Whatsminer P21E Unstable Output Voltage and Ripple

Drop miner power profile. Set the M30S+ to a reduced power mode (~80% of nameplate, or `Low Power` in MinerTool). Lower current draw means less ripple from a tired PSU. Run for 24 hours. Note hashrate, `HW%`, and pool reject rate. Often buys months of clean operation while you plan the proper repair, and confirms whether ripple is the bottleneck without spending a dollar.

Improve PSU airflow. P21E units jammed against the miner chassis or in a closed cabinet recirculate exhaust heat directly into the secondary filter caps. Pull the PSU `30 cm` clear, point a `120 mm` PC fan across the chassis vents, and verify clearance on every side and `10 cm` above. Run the rig for `90 minutes` and re-test. Free intervention; meaningful margin recovery on marginal units.

Drop ambient at the PSU intake. Move the rig to a cooler room, duct cold-side intake air directly to the PSU, or open a basement window in winter. Each `5 °C` of ambient reduction roughly doubles the failure-window margin on a tired electrolytic cap (Arrhenius rule).

Vacuum and inspect the PSU intake/exhaust grilles. Five years of dust on a continuous-duty PSU has materially clogged the airflow. Shop-vac the grilles, blow compressed air through the chassis, verify the fan spins freely with no grinding or wobble. Mandatory before spending any money on parts.

Verify line voltage at the wall under load. `200-240 V` input is what `P21E` prefers; tired P21E units fail far more aggressively on `110-130 V` where input current is roughly double and primary-side stress is materially higher. If your panel reads `108 V` under load on a `120 V` circuit, re-route to a `240 V` circuit if available.

Multimeter on DC output under load. Probe at the PSU-to-miner connector while the miner is at full power. Expect `12.0-12.5 V` sustained on `P21E`. DC sag below `11.8 V` indicates a primary-side issue (bulk caps or FET) — go directly to Tier 3 cap inspection. If DC is stable but you suspect ripple, continue to step 7.

Scope on the output rail (AC-coupled) under load. Use a `10×` probe, AC coupling on the scope, `20 MHz` bandwidth limit on, probe at the load connector. Run the miner full-load for `10 minutes` to thermally stabilize, then capture peak-to-peak ripple. Healthy `P21E`: `<60 mVpp`. Marginal: `60-100 mVpp`. Failing: `>100 mVpp`. This is the single most diagnostic measurement for `P21E_RIPPLE`.

Load-step the PSU. Drop the miner to `75%` nameplate, hold for `10 minutes`, re-measure ripple. Drop to `50%`, repeat. Plot ripple vs load. Linear scaling from low to high load indicates filter-cap aging (Tier 3 re-cap). Non-linear (ripple jumps disproportionately at high load) indicates primary-side bulk-cap death — different repair path, but still Tier 3.

Inspect output cable and connectors. With everything powered off, unplug, and visually inspect both ends of the PSU output harness. Look for arcing or pitting on contact pins, melted plastic, oxidation. Loose or oxidized connectors cause local heating and false ripple readings at the load. Clean with isopropyl alcohol, reseat firmly, listen for the click.

Tighten PSU output terminal screws. P21E uses a screw-down output terminal for the high-current `12 V` rail. Five years of thermal cycling can loosen these. With everything powered off and bled down, torque to MicroBT spec (typically `0.8-1.0 Nm`). Loose terminals = local resistive heating = false-positive ripple readings at the load connector.

Open the PSU and discharge bulk caps. Wait `5 minutes` minimum after power-off and unplug. Use a `10 kΩ 5 W` bleeder resistor across each bulk cap's terminals to drain residual `385 V DC`. Verify zero volts with a multimeter at every cap before touching anything. This is the step where people who skip safety lose fingertips or lives. Take it seriously — primary-side bulk caps store enough energy to stop your heart.

ESR sweep the entire cap bank. With caps discharged and PSU off, use a Peak `ESR70` or equivalent in-circuit ESR meter. Sweep every electrolytic — primary bulk, secondary filter, control-side. Note any cap with ESR `>2×` datasheet maximum. Photograph the board layout before you remove anything; mark polarity on the silkscreen with a fine marker.

Identify and source replacement caps. Common P21E primary bulk: `2 × 470 µF 450 V` (or `2 × 560 µF 450 V`) electrolytic, `105 °C` rated, low-ESR. Common P21E secondary filter bank: `4-6 × 2200 µF 25 V` to `3300 µF 25 V` low-ESR `105 °C`. Replace with `5000-hour 105 °C` parts (Nichicon `LGW`/`UPW`, Rubycon `BXC`/`ZLH`, Panasonic `EE`/`FC` series). Don't undersize; don't go cheaper than spec. Verify exact values against the actual board before ordering.

De-solder and replace the secondary filter caps first — they're the dominant ripple-driver. Through-hole, modest solder mass: temperature-controlled iron at `370-380 °C` plus solder wick or desoldering pump. Mark polarity before pulling each cap. Pull straight up to avoid lifting pads. Solder fresh caps in, verify polarity (reversed electrolytics vent at full power), trim leads, inspect under magnification for cold solder joints.

Replace the primary bulk caps. Same procedure, more solder mass, more discharge safety. Polarity matters absolutely on the `385 V DC` rail — reversed bulk caps vent catastrophically the moment power is applied. Triple-check before energizing. If you lift a pad while removing the original cap, this PSU is now a Tier 4 ship-to-D-Central job — through-hole pad reconstruction on a high-current rail is bench territory, not workshop.

Stop DIY and book a D-Central PSU repair when: lifted pads on bulk-cap removal, shorted or scorched FET, secondary controller IC abnormality, scope shows ripple `>180 mVpp` at idle (controller-side failure), secondary `12 V` rail measures wrong (e.g. `15 V`+, UVP gone, hashboards at risk on every cold-start), or your re-cap was completed correctly and ripple is still out of spec at full load. You're now in test-fixture territory.

D-Central bench process for `P21E`: full strip-down, ESR sweep on every electrolytic, FET and rectifier sanity check, controller IC verification, full primary + secondary cap replacement using `5000-hour 105 °C` parts, post-repair `4-hour` burn-in at full nameplate load on a programmable test fixture with scope-monitored ripple. Bench rebuild `$120-$220 CAD` plus parts, full warranty on the rebuild.

The retire-or-rebuild call. Worth rebuilding if (a) you have multiple `P21E`s on hand, (b) functional `M30S` / `M30S+` miners they're paired with, and (c) rebuild cost is under `50%` of going used-PSU price. If your single P21E is paired with a single aging `M30S+`, the better economic path is often a fresh `P22` or step up to a current-generation Whatsminer with a modern PSU. Fleet operators almost always come out ahead on the re-cap; single-rig home miners frequently better off retiring the PSU.