Goldshell – Temperature Monitor Shutdown Threshold

Cold-boot and read the actual numbers before touching anything. Power off, pull AC, wait 30 seconds, plug back in. Open the web UI at http://<miner-ip> with admin / 123456789 and let the miner run 15 minutes. Record chip temp per hashboard, ambient, fan RPM per fan, hashrate, and trip count in the log. If chip temp jumps >20 C in a single polling interval, suspect sensor failure (jump to Step 11). If chip temp climbs smoothly at <2 C/min from cold start, you have a real thermal event.

Vacuum the intake grille and fin face from the outside. Power off the miner. Use a vacuum cleaner with a soft-brush attachment to lift loose dust off the front intake grille and as much of the visible fin face as the nozzle reaches. Do not stick the nozzle into fan blades. This step alone resolves about 30% of trigger complaints from operators who haven't cleaned in 6-12 months. Cost: zero. Time: 5 minutes per face.

Drop ambient under 28 C at the front grille. Not room-middle — at the intake. Goldshell's 35 C ambient cap has zero hysteresis; you want 7+ C of headroom. Open a vent, run room exhaust, duct miner exhaust outdoors, or move to a cooler wall. A KD-MAX dumps ~3,360 W; in a small closed room ambient climbs 5-8 C in the first hour and the temperature monitor will not stop tripping until you fix the room. Re-test after 2 h of better airflow.

Reset firmware to factory defaults to clear any stale OC profile. In the web UI go to Settings -> Restore Factory Settings (or hold the RST pinhole for 5-10 s if web UI is unreachable). Re-pair to your network over Ethernet, change the default 123456789 password, re-add your pool. Stale aggressive overclock profiles can hold higher voltage / frequency than the silicon wants and push chip temp 3-8 C higher than stock — directly into the trigger threshold.

IR-cross-check the temperature sensor before touching the chassis. Fluke 62 MAX or equivalent IR thermometer aimed at heatsink fin tips above each hashboard while the miner is under load. If firmware says >95 C chip and IR says <65 C exhaust, the sensor is the problem — skip the airflow rabbit hole and jump to Step 11. Ten seconds with an IR gun saves three hours of cleaning the wrong thing.

Pull the side panels and inspect for visible dust mat between fins. Phillips #2 or Torx T8/T10 depending on revision. Shine a flashlight down through the fan grille at the fin stack. Run a fingertip across the fin tops. If you see a grey woven mat or fingertip lifts fibre, you are looking at packed-fin obstruction — the dominant root cause. If fins look clean, jump to Step 8 (fan check).

Deep-clean the fin stack with compressed air plus brush. Power off, AC unplugged, miner on its side. Compressed air at <=30 PSI blasted from the intake face outward at a 30-45 degree angle down the fin valleys. Vacuum the exhaust face simultaneously to capture dislodged debris. Soft anti-static brush for stuck mats. 5 minutes per side, repeat once. Goldshell's ~2 mm fin pitch traps lint at a brutal rate; this is the only technique that actually clears packed valleys.

Verify both fans run at nameplate RPM under load; replace any failing fan. Boot the miner, watch fan RPM in the web UI as it ramps from idle to full hashrate. Both should hit 4,500-5,500 RPM and stay there (KD-MAX: all four fans). Zero RPM, persistently low, or audible grinding/ticking = bearing gone, replace the fan. Stock spec: 120x120x38 mm, 12 V, 4-wire PWM. If the miner is past 2 years, swap both at once — second fan usually fails within 6-12 months of the first.

Re-seat hashboard data + power cables. Power off, AC unplugged. Open the chassis. Press each connector firmly home — vibration over time can back connectors out 1-2 mm and introduce intermittent contact, which sometimes shows up as bogus sensor readings or phantom trips. Verify no bent pins, no scorched plastic, no loose strands. Re-seat; do not over-flex the ribbons.

Cross-flash the latest stable Goldshell firmware over Ethernet only — never WiFi. Goldshell's KB explicitly bans WiFi firmware upgrades because the filesystem corruption rate is non-trivial. Plug Ethernet straight to the router. Download the latest stable .bin for your model from the Goldshell firmware page. Settings -> Firmware Upgrade -> upload. Do not power-cycle mid-flash. Older firmware versions have shipped with sensor-polling regressions; staying current rules that out.

