Whatsminer P221B Hex Fault Code Decoder (0x0040–0x2000)

SSH into the miner and run `btminer status`, or pull the JSON from the local API endpoint, to capture the exact `psu_fault` value. Convert decimal to hex if the firmware returns decimal (`python3 -c "print(hex(2112))"` returns `0x0840`). Note **every** bit asserted, not just the lowest-numbered one — the `psu_fault` field is a 16-bit OR. A miner showing `0x0840` is asserting `0x0800` (fan failure) AND `0x0040` (output OCP) simultaneously — two distinct fault families that need to be addressed in upstream-first order.

Power down completely at the breaker. Wait 60 seconds for the bulk-cap bleed cycle. Open the miner chassis per the model's service procedure (do not open the `P221B` itself yet — Tier 1 is busbar work between the PSU and the hashboards). Photograph every connection before disturbing anything so you can reassemble correctly. Inspect every M5 or M6 busbar bolt for: spring washer present and seated, bolt head not heat-discolored, copper not oxidized, no arc-tracking. Heat-discolored bolts (purple/blue oxide) have been carrying current on a marginal contact for weeks — replace, don't just retorque.

Clean any oxidized contacts. Green or black film on the copper means oxidation; current concentrates on the remaining clean area, contact resistance climbs, the spot heats further. Use a Scotch-Brite pad or a flat fiber-pen to polish the contact face, wipe with `99% isopropyl alcohol`, reassemble dry. Do not use sandpaper — it leaves abrasive grit in the contact face that traps moisture and accelerates re-oxidation.

Torque each busbar bolt to spec with a real torque wrench. MicroBT's air-cooling chassis spec is typically `4–6 Nm` on M5 busbar bolts (verify against the model's service manual — M50S/M60S can differ slightly from M30S+ revisions). 'As tight as I can get it with a 6-inch ratchet' is not a torque spec — it both undertorques on the awkward contacts and overtorques on the corner ones. Reassemble the chassis, reboot, run a 30-minute load test, recapture the `psu_fault` value. Most `0x0040`/`0x0100`/`0x0200`/`0x0400`/`0x1000`/`0x2000` clears land here.

If `0x0001` (input voltage low) is asserted, multimeter at the C19 inlet under load. Healthy `P221B` operating range is `220–240 V AC`. Any reading below `200 V` triggers `0x0001` and the PSU will refuse to boost properly. Common causes: shared-circuit voltage sag (neighbour's HVAC kicks on), undersized panel feeder, long run on undersized wire (a 30-meter run on 14 AWG to a `P221B` will sag visibly under 22 A draw), brownout in progress. Fix the electrical infrastructure — move to a dedicated 240 V circuit on its own breaker, ideally `30 A` on `10 AWG`.

If thermal bits (`0x0002`, `0x0004`, `0x0008`) are asserted, measure intake air temp at the PSU front grille with an IR thermometer. Healthy operating envelope: ambient at the intake `≤ 35 °C`. Above that and the internal sensors will trip even with a healthy PSU. Clean the intake filter, blow out the heatsink fins (compressed air, pulse — never let fan blades free-spin, back-EMF can damage the fan controller), verify the PSU fan is actually running (the hashboard fans are not the PSU fan). Quarterly intake clean is the right preventive cadence.

Verify the PSU fan is healthy. With the miner running, the PSU fan should be visually spinning, not making bearing whine, not vibrating against the chassis. If you can feel it scraping or hear it ticking: bearing wear, replacement scheduled. Don't wait for `0x0800` to assert — by the time the housekeeping micro reads zero RPM, the fan has been degrading for weeks and other components have been running hotter than spec the whole time.

Measure output rail at the busbar with the miner under full load. Multimeter on DC, probe at the busbar bolt or downstream of the hashboard input. Healthy reading = the negotiated setpoint, typically `13.0–14.5 V` on M50/M60 air-cooling profiles depending on the power profile in use. Outside the `11.5–15.5 V` P221B range = output regulation has drifted, the PSU is heading for trouble even if no bit is asserted yet. Schedule bench diagnosis.

