Whatsminer Error 275 – Power Supply Over-Temperature Warning

Confirm 275 is the only fault firing via the BTMiner API. Curl port 4028 with {"cmd":"status"} from a laptop on the same subnet and check the error_codes array. If [275] is alone, you are in the warning window - preventative work, miner still hashing, no rush. If [275, 202] or [275, 233] is present, you have crossed into hard-trip territory and this page is no longer the right page; see the linked critical-fault guides. If you cannot reach the API, pull the LED status with the miner running and proceed as 275-warning if the LED pattern matches the Whatsminer LED reference.

Take an IR thermometer reading at the PSU intake grille. Not room-middle. Not the hallway. At the intake. Target is <=30 C for any sustained-tune mining; <=25 C is comfortable margin. If reading is above 30 C, the warning is at least partially environmental and no amount of filter cleaning will fully fix it without dropping the ambient. Open windows, run an air conditioner, point a box fan at the intake from 60-100 cm away, or relocate the miner to a cooler space. Recheck after the air settles.

Pull the PSU intake filter and inspect. Kill AC at the breaker, wait 60 seconds for capacitor discharge, pull the PSU from its mount (M30S-class has remote PSU; M50S+/M60S+/M66S integrate the PSU into the chassis). The intake-side foam / mesh filter is the choke point. Visibly fouled = grey to brown coloration, more than half blocked. Shop-vac both sides at low suction, OR rinse under cool water and dry completely (24 hours minimum, do not reinstall wet), OR replace if torn or missing. Fresh filter is CAD $5-15. Reinstall, power up, hash for 30 minutes, recheck Code 275.

Listen to the PSU fan(s) under load. Power on the miner, give it 60 seconds to ramp up, put your ear close to the chassis. A healthy fan ramps to a strong, steady whine. Unhealthy patterns: wobble (bearings going), scraping (debris in the rotor), intermittent silence (electrical contact in the fan harness intermittent), or a fan that never ramps past idle (controller failure or fan dead). Wobble or scraping = replace soon. Silence or no ramp = replace immediately, you are minutes-to-hours from Code 202.

Inspect exhaust clearance behind the miner. Use a tape measure: from the rear of the chassis to whatever is behind it (wall, other miner, rack post, curtain) should be >=30 cm. Less than that and the PSU exhaust recirculates back to the intake, raising effective ambient by 5-15 C even in an otherwise cool room. Move the miner forward, the rack post back, or duct the exhaust away. This single move clears Code 275 for many home miners who never realised their setup was thermally cooking itself.

Log PSU telemetry via WhatsMinerTool for 60 minutes under steady-state load. WhatsMinerTool exposes T_psu internal, fan RPM, V_out, I_out, P_out. Hash at full tune, log all five for an hour. Healthy baseline: T_psu rises from ~30 C cold-start to steady 60-70 C within 15 minutes and holds there; fan RPM ramps with temp and stabilises; V_out steady at 12.0-12.3V. Unhealthy: T_psu continues climbing past 75 C (warning threshold), oscillates +/-5 C as MCU plays with fan speed, or trends slowly upward over the full hour (heat-soaking). The telemetry log is also what D-Central will ask for if you eventually ship the unit.

Verify PSU fan RPM against model spec. From the same WhatsMinerTool log: M30S-class P21 fans should be at 5,500-7,500 RPM at full load; M50S+/M60S+ class P221 at 6,500-8,500 RPM; M66S P222 at 7,000-9,000 RPM. Below those bands the fan is rolling off and is the root cause of the warning. If firmware-reported RPM is suspect (stale-reading bugs exist), unplug the PSU fan with the unit OFF, jumper its 12V and GND leads to a known-good 12V bench supply, confirm full-RPM operation. A fan that won't make spec on direct 12V is dead.

Replace a failing PSU fan. A standard MicroBT P21 fan is a 120mm x 38mm 12V server fan rated ~7,500 RPM. Direct OEM replacements are CAD $15-35 from D-Central or any server-parts supplier. Procedure: kill AC, wait 60s, remove PSU, unscrew the fan grille (typically 4 x M3 screws), unplug the fan harness, swap fan, plug new harness in (orient correctly - airflow direction is etched on the fan housing), screw grille back on, reinstall PSU, power up, verify telemetry shows the new fan ramping to spec. 15-30 minutes of careful work.

Improve room ventilation or relocate. If intake ambient is chronically >30 C and you cannot AC the room, the cheap fix is forced air circulation: a $35 box fan pointed at the intake from 60-100 cm away drops effective intake by 3-8 C depending on layout. Better fix is relocation - basement, garage with venting, dedicated mining room with a window fan exhausting hot air. Best fix for Canadian operators is converting to a Bitcoin Space Heater configuration during heating season: the miner runs cool because cold winter air is the heat sink, and the heat output offsets your home heating bill.

Set up duct work or shrouds for restricted spaces. D-Central stocks universal ASIC shrouds and duct adapters for operators in apartments, condos, basements, or rack-mount spaces who can't relocate the miner but need to push the exhaust elsewhere. A CAD $40-100 shroud + 1m of 15 cm flex duct extending the exhaust to a window or another room solves the recirculation problem permanently. Pair with a thermostat-triggered booster fan if the run is >3m.

