Goldshell – WiFi Upgrade Corrupts Firmware (Use Ethernet)

Plug in Ethernet. Run a Cat5e cable from your router or switch to the miner's RJ45 jack. Wait 60-90 s for the miner to pick up the new wired IP. Browse to `find.goldshell.com`, locate the new wired IP, log in, and confirm you are now operating over Ethernet. This single step prevents an estimated 60% of Goldshell brick events on the D-Central bench. It costs less than a coffee and takes less than five minutes.

Power down the miner before changing the network. Don't hot-swap from WiFi to Ethernet during a session. Power off cleanly via the UI's reboot/shutdown function or, if no UI option exists, pull the 12V barrel after a 30 s idle. Plug Ethernet, then power up. This avoids the worst-case race where DHCP resolves on both interfaces simultaneously and the routing table picks the wrong one.

Disable WiFi in the miner UI after a successful Ethernet handshake. Most BOX miners have a Network -> WiFi toggle. Setting it to off eliminates any chance the miner falls back to WiFi if the wired link drops momentarily during an upgrade. Belt-and-suspenders, but it is the path the experienced VoskCoinTalk crowd uses.

Verify the wired IP is stable for 10 min before clicking Upgrade. Refresh the miner status page every minute. If you see any `connection lost` flash, your wired link itself has a problem - bad cable, bad switch port, wrong duplex. Resolve the wired stability issue first, then upgrade.

Run a wired loopback test on the cable. Use a $20 RJ45 cable tester or a known-good laptop. Confirm all 8 wires are intact and no pairs are crossed. A failing pair drops Ethernet to 10 Mbps half-duplex, which is slow enough for a firmware upload to fail - though it fails more gracefully than WiFi will. Replace any cable that does not pass.

Check switch port LEDs. A solid green / amber link LED on both ends is mandatory. A blinking-but-no-link state means auto-negotiation failed - try a different switch port. Cheap unmanaged switches occasionally lose port-level state and need a power-cycle to clear it.

Verify your router is not dropping the miner from DHCP mid-session. Set a static DHCP reservation for the miner's MAC address in your router's admin panel. This eliminates the failure mode where the router renews the lease mid-upgrade and the miner briefly loses its IP.

Snapshot current firmware version + config before upgrading. Take screenshots of the dashboard, firmware version page, pool config, and worker name. If recovery is needed, you'll be able to email Goldshell support with exact context. This shaves hours off a recovery turnaround if anything goes sideways.

Email `hello@goldshell.com` BEFORE the upgrade if you are crossing major versions. Request the recovery `burn-*.img` for your model up front. Goldshell support takes 24-72 hours to respond - having the recovery image staged on a microSD card before you upgrade means a brick is a 30 min recovery, not a 4-day support-ticket wait.

Monitor the upgrade with `tcpdump` or Wireshark on a laptop. Filter for the miner's IP. You will see a sustained TCP stream during the upload. If you see packet retransmits or window-zero events, the upload is suffering even on Ethernet - stop, fix the cable / switch, retry.

Pre-stage the SD-card recovery image on a microSD before any major upgrade. Use a 4-32 GB microSD (NOT a 64 GB+ SDXC - the BOX bootloader chokes on exFAT-only cards). Flash the matching `burn-<model>.img` with Balena Etcher or `dd if=burn.img of=/dev/sdX bs=4M status=progress`. Keep the card in a labelled envelope next to the miner. Recovery time drops from days to 30 minutes.

Run the upgrade from a hardwired laptop, not a phone or WiFi-connected desktop. The browser issuing the upgrade `POST` should itself be on Ethernet. A WiFi-connected laptop driving an Ethernet-connected miner can still drop the upload mid-stream if the laptop's WiFi falters. End-to-end wired is the rule.

Validate firmware checksums before flashing. Goldshell publishes SHA-256 hashes on `goldshellhelp.zendesk.com` for each release. On Linux: `sha256sum <firmware>.img` and compare. On Windows: `certutil -hashfile <firmware>.img SHA256`. A mismatch means the download itself was corrupted - retry the download before clicking Upgrade.

Test the upgrade on one miner before rolling to a fleet. If you run 3+ Goldshell devices, upgrade one in isolation and run it 24 h before touching the rest. Goldshell firmware revisions ship with regressions - a known-good fleet beats a fully-current bricked fleet.

Disable any WiFi power-save features on the laptop driving the upgrade. macOS, Windows, and Linux all default to aggressive power-save on WiFi. If the upgrade is being driven from a laptop, plug the laptop's power brick in and disable WiFi power-save in OS network settings before clicking Upgrade. Even on a wired miner, a flapping driver-side connection can break the upload.

Stop and ship to D-Central if you have already followed Tiers 1-3 and the miner still bricks during a wired upgrade. Repeated wired-upgrade failures point at eMMC wear, control-board fault, or a hardware-level issue that DIY recovery will not fix. The D-Central ASIC repair service handles SD-recovery, eMMC reflow, and control-board swaps for the entire BOX-series lineup. The bench sees 8-12 BOX miners a month for this exact failure class.