Quick answer

The EEPROM stores each board's ID and calibration. If it can't be read, or three boards don't match, the controller refuses to run them.

EEPROM faults are often recoverable by re-imaging from a known-good board.

Full diagnostic flow

Did this start after mixing boards from different miners?

Yes, boards were swapped around → EEPROM mismatch. Boards must share consistent EEPROM data to hash together — synchronise all three to one valid image. [DIY with care] Error-code guide →
No, single original board → Reseat the 18-pin signal cable (it carries the EEPROM I2C lines). Recognised now?

Reseat the 18-pin signal cable (it carries the EEPROM I2C lines). Recognised now?

Now recognised → It was an I2C/cable contact issue. Done — keep an eye on that connector. [Fix it yourself]
Still not recognised → With an EEPROM tool, can you read the board's EEPROM (U5/U6/U10 by model)?

With an EEPROM tool, can you read the board's EEPROM (U5/U6/U10 by model)?

Reads but data is corrupt → Corrupt EEPROM. Read a known-good board's image and write it to this one to recover it. [Advanced DIY] Error-code guide →
Won't read at all → Failed EEPROM IC or a dead I2C bus. Replace the EEPROM and program it, or check for a stuck SDA/SCL. Bench work. [Send it in] Error-code guide →

This points at board-level damage. You can attempt it, but the realistic call for most owners is to send it to a repair bench.

Fix it yourself

Work the matching symptom decision tree and error-code guide step by step. Most cable, cooling, network and firmware faults are solvable on the bench you already own.

Send it to D-Central

Cracked solder, dead domains, shorted chips and EEPROM faults need a real repair bench. See transparent repair pricing or get a free repair quote — we diagnose and repair 38+ ASIC models in Laval, board-level, since 2016.

More tools: Kernel-Log Decoder · ASIC Fault Finder (650 codes) · All symptom trees

Related products, repair, and setup paths

Last reviewed June 12, 2026.