Back the tune off by one step. If miner is running at maximum tune profile (3.8 kW on M30S++, 4.5 kW on M60S+, 5.4 kW on M66S) and Tiers 1-2 are clean, the cooling environment is at its limit for that tune. Drop tune one step via WhatsMinerTool or the Web UI. Hash for an hour. If Code 275 stops firing, cooling cannot support max tune at this ambient. Two paths: keep the lower tune permanently (lose 10-15% revenue but gain reliability and PSU lifespan), or improve cooling materially (lower ambient, better ducting, fresh thermal compound) and re-push.

Refresh PSU internal thermal compound. For PSUs 4+ years old with chronic Code 275 despite clean filter / healthy fan / cool ambient: open the PSU case (warranty void at this point - confirm warranty status first). Inspect thermal pads and paste between primary-side switching FETs, secondary-side rectifier diodes, toroidal inductors, and their heatsinks. Failures: cracked pads, paste pumped out leaving voids, paste hardened to chalky consistency, missing pads. Procedure: clean old compound with 99% IPA + lint-free swabs, replace pads with matching-thickness silicone (1mm / 1.5mm / 2mm Arctic or Thermal Grizzly as appropriate), replace paste between FETs and heatsinks with Arctic MX-6. Reassemble carefully.

Inspect and replace PSU fan harness. A failing fan harness - corroded contacts, cracked insulation, intermittent break in a conductor - presents identically to a failing fan: low RPM, irregular telemetry, occasional silence. Unplug the harness at both ends, inspect under bright light and 10x magnification at the connector pins. Greenish corrosion on pins, blackened plastic at connector body, or visible cracks in the harness jacket = replace harness. Genuine replacements from MicroBT or aftermarket on Amazon CAD $10-25. If you can't source a direct replacement, a 4-pin PWM 12V fan extension cable (server / PC parts) usually works - confirm pinout before splicing.

Audit the PSU fan controller circuit (advanced). On a small percentage of cases the fan and harness are healthy but the PSU board's fan-control PWM circuit has degraded, commanding low duty-cycle even when temp warrants high. Symptom: T_psu rising past 75 C while fan RPM stays low and the PWM signal at the fan harness reads <50% duty cycle on a scope. Fix is component-level rework on the PSU board - replace a specific PWM controller IC or a degraded gate driver. Skill gate: hot-air SMD rework, datasheet reading, scope. If you don't have all three, ship to D-Central or replace the PSU.

Map ambient to internal-temp delta over 24 hours for chronic cases. For miners that fire Code 275 only at certain times of day or under certain conditions, log both intake-ambient and T_psu over 24-48 hours. The slope (delta-T_psu / delta-ambient) tells you the cooling system's efficiency - healthy P21 = ~1.0-1.2, meaning T_psu rises ~1.1 C for every 1 C of ambient rise. Slopes of 1.5+ indicate a marginal cooling system that will fire 275 whenever ambient peaks. Identifies whether you have an environmental problem or a hardware-degradation problem before you spend money on the wrong fix.

Implement automated 275-trigger alerting. Wire the btminer-api JSON polling into a lightweight Prometheus / InfluxDB / Grafana stack (community projects exist on GitHub). Alert the moment 275 appears in the error_codes array, before it has a chance to escalate to 202 / 233. Catching 275 in the warning band = 5 minutes of filter / fan work; catching it after the hard-trip cascade = 1-4 hours of downtime + potentially CAD $80-480 of bench repair. The monitoring is a one-time setup that pays dividends across the entire mining fleet.

Stop DIY when chronic Code 275 persists across all of Tiers 1-3. A fully-cleaned filter, a verified healthy fan, refreshed thermal compound, cool ambient, healthy exhaust path, and the warning still recurs means the PSU board's silicon is degraded - usually output-side MOSFETs whose R_DS(on) has drifted up over years of operation, dissipating more heat at the same load. This is bench-fixture territory. D-Central's process: pull the PSU, test against a programmable DC load at 50% / 100% / 0% cycles for 2 hours while logging T_psu and load efficiency. A drifted MOSFET shows up as elevated T_psu for measured power dissipation. CAD $80-180 typical, vs $220-540 for a full PSU replacement.

Stop DIY if the PSU is 5+ years old AND has a prior repair history. Silicon ages on a coherent curve. A 5+ year PSU with one fixed component has the rest of its silicon on the same wear trajectory; fixing the next failure buys months, not years. D-Central will quote repair vs replace with a photo-documented diagnostic so you can choose. For miners <3 years old, repair almost always wins; for >5 years, replacement usually does. The sweet spot for thermal-compound-refresh + fan-replace bench job is the 3-5 year band where silicon is healthy but consumables (paste, pads, fan bearings) are due.

Ship with full context. Pack the PSU (and chassis if you suspect downstream / harness issues - bench reproduction often requires the full stack), a copy of your last btminer-api status JSON, the WhatsMinerTool PSU telemetry log if captured, intake-ambient logs, service history (prior repairs, tune changes, recent moves). Match chassis serial to PSU serial in the ship note - they're on metal plates, don't guess. Canada-wide standard shipping, US / international welcomed. Turnaround 3-7 business days for isolated Code 275 repairs, longer if multiple faults surface on the bench.