Inspect for swollen or vented secondary capacitors through the vent grille (visual only, no opening). Bulging top or vented top on a top-side electrolytic = aged cap, replacement scheduled. This is a planning indicator, not an emergency fix. A `P221B` more than 4 years old is in cap-replacement territory whether you're seeing faults yet or not — the failure is gradual, not sudden, and a planned re-cap costs roughly `$25 CAD` in parts plus a bench-hour and adds 2–3 years of reliable life.

Power-cycle the miner with a full 10-minute discharge to clear any latched protect state (relevant for `0x0010`, `0x0080` primary-side bits that may have tripped on a transient). Cut breaker, wait 10 minutes, restart. Some primary-side latches clear with the full discharge cycle; if the bit reasserts within 60 seconds of restart, the failure is real and the PSU needs the bench.

Audit your miner's power profile against the `P221B` nameplate. The PSU is rated `265 A` continuous on the main rail. M50S++ at high power (~3850 W input) can pull `260+ A` on the busbar — at the edge of the envelope. M60S at peak draw is similar. If you're running an aggressive overclock and seeing `0x2000` (`295 A` for 10 min asserted), you are exceeding the PSU's continuous spec — pull the profile back to ~85% of nameplate. The longer-term hashrate-per-PSU-life math favours the conservative profile.

Tier 3 — discharge the bulk capacitor before opening the PSU itself. The `P221B` holds approximately `380 V DC` on the primary bulk cap after a normal power-down. Place a `10 kΩ 5W` resistor across the cap terminals for 30 seconds, then verify with a DMM that voltage is `< 10 V` before proceeding. Wear safety glasses. The `P221B` is a self-starting brick — there is no soft PS_ON signal; the cap charges the moment AC is applied.

Diode-test the bridge rectifier and PFC MOSFET on the primary side. Bridge across the AC pins both directions: ~`0.5 V` drop one way, open the other = healthy. Short either way = bridge replacement. PFC MOSFET drain-to-source: open both directions = healthy. Short = MOSFET replacement (and the AC fuse usually went with it — replace both). MicroBT does not publish reference designators for the `P221B` primary side; cross-reference with a board-level photo before sourcing parts. This step requires component-level skills and an isolation transformer for safe re-power.

Inspect the secondary rectifier stage. The output-OCP bits trip on the secondary side; a failing secondary rectifier diode can cause one phase to source more than its share, asserting `0x0100`/`0x0200`/`0x0400`. Visual inspection for discolored solder joints, lifted pads, or scorched silicon. Bench replacement only — not a kitchen-table repair. If you see secondary-side damage, this is ship-to-D-Central territory.

Plan a re-cap on units older than 5 years even if no fault is yet asserted. Use 105 °C-rated low-ESR caps (Nichicon HE / Rubycon ZLH or equivalent). Match capacitance and voltage exactly. Observe polarity. This is a one-hour bench job for a competent tech and adds 2–3 years to the unit's reliable life. Total parts cost typically `$25 CAD` at distributor pricing for a full secondary-side re-cap of a `P221B`.

Stop DIY and ship to D-Central if any of: multiple primary-side components shorted simultaneously, visible PCB scorching or lifted traces, internal fan dead and you don't want to open mains-side internals, no isolation transformer / current-limited bulb tester / scope, unit older than 6 years with cap bulge across multiple secondary positions, or the same fault bit reasserting on a swapped-in known-good `P221B` within minutes. Book at https://d-central.tech/services/asic-repair/ — flat-rate evaluation, parts at cost, return shipping included for North American customers.

When shipping to D-Central, package safely. Anti-static bag the PSU, double-box with at least `5 cm` of foam on every side. Include a note with: chassis model, observed `psu_fault` hex value (every bit you've seen asserted), BTMiner firmware version, mains voltage at the inlet under load, ambient temperature at the intake, and what diagnostic steps you've already tried. This information saves diagnostic time on the bench and saves you money on the final repair quote. Photograph the unit before shipping for your own records.

Decide repair vs replace based on cost math. If parts plus labor exceeds 60% of replacement cost (`~$200 CAD` for a tested second-hand `P221B`, `$300+ CAD` for new), replace. D-Central stocks tested replacement units when available; new pricing varies by region and supply. Beyond 6 years of age, default to replacement unless the unit has sentimental value or the chassis has a custom modification that makes the original PSU non-trivial to re-source.