Swap the temperature sensor cable to a different hashboard's port (KD-MAX, LT6, HS5 only — models with discrete sensor harnesses). Power off, AC unplugged. Disconnect the suspect sensor cable, plug it into a known-good hashboard's sensor port. Boot, observe. If the high reading follows the cable to the new board, the cable is the failure (<$5 part to replace). If the high reading stays with the original board, the on-board sensor IC has failed — Tier 4. Models without discrete sensor cables (BOX-series, single-board SKUs): the sensor is on-board by default, jump to Tier 4.

Pull the suspect hashboard; remove old TIM; re-paste with Arctic MX-6 / Kryonaut / PTM7950. Power off, ESD strap on. Remove hashboard mounting screws (Torx T8/T10). Carefully separate heatsink from ASIC carrier with a gentle twist motion — old TIM may bond surfaces, do not pry. Scrub all old paste off both surfaces with 99% IPA + lint-free wipes until both are mirror-clean. Apply a rice-grain dot of fresh TIM per ASIC die, re-mount heatsink, re-torque clamp screws to a consistent feel (snug-plus-quarter-turn, do not crank).

Inspect heatsink clamp tension; re-torque per Goldshell spec. Goldshell uses spring-loaded standoffs to clamp heatsink to hashboard at controlled pressure. After 2-3 years of thermal cycling, springs can lose preload by 10-20%. Visually compare clamp height across all chips on a board; any clamp sitting notably higher than its neighbours has lost preload. Re-torque all clamps to consistent feel. Uneven clamp pressure produces uneven chip temps; the hot chip drags the board's reported temp toward the trigger threshold even when everything else is healthy.

Verify thermal-pad coverage on PCH / VRM regions; replace if degraded. While the heatsink is off, inspect secondary thermal pads between heatsink and PCH / VRM ICs (the smaller power-control chips, not the main ASICs). Healthy pads = pliable, smooth, full coverage. Degraded = cracked, hardened, gaps. Replace with 1.0 mm or 1.5 mm thermal pad rated 8.5 W/m K minimum (Gelid Extreme, Thermalright Odyssey). Match the thickness of the pad you removed — wrong thickness is worse than no pad.

Burn-in 4 h after re-paste; verify chip temp stays below the model's stock baseline + 5 C. Re-assemble, power on, point at your normal pool. Watch chip temps for the full 4 h. Successful re-paste should hold chip temp 5-15 C below the pre-fix baseline at the same ambient. If chip temps still climb past the model's normal range with healthy fans, clean fins, and fresh TIM — silicon is the failure. Document burn-in numbers; you'll send them with the unit if Tier 4 is needed.

Stop DIY when chip temp still trips after verified deep clean AND verified re-paste, or sensor cable swap isolates the failure to the on-board sensor IC, or KD-MAX/LT6 chip-temp delta between boards is >15 C after a full clean + re-paste. You are now in component-level repair territory — chip-level rework, sensor IC replacement, or control-board work. Pack the miner safely (anti-static bag for boards if shipped separately, double-box with >=5 cm foam) and ship to D-Central.

D-Central bench process: IR thermal-cam diagnostics on every Goldshell repair to identify failed chip positions and drifted sensor ICs before any rework. Failed sensors get replaced as discrete components (LM75-class or PMIC thermal diode, depending on board revision). Failed chips get desoldered with hot-air rework, replaced from our salvage donor stock (CK / KD / LT / HS boards), and reflowed at 220 C preheat / 245 C peak. Control-board issues are rarer but occur — we have known-good board swaps for the entire Goldshell lineup. No 6-week China round-trip via Zeus, no shipping a KD-MAX across an ocean.

D-Central runs a 24 h burn-in at customer's preferred hashrate (or nameplate if unspecified) before return ship. We log chip temps, fan RPM, hashrate stability, pool acceptance ratio, and trip count over the full 24 h. Anything that fails burn-in goes back on the bench. You get a one-page report with the unit when it ships back — pre-repair vs post-repair temps, what was replaced, recommended ambient + maintenance interval to keep the temperature monitor quiet going forward. Typical turnaround: 5-10 business days from receipt.