Goldshell CK6 – Hashboard Hashrate Below 80%

Run a clean 60-minute baseline. Power off the CK6 at the wall. Wait 60 seconds for full capacitor drain. Power on. Watch fans ramp and the green LED come solid. Open the web UI dashboard and start a 60-minute timer. Do not touch the miner, do not change settings, do not refresh obsessively. After 60 minutes, read Hashrate Avg (not Realtime, which lies). CK6 nameplate is 19.3 TH/s; the 80% threshold is 15.4 TH/s. If the 60-minute average is at least 17 TH/s, you were watching a transient. If it is stuck under 15.4 TH/s, you have a real fault. Log wattage, ambient temp, dashboard hashrate, and firmware version before doing anything else.

Switch the pool to Dxpool CKB as a control test. Web UI, Pool Configuration. Screenshot existing settings. Set Pool 1 URL to stratum+tcp://ckb.ss.dxpool.com:7777, Worker to YOUR_DXPOOL_USERNAME.ck6-01, Password to x. Save. Cold boot. Wait 30 minutes. Open the Dxpool dashboard. Read your CKB worker effective hashrate. If it climbs toward 19 TH/s, the original pool was outside Goldshell's whitelist and silently rejecting valid shares (Zendesk KB 17169151093657). F2Pool CKB is the other officially supported option. NiceHash, Prohashing, and self-hosted stratum are not on the compatibility list.

Verify worker-name format and strip pasted unicode. CK6 cgminer is picky about formatting. The field must be exactly username.worker with one literal dot, alphanumerics and hyphens only on the worker suffix. No spaces, no underscores in some firmware revisions, no special characters, no emojis, and absolutely no zero-width characters from a copy-pasted PDF or chat message. Retype the worker manually in the web UI rather than pasting. Save. Cold boot. Confirm shares are accepted on the pool dashboard.

Confirm ambient temperature and intake clearance. CK6 hashboards throttle once junction temperature crosses the firmware ceiling. With ambient above 30 C or the intake choked with dust, the firmware quietly drops hashrate to keep silicon alive. Move the miner out of any closed cabinet, give the intake a minimum 20 cm of clear airflow, and re-test. If the room is hot - Canadian basement in July, garage in summer - improve ventilation before chasing hardware. A box fan blowing across the intake is a legitimate diagnostic tool.

Factory reset via the web UI as a clean-slate baseline. Web UI, Setting, Reset. Wait for reboot. Reconfigure pool, worker, and password from scratch. No restored config files. Stale cgminer config entries accumulate across firmware updates and can leave a chip in a partially-disabled state that survives soft reboots. A factory reset followed by hand-typed config is the cleanest software-side test before opening the case. Screenshot known-good settings after success for future text-based reference.

Hard-wire ethernet and disable WiFi for any further work. Goldshell officially requires ethernet for firmware operations and recommends it for sustained mining. WiFi flashes brick devices at unacceptable rates per Zendesk KB 16805936980633. Run a temporary cable drop if your CK6 lives on WiFi today. This is non-negotiable before touching firmware in Step 8. The single highest-ROI prevention rule on the Goldshell BOX/CK line.

Change default admin / 123456789 credentials. The CK6 ships with the same default credential as the rest of the BOX/CK line. This is a documented pool-hijack botware target if the miner is port-forwarded to the public internet (see Andreas Mai security writeup). Before any further diagnostic work, change the password via the web UI or passwd over SSH. Mining Hacker sovereignty baseline - non-optional.

Downgrade firmware to the community-confirmed stable release. Check Goldshell firmware GitHub issues for the current consensus stable revision for CK6. Download from the official Goldshell firmware mirror or GitHub release page. Web UI, System, Firmware Upgrade, over ethernet. Wait the full 5-10 minute flash cycle without touching anything. Confirm version string after reboot. Cold boot. Run the 60-minute baseline again. Firmware regressions account for ~25-30% of underperforming tickets that turn out not to be silicon.

