50 MHz` handheld scope (Owon HDS242, Rigol DHO804) on the input rail during a frequency ramp reveals ripple and transients a DMM averages out. Healthy regulated PSU: `< 100 mV` pk-pk ripple on `Vin`. Surge-damaged TPS546: often `> 500 mV` ripple or intermittent undershoots that trigger `Power Fault Detected` without failing any DMM-averaged reading. Probe `Vcore` near BM1366 decoupling caps — stock `1.25-1.35 V`, clean and stable. Lost transient response = regulator damaged.

Check the TVS diode at the input. Locate the TVS near the barrel jack or USB-C VBUS (typically SOD-323 or SMA). DMM diode-test in-circuit, board unpowered. Healthy TVS: open-circuit both ways at normal DMM voltages. Failed-short TVS (common post-surge — it did its job but didn't survive): reads as a short. Desolder and replace with same-spec part (`$0.50-$2` each). A surprisingly common single-component fix on surge-hit Bitaxes that otherwise look dead. Restores input protection for the next transient.

Replace the 5 V LDO (Ultra/Supra/Gamma) or the TPS546 regulator (Hex, Gamma/GT) if damaged. Identify the part from board silkscreen and the bitaxeorg schematic repo. Hot-air desolder: preheat bottom `~150 °C`, top-side `~310-330 °C`, lift part, clean pads with braid and flux, reflow new part down. Keep peak temp under 3 seconds or adjacent caps lift. Post-replacement, DMM-verify the rail before powering the full board. Hot-air-station job, not an iron job.

Reflow or replace the BM1366 if `count_asic_chips = 0` after clean firmware flash and clean rails. Reflow first — transient damage occasionally cracks a solder joint on the BGA rather than killing silicon. Preheat `~150 °C`, top-side `~310-330 °C` for `~30 seconds`, cool naturally, retest. If reflow fails, replacement: desolder, clean pads, place fresh BM1366 from D-Central graded inventory or trusted supplier, reflow down. On Hex with `count_asic_chips = 1-5`, replace the dead chips (usually the ones closest to the input rail) — the others are typically fine.

Inspect ESP32-S3 for USB-C-coupled ESD damage. If the board powers up but the ESP32-S3 no longer enumerates over USB-C, check the VBUS ferrite, the ESD protection diodes on D+/D-, and the `3.3 V` rail to the MCU. ESP32-S3 tolerates substantial ESD thanks to on-die protection, but a direct mains transient through the USB-C shield can punch through. Clean `3.3 V` at the MCU with MCU silent = MCU damaged, and replacement requires BGA-adjacent rework (Tier 4).

Re-flash firmware after any component replacement. Any hot-air rework can briefly raise flash temperature and corrupt data. After any Tier 3 swap, re-flash the factory image via Web Flasher before full-load validation. Use the correct factory image for the board revision (flashing the wrong variant is recoverable but confusing). Verify `count_asic_chips` matches expected before declaring the repair complete. Run 30-minute full-load burn-in before returning to production.

Stop DIY if the input rail shows a continuity-beep short with the board fully unpowered. A shorted `Vin` rail means a component failed shorted internally — desoldering BGA parts and large electrolytics without a hot-air station and a rework microscope is a reliable way to lift pads and turn a `$45` parts-swap into a `$150` replacement board. Ship the board to D-Central. Our bench handles this class of work every thunderstorm season.

Stop DIY on visible thermal damage. Scorched packages, bulged caps, blackened LDO or TPS546, charred PCB — bench-rework jobs. Iron-level work compounds the damage; hot-air with proper board preheat and a rework microscope is mandatory. The board is usually worth saving (ESP32-S3, OLED, EMC2101, connectors, PCB are all reusable), but only with a proper process. Pack anti-static, ship to D-Central.

Stop DIY if the ESP32-S3 doesn't enumerate in forced-bootloader mode. A dead ESP32-S3 needs replacement — the part is a QFN package, technically reworkable with skilled iron-plus-hot-air, but post-replacement bring-up (fuse verification, peripheral re-init) is bench-only. D-Central can also harvest the OLED, connectors, and passives to another board if ESP replacement isn't economically worthwhile for your specific case.

Ship to D-Central with the PSU and the surge context. Pack the Bitaxe in an anti-static bag, include the original PSU (so we can reproduce the fault with your exact stack), and attach a note describing the surge event (thunderstorm? UPS switch? other devices affected?), your diagnostic steps taken, and observed symptoms. Canada-wide shipping; US/international welcomed. Turnaround `5-10` business days. D-Central pioneered Bitaxe ecosystem — original Mesh Stand, first heatsinks, deep bench history on surge-damaged boards through Canadian thunderstorm season.

Consider the economics before committing to a full repair. Tier 4 surge-damage repair averages `CAD $60-$140` depending on which components failed. Ultra replacement: `CAD $130-$180`. Hex replacement: `CAD $220-$320`. If multiple components failed (TPS546 + BM1366 + ESP32-S3 all dead after a severe strike), replacement is often the better economic call. Bitaxe is open-source hardware — bench time to recover a triple-component-death isn't always worth it. D-Central will quote honestly; ask us before committing.