Whatsminer M50S – Low Hashrate

Log a 15-minute baseline in WhatsminerTool (realized hashrate, each chain, temp, fan RPM, PSU input / output). Screenshot or CSV it.

Export logs and grep for the code set (`2340`, `2350`, `236`, `238`, `268`, `430`, `520`, `540`–`542`, `6xx`, `5110`–`5112`, `100001`–`100003`).

Check Miner Info → Diagnosis for chip count below `156` per chain.

Hard power-cycle the miner for 60 seconds at the breaker.

Shop-vac the intake filter and grille; verify no recirculation; confirm intake ≤ `30 °C` at the grille.

True-RMS meter on AC at the PSU C19 inlet under full load. Target `230–245 V` on 240 V, `202–212 V` on 208 V.

DC-probe the PSU-to-hashboard connector under full load. Target `14.6–15.0 V` sustained on M50S-class PSUs.

Cross-check PSU-reported input voltage vs meter. Drift `>5%` = swap PSU.

Re-seat every 40-pin ribbon cable at both ends. Re-seat the hashboard DC connectors. Inspect for blackening, bent pins.

Swap the suspect hashboard to a known-good slot. Observe 30 minutes. Follows board = board fault; stays in slot = control path.

Flash the current MicroBT-signed stable for your hardware revision via WhatsminerTool. Do not attempt DCENT_OS or Antminer firmware. If on custom firmware with `100001`–`100003` codes, revert to signed stock.

Refresh thermal paste on all hashboards. Arctic MX-6 or Thermal Grizzly Kryonaut. Thin uniform layer. Inspect PMIC / voltage-domain thermal pads for cracks or hardening — replace if in doubt.

Reflow the worst-performing chip (once an `upfreq_test.log` position is isolated). Preheat bottom side to `~150 °C`, top-side hot air at `310–330 °C` for `~30 s`. `BM1398`-class BGAs tolerate a reflow cycle well.

Replace a bad chip if reflow doesn't hold. Source a bin-matched replacement — mismatched bins drag the whole chain down (`430`/`520` territory). Bin compatibility tables are not published by MicroBT; confirm against D-Central's internal decoder before replacing.

Replace the 40-pin ribbon cable and / or the adapter plate if re-seat fixes the problem but it returns within 90 days — the ribbon is worn.

Stop DIY. Triggers: chip-index failure reproduces on two different boards (PCB-level); visible heat damage, bulged caps, or burnt smell; signature / integrity damage that won't clear after rollback; miner dropped, submerged, or struck by lightning. Book a D-Central ASIC Repair slot — turnaround 5–10 business days Canada-wide.

Bench process at D-Central: test fixture with programmable load, chip-level isolation via WhatsminerAPI (`TCP 4028`) scripted sweep, bin-matched chip replacement from D-Central's graded inventory, reflow + reseal, PSU sensor verification and rotation if drift is suspected, 24-hour nameplate burn-in post-repair.

Ship safely. Hashboards in anti-static bags. PSU wrapped separately. Double-box with `≥ 5 cm` foam on every side. Include a printed note with observed symptoms, firmware version, log snippets (highlight any codes), electrical context (`120 V`? `240 V`? `208 V`? what else is on the circuit?), and contact info. Saves us diagnostic time, saves you bench-hour charges. Key rules of the road: - Do not reflash DCENT_OS on a Whatsminer. DCENT_OS is D-Central's open-source Antminer firmware — it runs on Bitmain-class silicon and control boards. Flashing it to M50S-class hardware will brick the control board and is a common cause of "paperweight" customer tickets. - Rule out power before you touch silicon. The biggest avoidable cost we see is operators ordering hashboard repairs for what was a sagging `12 AWG` extension cord. - Baseline before changing anything. A 15-minute Status log saves you from chasing ghosts after you've already swapped three things.