Whatsminer M30S++ – Buck-Boost Regulator Failure

Hard AC power-cycle for 60 seconds, not at the chassis switch. Kill AC at the mains or PDU, wait a full minute for primary-side bulk caps in the P21 PSU and downstream board capacitors to discharge. Boot once. If the fault clears and stays clear for hours, you may have caught a transient power-up race. If it returns within minutes, the fault is hardware - proceed. Log the timestamp; recurrence pattern matters for diagnosis.

Confirm fault via BTMiner API, not just Web UI. From a laptop on the same subnet: `curl http://<ip>:4028 -d '{"cmd":"summary"}'`. Parse the JSON for per-board hashrate and chip count. The Web UI sometimes shows stale state from the previous boot; the API always shows current. Log timestamps so you can spot patterns later - single-event vs recurring matters for the diagnostic tree.

Check `log/btminer.log` over SSH for power and chip-init errors. Look for lines matching `hashboard X: power init failed`, `hashboard X: voltage out of range`, `hashboard X: chip id read failure`, or `hashboard X: enumerate timeout`. The exact wording locates which stage of the boot sequence the affected board fails at, which narrows the suspect failure mode before you ever open the chassis.

Verify chassis intake ambient with an IR thermometer at the front grille - not room-middle. Above 35 C ambient on an M30S++ stresses every power stage on every board, and a marginal buck-boost converter will fail under heat that a healthy one tolerates. Canadian operators October-April: this is essentially never your problem. Warm-climate or summer operators: this is often the trigger that revealed an already-aging stage.

Roll firmware back if the fault appeared right after a BTMiner firmware update. SD-card recovery procedure from MicroBT support, previous BTMiner version. A handful of builds shipped power-profile tweaks that triggered marginal stages on M30S++ hardware. Verify firmware-to-hardware-revision match before flashing - wrong firmware bricks the control board.

Swap the suspect board into a known-good slot and a healthy board into the suspect slot. Power off at AC mains, label the three slots 0/1/2 with tape, swap, boot, observe `summary` and `btminer.log`. Fault follows the board = board fault, continue. Fault stays in the slot = control-board or chassis-wiring fault, switch to the control-board troubleshooting tree and stop here.

Inspect the buck-boost row visually under bright light + 10x loupe. Board out, anti-static mat. Walk the row of MOSFETs, inductors, and output cap banks (typically 6-10 stages depending on hardware revision). Look for discoloration, lifted pads, bulged caps, cracked MLCCs, scorched silkscreen, melted inductor shrouds. Smell the board - fried MOSFET, saturated inductor, and cooked electrolyte each have distinct signatures. Photograph anything suspicious.

IR thermal scan under brief boot. Reinstall the board with the chassis open. Boot the miner. Within 30-60 seconds of power-up, scan the buck-boost row with an IR thermometer or thermal camera. Working stages run 50-75 C; failed-open stages stay at ambient (~25-35 C); failed-shorted stages spike past 100 C in seconds. Kill power IMMEDIATELY if you find a hot-shorted stage - every additional second of power risks adjacent components.

Multimeter check on the suspect rail with board powered briefly. Probe at the inductor-output pad (or labeled test point if your board revision has one). Compare to the same point on a known-good domain on the same board to set expectation. 0 V = open MOSFET or open inductor; bus voltage (12-15 V) = shorted high-side MOSFET, KILL POWER NOW; oscillating = controller / feedback fault. Healthy is 0.30-0.42 V per BM1262 domain.

Multimeter check off-power for shorted components. Power off, wait 60s, verify discharged. Continuity / diode mode. Check each MOSFET in the suspect stage source-to-drain both directions; low resistance = shorted, mark for replacement. Check the inductor for continuity (open = replace; healthy <50 milliohm). Check output caps to ground for shorts; identify cracked MLCC by individual probing or by visual inspection under loupe.

Replace the failed buck-boost MOSFET. Hot-air rework station: preheat bottom-side to 150 C, top-side hot air 310-330 C for ~30s, lift the failed MOSFET with tweezers, clean pads with flux + braid, place new MOSFET (orientation matters - check silkscreen and pin-1 dot), reflow. Match the original's rated voltage, current, and package precisely. D-Central stocks salvaged-grade and new-old-stock MOSFETs for M30S++ buck-boost stages.

Replace a saturated or open inductor. Hot-air the failed inductor off (preheat 150 C bottom, top-side 320-340 C - power inductors have substantial thermal mass and need higher temps than MOSFETs). Clean pads. Place new inductor matching inductance (typically 0.22-1.0 uH for this stage), saturation current rating (must equal or exceed original), and DCR (must equal or be lower than original). Wrong-spec inductor = the rail won't regulate cleanly under load.

Replace cracked MLCC or dried bulk capacitor on the output bank. MLCCs: hot-air or fine-tip soldering iron at 360 C, lift, clean, place - match capacitance, voltage, and dielectric (X7R for this application). Bulk caps (polymer or electrolytic): match capacitance, voltage, ESR, and ripple-current rating. Watch polarity on electrolytics. Replace the whole bank on a domain rather than just the visible failure - caps in the same bank age at similar rates.

Reflow suspect joints if no failed component is identified. If Step 10 finds no failed component but Step 9 confirmed a dead rail under load, the fault is most likely a cracked solder joint at a MOSFET source pad or inductor pad. Add flux to every joint in the suspect stage, hot-air the entire stage at 320-340 C for 30-45s, let cool naturally. Re-test under load. Reflow restores ~60% of joint-cracking faults; the remaining 40% need test-fixture diagnosis.

Apply fresh thermal interface to the chip-domain heatsinks before re-assembly. While the board is out, refresh thermal pads or paste under the BM1262 chip array if dried, cracked, or unevenly compressed. Arctic MX-6 or Thermal Grizzly Kryonaut for paste; properly-rated thermal pads for any pad-interfaced positions. Don't substitute thinner / thicker pads - pad thickness matters for chip-to-heatsink contact pressure.

Stop DIY when any of: PWM controller IC is suspected; two adjacent buck-boost stages have failed (cascade damage pattern, suggests upstream fault); you have reflowed a stage and the fault returned within 30 days; visible damage extends to traces / vias / multiple components; ESR test shows multiple caps out of spec on the same domain; downstream BM1262 chips were exposed to bus voltage during a shorted-MOSFET event. You are now in test-fixture territory. Book a D-Central bench slot.

D-Central bench process: programmable DC source replaces your P21 / chassis bus so we can ramp voltage and current independently and reproduce intermittent faults that home gear can't. Per-domain rail measurement with isolation between stages. Salvaged-grade and new-old-stock parts inventory for M30S++ buck-boost stages. PWM controller swap when needed. Post-repair burn-in: 24 hours at full nameplate hashrate on a test rig before the board ships back. Photo-documented diagnostic sent with quote.

Ship safely. Anti-static bag the hashboard. Double-box with at least 5 cm of foam every side. Include in the box: print of your last `summary` API capture, `btminer.log` extract showing the fault, note describing observed symptoms / firmware version / recent events / contact info. Match chassis serial to hashboard serial in your ship note - both are on metal plates, don't guess. Book at d-central.tech/services/asic-repair/. Turnaround 5-10 business days for isolated single-board buck-boost faults.