A single P-channel MOSFET in a SOT-23 package killed a 168 TB Seagate LaCie Thunderbolt 3 RAID array belonging to the Slow Mo Guys. The unit, acquired in 2019 and now out of warranty and out of production, was dead on arrival with no fan activity and no power rails coming up. Seagate support offered nothing, so the device went to repair channel Mend it Mark.
With no schematics and no service documentation, Mark traced the fault methodically: check power rails, identify the controller, follow the signal chain. The culprit was an NXP LPC11U6x-series MCU acting as power management, which was withholding rail enable signals while waiting on fan controller feedback from the SATA backplane. One faulty SI2319-compatible MOSFET near a fan connector broke that feedback loop and took the entire 168 TB system down with it. The part costs cents.
The array powered up after the swap, but the full test comes when the drives go back in. Read the original for the step-by-step diagnostic logic, which is the actual value here: a clean example of fault isolation on an undocumented proprietary system. The broader point is harder to ignore. Seagate killed a product, pulled the documentation, and left owners of expensive hardware with no path to repair. One open schematic would have made this a 20-minute fix instead of a teardown project.
[READ ORIGINAL →]