Whatsminer M50S+ – Voltage Domain Overcurrent Trip

Hard power-cycle at the breaker for 60 seconds, then reboot with stock OC profile via WhatsMinerTool. Aggressive OC profiles on aged M50S+ boards are the single most common trigger of WM_M50SP_OCP. About 25-35% of cases on overclocked rigs clear here, with the long-term answer being to rebuild the OC slower. Monitor hashrate after — if stock hashrate is acceptable you may be done; if it's well below nameplate the chip drift has progressed and Tier 3 is needed.

Verify intake air temperature with an IR thermometer at the front grille. M50S+ wants intake ≤ 30 °C and chip temp ≤ 90 °C. Hot intake compresses margin to OCP. Move the rig to better airflow, clean the intake filter, blow out heatsink fins, and re-test. Common Canadian summer pattern: garage rig that ran fine all winter throws OCP in July when ambient hits 35 °C.

Audit wall voltage with a plug-in voltage meter or multimeter. Whatsminer PSUs feeding M50S+ want stable 220-240 V. Sagging circuit during steady-state hashing causes PSU compensation that can present as domain instability resembling OCP trip. If voltage drifts ±10 V or sags below 215 V under load, address the electrical feed before continuing.

Restart via WhatsMinerTool soft restart, not chassis power button. Soft restart cycles BTMiner without bouncing PSU. Diagnostic value: trip clears on soft restart and runs hours then re-throws = BTMiner state or OC config part of the problem. Trip returns immediately on soft restart = cause is hardware on the load side.

Check BTMiner firmware version against latest stable for M50S+ at the WhatsMinerTool firmware repository. Some 2023.x BTMiner builds shipped with overly aggressive domain probing that elevated nuisance OCP trips on aged boards. Roll forward or back one stable version and observe — never cross-flash from a different M-series model image.

Pull per-board and per-domain current via the BTMiner API. Query TCP 4028 with cmd:get_miner_status and cmd:get_psu. Record values, repeat every 30 seconds for 5 minutes. Healthy M50S+ pulls ~85-95 A per board distributed evenly across domains. A board climbing past peers is the suspect. A domain climbing past peers within a board is the smoking gun — that domain contains the failing component.

Power off, breaker off, 2-minute discharge, then re-seat all hashboard data and power harnesses. Inspect each connector for blackening, bent pins, oxidation. Re-seat firmly, listen for the click. A noisy harness can cause BTMiner's domain telemetry to misread, commanding a defensive shutdown that mimics an OCP event. Free fix; clears a small but real percentage of cases.

Hashboard slot swap. With the chassis open and labelled, move the suspect hashboard to a known-good slot (board from SM0 to SM2, etc.). Boot, observe whether OCP follows the board or stays in the slot. Board-following = bad board, isolate further. Slot-following = control board, harness, or power-distribution to that slot is the cause.

Rebuild OC profile slowly from stock. Baseline at stock voltage and frequency for 30 minutes — no trip is the goal. Add 25 MHz at a time, watch 15 minutes between steps, stop at the step before OCP returns. That's the new silicon-lottery ceiling for this rig — on aged boards lower than the original. No two M50S+ boards drift identically; per-rig calibration, not a forum profile to copy.

Measure per-board DC current under load with a clamp DC ammeter (Fluke 376 FC or similar) around each board's positive supply lead while hashing at full load. Compare live measurement to BTMiner API report — drift greater than 10% between actual and reported indicates PSU sensor degradation, not domain OCP, and diagnosis pivots to Whatsminer Error 205 / 268 territory.

Audit PSU output voltage at the connector under load. Multimeter on DC, probes at the PSU-to-board connector with rig hashing. Expect ~14.5-15 V steady. Sag below 14 V under load = PSU is tired or bus-bar contact is degraded; fix the supply side before chasing the domain trip.

Verify chassis fan health and intake clear path. With chassis open, confirm intake and exhaust fans both rotate at full RPM during hashing. Confirm heatsink fins not packed with dust. Apply compressed air or soft brush. Verify no obstruction in front of the front grille within 15 cm. Thermal-margin restoration alone clears a meaningful fraction of OCP trips on dusty rigs.

Thermal scan the suspect board under load with a FLIR ONE Pro or equivalent. Run the rig long enough to approach the trip event (don't let it actually trip — pull the load before OCP fires if you can predict the timing). Hot chip is the failing chip — expect 10-25 °C above peer chips on the same chain. Mark the position. Also scan buck-stage MOSFETs — a hot MOSFET indicates a failed PWM phase, distinct from chip drift.

Reflow the worst chip. Remove heatsink, clean chip and heatsink contact face with isopropyl alcohol 99%, apply flux to the BGA perimeter, reflow with preheat-plus-hot-air (preheat bottom 150 °C, top-side hot air 310-330 °C for 30 seconds). Cool naturally. Re-apply Arctic MX-6 or Thermal Grizzly Kryonaut, reseat heatsink, reassemble, burn in 24 hours at stock OC. Reflow holds 6-18 months on a drift-aged chip; if it doesn't hold for 30 days the chip itself needs replacement.

Replace any bulged electrolytic caps or cracked MLCCs in the affected domain. Soldering-iron + hot-air rework, not reflow. Replacement MLCCs are pennies; electrolytics on M50S+ buck stage are typically 220 µF / 16-25 V low-ESR polymer caps, $1-3 each. Match capacitance, voltage rating (or higher), and ESR class — a wrong cap creates new domain instability. Only attempt if you can read schematics; otherwise bench-shop territory.

Replace a failed PWM phase MOSFET pair if FLIR identified it in step 13. Identify the high-side / low-side MOSFET pair on the failed phase, desolder with hot air, replace with matched-spec parts (M50S+ buck stage uses common DrMOS packages — part number silkscreened or in datasheet for that PCB revision). Intermediate-advanced rework; most home miners stop here and ship to D-Central. Document with photos, work clean (ESD mat, wrist strap), burn in 24 hours before returning to OC tuning.

Stop DIY and ship to D-Central when any of these are true: (a) two or more drifted chips identified on the same board via FLIR, (b) reflow on one chip doesn't hold 30 days, (c) failed component is the buck-stage IC itself, (d) trip pattern follows the slot rather than the board (control board suspected), (e) capacitor bulging or burnt-component smell on inspection, (f) the rig is one of multiple in a fleet showing same drift pattern at similar age, or (g) your bench skills don't include hot-air rework, BGA reflow, or fine SMD soldering.

Ship safely. Power off, breaker off, full discharge. Remove each hashboard, bag individually in ESD-safe anti-static, separate with foam. Box separately from chassis if shipping boards only. Include a written note: BTMiner firmware version, observed trip pattern (cold-boot vs thermal vs OC-only), every diagnostic step already tried, per-board current readings from the API, and any FLIR scan photos. Pre-diagnosed boards quote materially lower because we don't redo the work you've already done.