Whatsminer Error 272 – Excessive Power Output Warning

DMM the 12V busbar before believing the dashboard. Calibrated multimeter, DC volts mode, probe at the control-box input under full hashing load. Read for 60 seconds. Healthy: 12.0-12.3V mean. Warning-zone: >12.5V sustained. If your meter reads 12.1V while the dashboard reports 12.7V, the sense path is lying - completely different repair than a real overvoltage. This 30-second test sorts the diagnostic into the right tier before you spend any money. Anyone with a CAD $40 Klein DMM can do this; you do not need a Fluke for the first pass.

btminer-api query the full error_codes array. Curl port 4028 with {"cmd":"status"} and read what is actually flagged. [272] alone = warning only, you have weeks. [272, 235] or [272, 233] = warning AND hard trip have both fired = days, not weeks, before the miner stops hashing. The Web UI sometimes collapses paired errors and only displays Code 272; the API tells the truth. This single command gives you the urgency calibration for the rest of the work.

Power-cycle at AC mains for 60 seconds. Kills any stale latched fault flags. About ~5% of Code 272 events are stale software state from a transient overshoot the PSU has since recovered from. Wait the full minute for primary-side bulk caps and secondary-side output caps to discharge - a chassis-button reboot leaves the PSU MCU's fault register set on some firmware revisions. If Code 272 clears on AC cycle and stays clear for 24h+ of continuous hashing under full load, you had a stale flag. If it returns within hours, the fault is real and you are moving down the tiers.

Check service-history context. Was the firmware updated in the last 7 days? Was the tune profile changed? Was the control board reset, EEPROM cleared, or a WhatsMinerTool configuration push applied? Recent firmware / config events point at categories 2 (corrupted calibration) or 3 (wrong PWM command) - both are software fixes. No recent events = the warning is most likely category 1 (silicon drift) and you are heading toward PSU replacement. Documenting this saves an hour of wrong-direction diagnostics.

Back off the tune by one step and observe. If the warning correlates with high-tune profiles (3.8 kW on M30S++, 4.5 kW+ on M50S++, 5.4 kW on M60S+), revert to the previous step via WhatsMinerTool or the Web UI. If Code 272 clears at the lower tune, the PSU's regulation margin is gone at the higher load - confirming a category-1 ageing PSU near end of life. Run at the lower tune to extend remaining PSU life while you order a replacement; revenue loss from 10-15% lower tune is dramatically less than the cost of a Code 235 cascade taking a hashboard cap with it.

Log V_out telemetry for 30-60 minutes under load via WhatsMinerTool. The desktop app exposes the live PSU telemetry the Web UI hides. With the miner running at tune, log V_out, I_out, P_out, and PSU-internal temperature. Healthy baseline: V_out steady at 12.0-12.3V, ripple <0.1V, no excursions. Code 272 precursor: V_out creeping above 12.4V over the log window, occasional spikes to 12.6V+ during load steps, or progressive climb across the 60-minute capture. Save the log - it becomes documented fault evidence for warranty / RMA / repair-vs-replace decision and the baseline for comparing the replacement PSU after install.

Re-flash the same firmware via WhatsMinerTool / SD recovery. This is the cheapest fix for category 2 (corrupted calibration blob). Via WhatsMinerTool: Configure -> Firmware Update -> choose your model's current stable build -> flash. Via SD card: write the recovery image with Etcher or Rufus, insert into the control board, hold the recovery button while powering on, wait 5-10 minutes for re-flash. Boot, run 1 hour at full tune, re-check Code 272. VERIFY firmware image matches your exact model and sub-revision - cross-flashing wrong-model firmware bricks the control board because MicroBT version-checks on the CB identity EEPROM.

Roll back one stable firmware version. If category 2 corruption keeps coming back after re-flash, the current firmware build itself has a calibration-write regression. Check the WhatsMiner support portal for prior stable releases for your model, download the previous one, flash it, observe. If Code 272 clears on the older firmware, stay there until MicroBT issues a fix on the newer build. Document the version transition for your service log - and report it on the WhatsMiner forum so the next operator sees the pattern.

