NerdQAxe – Overheat Mode at 75C Shutdown

Cold power-cycle the NerdQAxe: unplug the 12 V barrel / XT30 connector for a full 30 seconds, then repower. This guarantees the firmware's latched Overheat Mode flag is cleared — some builds keep the flag set across soft restarts from the web UI. Expect the device to boot, attempt to resume mining, and climb back to target temp within 3-5 minutes. If it re-trips immediately, skip to Step 4.

Lift the device off carpet, wood, fabric, or any heat-trapping surface. Set it on ceramic tile, bare metal, or a silicone mat with 30 cm of clear airspace on every side. Point a cheap USB or desk fan at the heatsink from ~1 m away at low speed. Every 5 degrees C drop in ambient buys you roughly 5 degrees C at the VREG sensor — this alone fixes a large share of 'failed' NerdQAxe++ reports in D-Central's support queue.

Drop ambient temperature if the room is above 27 degrees C. Crack a window, run A/C, or relocate the miner to a cooler basement or garage corner. Measure ambient at the intake grille of the NerdQAxe itself with an IR thermometer, not across the room — the intake is what the heatsink sees.

Revert to stock overclock profile. In the web UI set ASIC Frequency = 525 MHz and Core Voltage = 1150 mV (NerdQAxe++ stock). Reboot and monitor VREG temp for 30 minutes. If VREG settles at 60-68 degrees C at stock, your overclock was the trigger — rebuild it slower later (Tier 3). If VREG still crosses 75 degrees C at stock, the problem is thermal/electrical not tuning — continue.

Flash the latest stable esp-miner-nerdqaxeplus or esp-miner-nerdqaxeplus2 firmware from shufps' GitHub releases via the browser-based web flasher (https://github.com/shufps/nerdqaxe-web-flasher). Older builds reported VREG temp 5-7 degrees C higher than actual, producing false-positive trips; recent builds are accurate and expose a Max VREG Temp slider for user-tunable protection thresholds.

Verify PSU output under load. Put a multimeter in DC mode and probe the 12 V barrel jack at the device side while the miner is hashing at full power. Expect >= 11.8 V sustained. Then read the PSU label — it must be 12 V and at least 10 A (120 W) for NerdQAxe++. Swap to a known-good 12 V / 15 A unit (Meanwell GST160A12 or D-Central-spec brick). We have measured 10-15 degrees C VREG drops from a PSU swap alone on undersized-PSU NerdQAxe++ cases.

Re-seat the XT30 or barrel connector. A loose or oxidised connector adds 50-100 m-ohms of contact resistance, enough under 5 A of load to produce visible droop and VREG heating. Unplug, inspect both halves for blackening or oxidation, reconnect until you hear and feel the click. On XT30 connectors pay attention to whether both blades fully seat — partial seat is common.

Open the chassis (four Phillips screws on NerdQAxe++; add two heatsink clip screws on ++ Remastered). Inspect the tower-cooler mount. It must sit square on the BM1370 die, with the spring screws seated evenly — a 1-2 degree tilt leaves one corner of the die uncovered and the VREG domain stealing the contact. Re-seat if askew; do not over-torque the springs.

Re-apply thermal paste on the BM1370. Remove the heatsink carefully. Clean both the die and the cold plate with IPA 99% and lint-free wipes until the surfaces shine. Apply a single rice-grain bead of Arctic MX-6 (or Thermal Grizzly Kryonaut for warmer rooms) to the centre of the die. Mount the heatsink straight down — do not twist — and tighten the spring screws in an X pattern hand-tight, not torqued. Boot and watch VREG temp for 30 minutes.

Replace the VREG-side thermal pad. Peel the old silicone pad between the PCB VREG region and the backplate / chassis shell. Wipe residue with IPA. Cut a fresh 1.5 mm or 2.0 mm silicone pad (Thermalright Odyssey 12.8 W/mK or Arctic TP-3) to cover the full VREG footprint. Close the chassis firmly so the pad compresses 15-20% — that compression is what makes it conduct. A pad that is too thick traps air and performs worse than a correctly-sized pad.

Tune overclock on a temperature budget. On stock paste + ambient <= 24 degrees C, NerdQAxe++ typically holds 550-565 MHz at 1170-1185 mV with VREG at 66-70 degrees C. Add +5 MHz and +5 mV per 15-minute stability window. Stop at the last setting before VREG sustained temp crosses 68 degrees C. Write the profile down — silicon lottery means every unit overclocks slightly differently. Do not push hashrate past your thermal budget; chip longevity costs more than 50 GH/s of headroom.

Upgrade the fan. Stock 92 mm fans on some NerdQAxe++ production batches rattle at 100% duty and cool 8-12 degrees C worse than a Noctua NF-A9 PWM. Swap: 4-pin PWM header, 92 x 25 mm, 12 V. Route cable away from the ASIC. Use the web UI Fan Override control to verify full duty spins up the new fan correctly.

Add a chassis-exit fan. On NerdQAxe++ Remastered enclosures, drill out one side panel for a 40 mm exhaust fan wired to the spare 5 V pad. Turns a saturating box into a through-flow duct. Community build logs on the NerdQAxe Discord document 4-8 degrees C VREG drops from this mod alone. Not factory-endorsed — document in case you sell the unit.

Reflow the BM1370 (rare, last-resort before chip replacement). If paste, pad, PSU, fan, firmware, and tuning are all clean and the VREG still trips at stock in a cool room, the chip-to-PCB BGA may have cold joints. Preheat bottom side to ~150 degrees C, top-side hot-air at 310-330 degrees C for ~30 s, cool naturally, re-paste, reassemble. Applies to <5% of cases we see but closes the loop before Tier 4.

Inspect on-PCB buck converter capacitors and MLCCs. The 12 V to core-voltage rail runs through a buck converter stage on the NerdQAxe++ PCB. Bulging electrolytics or cracked MLCCs near the buck IC raise ripple and push the VREG domain hotter. Loupe inspection first; soldering-iron swap if you find damaged parts, matching exact ratings (voltage, capacitance, package size).

Stop DIY and book a D-Central ASIC / Open-Source Miner Repair slot when two rounds of re-paste + re-pad did not drop VREG below 70 degrees C at stock in a 22 degrees C room, or you see burnt components / cracked MLCCs, or the device refuses to boot / chain-breaks after a thermal event. This is bench-tool and thermal-imaging territory — not home-workshop territory. Ship address and intake form at https://d-central.tech/services/asic-repair/.

D-Central bench process: thermal-imaging audit (FLIR) under load to identify the exact heat source — VREG integrated on BM1370, on-PCB buck converter, or input-side filter. Reflow or component replacement as indicated, sourcing BM1370 from graded inventory if the chip itself is dead. Full 12-hour post-repair burn-in at the user's specified overclock with logged VREG and ASIC temps. Results report delivered with the unit.

Ship safely. Place the NerdQAxe in an anti-static bag, double-box with at least 5 cm of foam on every side. Include a note with: firmware version at time of failure, overclock profile you were running, ambient room temp, PSU make/model/amperage, date of purchase, and a short description of symptom onset. That metadata cuts 30-45 minutes off our diagnostic time, which cuts the invoice.