NerdQAxe – 0.0 GH/s With WiFi Connected

Cold power-cycle for 30 seconds. Unplug the 12V input at the PSU and the XT30 or barrel at the miner (capacitor discharge is faster with both ends open). Count 30s out loud. Plug PSU first, then miner. Watch the dashboard for 2 full minutes before declaring failure - the stratum pipeline takes time to rebuild. Do NOT use the web UI `Restart` button for this diagnostic; the BM1370 chain is more reliable via cold cycle.

Verify PSU specs on the label. Physically read your 12V brick. You need `12V / >=10A / 120W` for NerdQAxe++, `12V / >=5A` for NerdQAxe (non-++). If label says `12V / 8A` or below, stop - you've found the root cause. Swap for a D-Central-recommended NerdQAxe PSU or a proven brand (Mean Well, Lite-On). Undersized supplies may work at boot and fail once the ASIC chain pulls full load.

Re-seat the 12V connector (barrel or XT30). Unplug, firmly reconnect, wiggle-test once hashing resumes. Any power dip from finger pressure means the connector is worn or the cable is stressed. A loose XT30 is the #2 false-positive for silent zero behind undersized PSU. Replace the cable or reflow the board-side connector if intermittent.

Switch pool. In the web UI, change pool URL to `public-pool.io:21496` or `solo.ckpool.org:3333` (well-tested solo Bitcoin pools). Save, reboot. If hashrate comes up on the new pool but not the old one, your original pool was blocking jobs (IP ban, rate limit, or template-refresh stall). You can swap back after 30 minutes or stay on the new pool permanently.

Factory-reset NVS via the web UI. `Settings -> Reset to Defaults`. Reconfigure WiFi SSID/password and pool URL. Cold power-cycle. A corrupted NVS block (typically from aggressive OC experimentation) can cause silent boot failures that mimic this bug. Lowest-effort firmware-side fix, costs nothing.

Measure 12V rail with a multimeter under load. DC setting, probe on the miner-side of the 12V connector while the unit is trying to hash. Expected: `11.7-12.3V` sustained. Below `11.4V` = PSU sag confirmed regardless of label. Swap PSU and retest. Label-rated `10A` cheap supplies can sag worse than label-rated `6A` quality units - nameplate is a starting point, not a guarantee.

Capture boot serial log. USB-C, `115200 8N1`, terminal of your choice. Cold-cycle and record the first 20 seconds. Three lines matter: `asic_chain: BM1370 x N detected`, `stratum: subscribed`, `stratum: notify received`. Missing any of those tells you which subsystem failed. Save the log as a text file - if you escalate to D-Central or Discord, this is the single most useful attachment.

Flash the latest stable firmware. Pull the current `ESP-Miner-NerdQAxePlus` release from shufps on GitHub (or D-Central-curated build). Verify the build targets `BM1370`, not `BM1368` or `BM1366` - mismatched chip targets produce this exact silent-zero signature. Flash via OTA if possible, otherwise via the NerdQAxe web flasher over USB-C. Post-flash: factory reset NVS, reconfigure, cold-cycle.

Test on a different network. Phone hotspot for 5 minutes. If hashrate comes up on hotspot but not on home WiFi, your router is rate-limiting outbound TCP on the stratum port, your ISP is rate-limiting, or your home network has MTU / fragmentation issues. Check router logs for dropped connections from the miner's IP. Bypass: static route or QoS priority for the miner's MAC address.

Confirm chip type via web UI. `System Info` panel should list `4 x BM1370`. If it shows `BM1368` or `BM1366`, your firmware is wrong for the hardware - the chain enumerates but won't hash because protocol framing mismatches. Reflash the correct target. If it shows `3 x BM1370` or `2 x BM1370`, jump to Tier 3 - you have a chain-level hardware fault, not firmware.

Log-capture a full 10-minute session. With serial capture running, let the miner sit at `authorized, 0 GH/s` for 10 minutes. Look for recurring patterns: is the stratum client reconnecting every few seconds? Is the chain reporting `chip timeout` repeatedly? Is the watchdog resetting the ESP32 without telling the UI? These patterns pin root cause to one subsystem and direct the next action.

Oscilloscope the 12V rail under load. DC-coupled, `20 MHz` BW, probe at the miner-side connector. Expected: `11.7-12.3V` DC, under `200 mV` peak-to-peak ripple, no transient dips below `11.4V` even during brief hashrate spikes. Ripple above `500 mV` or transient sag below `11V` indicates a failing PSU filter cap or undersized supply. Swap PSU before touching anything else.

Scope VCORE at the BM1370 package. Probe on the output ceramics of the TPS546-class regulator on the miner board. Expected: `1.05-1.15V` DC, under `20 mV` ripple. If VCORE reads near `0V` or oscillates wildly, the regulator isn't delivering - indicates a TPS546-level failure similar to the Bitaxe Gamma `VCORE init failed` path. VRM swap is fine-pitch QFN rework, hot-air only.

Re-apply thermal paste on all four BM1370 chips. Degraded or gap-filled TIM can cause electrical contact issues with the mount, not just thermal - ground/mount path matters for chain enumeration. Disassemble, remove old paste with IPA 99%, apply Arctic MX-6 or Kryonaut in a uniform thin layer (rice-grain per chip, spread by heatsink torque not finger), reassemble to spec torque. Repair for NerdQAxe++ units `>12 months` in service.

Chain-break reflow. If Step 10 showed fewer than 4 BM1370 detected, the failing chip is cold-soldered, cracked from shipping, or silicon-dead. Reflow the suspect BM1370: preheat `150 C` bottom-side, top-side `310-330 C` for ~`30 s`, cool naturally, re-paste, reassemble. If reflow doesn't recover, replacement from graded stock is next - see the chain-break dedicated page for full procedure.

Inspect the XT30 or barrel jack on the board side. Cold solder joint on the power connector is an underreported cause of `silent zero under load` - the connector passes continuity at multimeter currents but fails at `8A+`. Reflow or re-solder the connector to board pads. Use lead-free or leaded solder matching the board's original build; don't mix.

Stop DIY. You've swapped PSU, confirmed clean `12V` rail under load, flashed latest correct firmware, factory-reset NVS, tested two pools and two networks, and hashrate is still zero. OR you see `<4` chips enumerating and lack a preheat/hot-air station. OR you see burn marks, discoloration, or burnt smell anywhere on the board. The NerdQAxe is `$200-$400` hardware - extensive DIY beyond this point often costs more than the bench fee. Book a D-Central repair slot.

Ship safely to D-Central. Anti-static bag around the PCB, bubble wrap, rigid outer box with `>=5 cm` foam on all sides. Include a one-page note: exact firmware version string, PSU spec (brand + label rating), serial-log capture on USB stick or pasted, pool and worker name used, whether any cold power-cycle ever transiently recovered hashrate, and approximate purchase date. Canada-wide and worldwide shipping from Montreal. Return turnaround `5-10` business days.