Factory-reset the control board to clear stuck PWM commands. If categories 1 and 2 are ruled out (PSU is young, firmware is fresh and re-flashed), category 3 (wrong PWM command holding regulation high) is next. Hold the reset button on the control board for 10 seconds per the WhatsMiner factory reset procedure - wipes the EEPROM tune-profile state, the saved network config, and any control-board adjustment commands cached in flash. Re-apply your intended tune profile from clean firmware. If Code 272 clears, the tune-profile EEPROM was holding a stale command sequence.

Cross-reference with peer miners. If you operate a fleet of multiple Whatsminers, check whether Code 272 is firing on others - same model, same firmware, same age cohort. Multiple miners flagging simultaneously after a firmware update = the firmware is the cause (categories 2 / 3). Single miner flagging in a healthy fleet = the PSU is the cause (category 1 / 4 / 5). This single comparison takes 5 minutes and pins the failure category before any rework or replacement.

Scope the 12V rail steady-state and load-step response. Bench-grade oscilloscope (100 MHz bandwidth, 1 GS/s, voltage probe rated >50V), probe across the 12V busbar at the control-box input. Steady-state capture: confirm 12.0-12.3V DC, <100 mV switching ripple. Load-step capture: trigger on a hashboard ramp event (start/stop via WhatsMinerTool or tune toggle), measure overshoot and settling time. Healthy: <300 mV overshoot, <5 ms settle. Marginal: >500 mV overshoot, >50 ms settle, ringing visible. Marginal regulation loop = PSU at end of life regardless of steady-state reading. This is the highest-fidelity diagnostic for distinguishing 'drifted but stable' from 'drifted and unstable' - the latter is <30 days from Code 235.

Boot into BTMiner recovery / SD card mode for low-level firmware re-flash. If categories 2 / 3 persist through Tier 2 fixes, the control-board flash itself may be wearing out. SD-card recovery flash forces a full reflash including bootloader, factory data, and calibration tables. Per the WhatsMiner bootloader recovery procedure, write the recovery image, insert SD, hold recovery button while powering, wait 10 minutes. After recovery, re-apply your tune profile, observe Code 272. If recovery fixes a long-running fault that re-flash did not, the control-board flash had wear-induced bit errors - a control-board replacement is the next step if the warning returns.

Replace the secondary-side feedback divider on the PSU board (skill gate). If Tier 2 + sense-path check identified a drifted divider, the fix is component-level rework. Remove the PSU from chassis, identify the silkscreened divider resistors (usually two 0.1% SMD resistors near the secondary-side reference IC, typical part numbers R-FB1/R-FB2 or model-specific equivalents). Hot-air remove and replace with 0.1% precision parts of the same value. Re-install, bench-test, then chassis install and verify V_out. Requires SMD rework skill, magnification, hot-air station. If you do not have these, ship to D-Central - divider-replacement bench rework is CAD $80-150.

Replace the regulation opto-isolator (skill gate). Similar to step 13 but for the PC817-class opto crossing the primary-secondary isolation. Identify on the PSU board (typically labelled U-OPT or model-equivalent, near the transformer with traces crossing the isolation slot), hot-air remove, replace with same-spec part. CTR target >0.5 for a fresh opto. Install, bench-test, chassis install, observe V_out. Same skill / tool gate as step 13. If you replace one without the other and the second component is also drifted, you will be back inside 30 days - D-Central's bench process replaces both as a paired job because their failure modes correlate.

Replace the secondary-side reference IC (skill gate). If categories 1 + drift are confirmed and steps 13 / 14 did not fully restore regulation (or if your scope captures show the regulation target itself shifted, not just the loop response), the bandgap / TL431-class reference IC has drifted past spec. Identify on PSU board, hot-air remove, replace with same-spec part. Trickier than divider or opto because the reference is often in a small SOT-23 / SOT-89 package with critical thermal coupling to surrounding components. D-Central's bench rework typically replaces all three (divider, opto, reference) as a paired job because the failure modes correlate and partial replacement extends life by months only.

