Antminer – Power Surge Damage

Unplug from the wall. Don't toggle the PSU switch, don't try another outlet, don't poke anything until you've done a proper inspection. A damaged PSU in a second power-on can take out a previously-healthy control board, and a damaged control board can cascade voltage glitches into hashboards. Power-off-and-leave-off is the single highest-ROI move on a surge-damaged miner. Document the event: time, weather, what tripped (breaker, GFCI, UPS).

Log what you see. Take phone photos of the PSU, control board (both sides if possible), every hashboard, and every cable connector. Note any burnt smell, soot, or discolouration. This record matters if you ship to D-Central — symptom-rich intake notes cut diagnostic time and your invoice. Save photos with event timestamp in the filename for insurance claims if applicable.

Check the circuit and other loads on it. Was the breaker tripped? Test other equipment on the same circuit — if a Wi-Fi router or monitor on the same outlet also died, the surge was upstream of the miner's PSU and simply swapping the PSU won't protect the replacement. Fix the circuit or add whole-panel surge protection before repowering anything on that line.

Check pool-side dashboard and firmware state remotely. Before disassembly, check your pool dashboard to confirm the miner dropped off at the event timestamp (not earlier, which would indicate a pre-existing fault). Note the last-reported hashrate and HW%. This establishes a clean before picture so you can tell a surge event apart from a slow-death fault.

Dead-bench the PSU. Disconnect the PSU from the miner completely. Plug it into AC alone. A healthy APW9/APW12 spins its fan for 1-2 seconds at plug-in, then idles. Put a multimeter on DC across the 12V output pins with no load — expect 12.0 +/- 0.3 V. If dead, or voltage is wildly out of spec, the PSU is your first replacement: CAD $150-$450 depending on model and new-vs-used.

Inspect the PSU input stage (warranty already void). Open the PSU shell and look for a blown MOV on the AC input — black/green disc component, usually cracked or charred after a surge clamp event. Check the primary-side fuse (ceramic cartridge near AC input). A blown fuse alone is sometimes a five-dollar repair, but only if you confirm the rest of the primary side is undamaged.

Isolate the control board first, BEFORE the hashboards. With a known-good PSU, connect it to the control board only — leave all hashboards disconnected. Power on. The control board should boot, negotiate ethernet, and serve its web UI within 90 seconds. If it doesn't, you have a CAD $80-$140 control-board replacement, which is substantially cheaper than assuming the hashboards are dead and shipping the whole miner.

Hashboard-by-hashboard re-introduction. With the control board confirmed good, connect hashboard 0 only, power on, wait 90 seconds, check the dashboard for full chip count (63 on S9, 76 on S19-class). Then power off, repeat for boards 1 and 2. Never introduce multiple suspect hashboards in parallel — you can't isolate the fault, and one bad board can load the PSU enough to hide a second marginal one.

Re-seat every PCIe and data cable. Power off at the breaker. Disconnect the 6-pin PCIe power cables from each hashboard. Inspect the connector housings for melt marks or blackened pins — this is a classic surge artefact. Inspect the data ribbon cables for cracked insulation at the strain-relief point. Replace anything visibly damaged. Listen for the click when reconnecting.

Check line voltage at the panel under load. Once the miner is running again, put a multimeter at the outlet under full mining load. 235-245 V expected on 240 V split-phase; 202-212 V on 208 V commercial. Low line voltage after a surge event often means your service entrance took damage too — if readings are wonky, stop and call a licensed electrician before you lose the replacement miner.

Replace input-stage MOSFETs and TVS diodes on the hashboard. If the hashboard's 12V input domain has a burnt MOSFET or cracked TVS but the ASIC chain is intact, this is a hot-air + solder-iron repair in the CAD $40-$100 parts range. Desolder the damaged part, clean pads with braid + flux, place the replacement with tweezers, reflow. Use a thermal camera afterward under load — the newly-installed part should not be the hottest thing on the board.

Replace bulged electrolytics and cracked MLCCs. Bulged electrolytic caps near the 12V input or voltage-domain ICs are a cheap fix (CAD $5-$15 in parts), but they're a sign the surge stressed the whole input stage. Replace all of them on the affected board, not just the obvious ones — the others will follow within weeks. Cracked MLCCs are harder to spot; a multimeter on capacitance-check mode against a known-good board often reveals open caps that look visually fine.

Flash DCENT_OS for per-chip diagnostics on surviving hashboards. DCENT_OS — D-Central's own open-source Antminer firmware — exposes per-chip HW%, tuning, autotuning, and stratum v2 with no licensing BS. Braiins OS+, LuxOS, and Vnish are equivalent alternatives. Stock Bitmain firmware hides the one number that matters post-surge: which chip position is on its way out. Land at https://d-central.tech/dcent-os/ to flash.

Reflow marginal chips before replacing them. If DCENT_OS shows one or two chip positions with elevated HW% post-surge, reflow before condemning. Remove heatsink, flux the BGA, preheat bottom side to ~150 C, top-side hot air at 310-330 C for ~30 seconds. BM1398/BM1362/BM1368 tolerate a reflow cycle well. Re-apply thermal paste (Arctic MX-6 or Kryonaut), reassemble, re-verify per-chip HW% after 20 minutes of stable hashing.

Replace the control board's SPI flash if it was brown-out-corrupted. On S9/S17 control boards with soldered-on SPI flash, a surge brown-out mid-write can corrupt the boot image. Re-flashing over SPI with a CH341A programmer plus a known-good image restores the board without replacing it. Pull images only from trusted sources. This is a soldering-adjacent skill, not the first tool you reach for — but it can turn a CAD $140 control-board replacement into a CAD $10 repair.

Stop DIY when damage crosses subsystem boundaries. If the surge took out the PSU AND the control board AND multiple hashboards, or if you find burnt components on the ASIC chain itself (soot on BGA pads, cracked chip packages), you're in test-fixture territory. Book a D-Central ASIC Repair slot. Our bench has programmable-load test fixtures, official Bitmain test binaries for per-chip isolation, and salvaged-grade / new-old-stock BM1398/BM1362/BM1368 inventory to replace chips without buying a whole board.

D-Central bench process. Inbound intake photos and log capture; surge-damage specific checklist (MOV / primary fuse / control-board SPI / hashboard input stage / domain-by-domain rail test); component-level repair of whatever survived; chip replacement on anything that didn't; post-repair 24-hour nameplate burn-in before the miner ships back. Turnaround 5-10 business days, Canada / US / international. Expect CAD $150-$700+ depending on subsystem count damaged.

Ship safely. Pack PSU, control board, and hashboards separately in anti-static bags. Double-box with at least 5 cm of foam on every side. Include a note with event timestamp, weather conditions, what tripped (breaker/GFCI/UPS), firmware version, and contact info. Canadian shipments via Canada Post expedited; US via UPS Ground; international via DHL. The note shortens diagnostic time, which shortens your invoice.

Install prevention BEFORE you put the repaired miner back in service. Type-2 SPD at the service entrance (CAD $200-$600 plus licensed electrician labour), per-circuit surge strip with joule rating >= 2000 J, pure-sine-wave UPS (850 VA+) feeding at minimum the network switch and control electronics. The upstream environment that killed the first miner will kill the replacement if untouched.

Log line voltage for 24-48 hours post-repair. A USD $30 outlet logger plugged in for 48 hours catches degrading MOVs, failing UPSes, and utility-side harmonic creep before they eat the repaired miner. If logs show any sag below 220 V on 240 V split-phase or below 195 V on 208 V commercial, escalate to the utility or a licensed electrician before long-term mining operation resumes.