SSH log dive. SSH is usually enabled on CK6 firmware. After changing the default password, ssh admin@<miner-ip>. Once in, run tail -n 200 /tmp/cgminer.log and dmesg | tail -n 100. Look for chip X disabled, chain Y partial, nonce error, ASIC frequency tune fail, I2C error. Count disabled chips. Note any chain that reports a smaller-than-expected chip count. Paste anomalies to the D-Central support Discord or a GitHub issue. Pattern-matching the log against known firmware bugs is a five-minute Mining Hacker community win.

Multimeter check on the 12 V rail under load. With the case open (miner powered on this time, be careful around the PSU), measure the 12 V rail at the hashboard input connector. Healthy reads 12.0 V plus or minus 0.2 V under load. Dipping to 11.4 V or lower under hash-rate ramp = PSU is sagging and chips at the end of the chain are starving. Replace the PSU with a known-good unit or a bench supply rated for 4,000 W on 12 V and re-test before assuming silicon damage.

Re-seat hashboard FPC ribbons. Power off, unplug, wait 5 minutes for caps to drain. Open the case. Locate the ribbon cables between the control board and each hashboard. They are small FPC connectors with a hinged latch - handle with care, the ribbons tear if folded against the grain. Pop each latch, gently remove the ribbon, inspect for tears or oxidation, re-seat fully, close the latch. Reassemble. Cold boot. If hashrate jumps back to nameplate, the ribbon was partially seated. A reseat fix often comes back within a few months and warrants a ribbon replacement.

Repaste the hashboard heatspreaders. CK6 ships with factory thermal paste that pumps out after 12-18 months. Power off, unplug, drain caps. Remove heatsink mounting bolts in a star pattern (don't crank one corner - uneven pressure cracks chips). Lift the heatsink. Clean old paste with isopropyl alcohol (>=99% IPA, lint-free wipes). Apply fresh thermal interface - Arctic MX-6, Honeywell PTM7950 (phase-change pad, best for ASICs), or Thermal Grizzly Kryonaut. Reassemble in the same star pattern with even torque. Cold boot, run 60-minute baseline. Repaste alone often recovers 5-15 percentage points on aged CK6 hashboards.

Per-chip / per-chain isolation across hashboards. If you have access to a second CK6, swap individual hashboards between the two units. The fault following the board = bad hashboard. The fault staying in the same slot regardless of board = control board / power stage issue on the suspect unit. This is a diagnostic test, not a production fix - return everything to its original chassis once you know which side is bad, then plan the replacement properly.

Replace the hashboard. Replacement CK6 / CK5 hashboards are stocked by parts specialists like Zeus Mining and BT-Miners. CK6 hashboard replacement runs roughly CAD $400-700 depending on revision and supplier (verify against current quotes - supply is volatile). Swap involves removing the heatsink, disconnecting the ribbon and power leads, mounting the new board, reseating thermal interface, and re-cabling. Document firmware version before and after; a fresh board may ship with a different factory firmware that needs re-flashing to your known-good revision.

Replace the control board. If isolation points to control board failure (rare on CK6, more common on aged KD/LT/HS units), replacement control boards are available through Zeus Mining and similar parts specialists at CAD $100-180. Swap is screws + ribbon + power. Fresh control boards ship with factory firmware - re-flash to your known-good before pointing at a pool, and re-secure the default admin password immediately on first boot before leaving the unit on the network.

Ship to D-Central ASIC Repair when Tier 1-3 fixes are exhausted and the hashboard still reports under 80% of nameplate, the hashboard shows visible damage (scorched pads, lifted components, swollen caps, corrosion), per-chip isolation suggests reflow or chip-replacement territory, you don't have the bench tools to do safe FPC work or thermal repaste, or the cost of replacement parts exceeds the residual value of the unit. D-Central runs the only dedicated Goldshell repair vertical in Canada with Canadian-dollar pricing and Canadian shipping. Book at d-central.tech/services/asic-repair/ and ship with model, firmware version, symptom log excerpt, and a photo of the hashboard.