Replace output bulk capacitors (Tier 3 if not at end-of-life PSU). If scope captures show elevated ripple (>200 mV peak-to-peak at switching frequency) but steady-state is OK, the secondary-side bulk electrolytic caps (typically 2-4 x 470 uF / 25V or model-equivalent) have lost ESR. Hot-air remove and replace with same-spec or upgrade to low-ESR equivalent. Improves ripple, may marginally improve regulation. Cost-effective only on PSUs <3 years old; 4+ year PSUs are best replaced wholesale because the rest of the silicon is on the same ageing curve.

Inspect hashboard buck regulators and decoupling caps for downstream damage. Even at Tier-3 fix territory, Code 272 warning that has been operating for weeks-months may have already inflicted damage on the hashboard buck regulators (12V -> 0.4V chip rail) and the decoupling caps adjacent to the BM chips. Pull each hashboard, inspect under 10x-20x magnification: bulged caps, blackened buck packages, lifted pads near the chip rail. If found, the hashboard needs component-level repair regardless of whether the PSU fix restores V_out cleanly. This is the cascade-damage path that makes early Code 272 action so much cheaper than late.

Stop DIY when the PSU is 4+ years old AND Code 272 is firing. End-of-life PSU silicon ages on a coherent curve - even if you successfully replace one drifted component, the rest of the silicon is on the same ageing trajectory. You will be back inside 6 months. At this stage, replacement is dramatically cheaper than chasing repeat repairs. D-Central stocks graded-salvage and new MicroBT P21 / P221 / P222 PSUs at ~60-100% of new pricing; book a slot for chassis-aware install (matched to your specific control-board firmware revision and tune profile) so the warning stays cleared on first boot.

Stop DIY when scope captures show marginal regulation loop AND you do not have SMD rework capability. A regulation loop running at the edge of stability is <30 days from Code 235 and possible hashboard cap damage. Iron-only rework on a P21 / P221 / P222 PSU board lifts pads and makes the unit unrepairable - D-Central has seen multiple chassis arrive with iron-damage from owners who tried home-repair and made the bench-rework path impossible. Hot-air + preheater is the minimum competent toolset for this rework; without it, ship.

Stop DIY when you see hashboard cascade damage (bulged caps, blackened buck regulators, lifted pads near the chip rail). Cascade damage from prolonged >12.5V operation requires full-chain inspection - replacing only the visibly-broken cap leaves the adjacent damaged-but-not-yet-failed components ready to fail next. D-Central's bench process inspects the full 12V -> 0.4V regulation chain under magnification, replaces all damaged components in a single rework pass, and burn-tests before return. Iron-only home rework on a damaged hashboard makes things worse ~80% of the time.

Ship with full context. Pack the chassis with the suspect PSU (we need your exact stack to reproduce the fault - cross-stack substitution will miss marginal control-board PWM interactions), a copy of your last btminer-api status JSON capture, the WhatsMinerTool PSU telemetry log (especially any time-series V_out capture showing the drift), service history (recent firmware updates, tune changes, prior 272 / 235 / 233 events), and any scope captures if you got them. Match chassis serial to PSU serial in your ship note - both are on metal plates, do not guess. Canada-wide standard shipping; US / international welcomed.

Discuss repair-vs-replace economics for Code 272 specifically. A new MicroBT P21 runs CAD $220-320, a P221 $340-420, a P222 $420-540; graded-salvage units run ~60% of new. A bench-level paired-component repair (divider + opto + reference + caps) runs CAD $120-200 depending on PSU model and component condition. For Code 272 warning caught early on a 2-3 year PSU, repair is usually cost-effective - you get 2-3 more years of clean regulation. For 4+ year PSUs already throwing 272, replacement wins because the rest of the silicon will fail next. D-Central quotes up front with photo-documented diagnostic so you make the call on your specific unit. For hashboard cascade damage caught alongside 272, the hashboard repair is the bigger line item - CAD $100-250 per board for component-level rework, dramatically cheaper than a $400-900 replacement.