I too sometimes get these 'untouched' and 'brand-new' items on my desk.
Never take a customers word for it.
Buzzing sounds are usually caused by coils, which in turn points to the power management MT92.
The coils are probably fine but the MT92 or some capacitors might not be,
Did you also check what the voltage rails look like on a scope?
Wanted to give you an update, it’s gone from “assumed dead NAND” to a board where everything tests healthy and it boots OFW, so I’m now leaning toward a transient fault rather than a dead component, and I wanted to share the findings. I just wanted to add that should have been more clear on the noise, it was loud, and I think it was coming from the speakers
On the testing: all the main rails checked out with no shorts against a reference OLED. I went over the M92T36 thoroughly, taking diode readings on all four sides pin by pin against a reference board, and every pin matched within tolerance — the chip’s healthy and charging behaviour is normal. The 3V3 rail initially read low in diode mode (around 0.20V vs ~0.34V on the reference), which had me chasing a leak, so I did a voltage injection on the rail at 1.1V with a 0.5A limit and it only drew 1mA — no meaningful leak at all. Re-measuring diode mode afterward gave 0.39V, which is normal, so that initial 0.20V looks to have been a measurement artifact, though possibly intermittent. Then I pulled up the eMMC info in Hekate and the module reads completely healthy — full CID and Ext CSD, HS400 at 400MB/s, Estimated Life 100% on both, reserved usage normal, and all partitions present and correctly sized. So the eMMC clearly isn’t dead despite the fault appearing mid-backup. And then it booted straight into OFW with no issue.
My working theory at this point is a transient or intermittent short — possibly a stray solder ball or flux bridge that was active under load during the backup, which would explain why it surfaced during sustained eMMC access, why the rails were clean when I probed them cold, and why it’s now booting fine. It could have since been dislodged, or it might still be intermittently present. The slightly wandering 3V3 reading across sessions might point that way too.
A few things I’d love input on if anyone’s been here: has anyone had 2162-0002 turn out to be a transient short rather than a dead component, on a board that otherwise tested this clean? Would you chase the “fault only under sustained load, clean when cold” pattern as an intermittent SoC/RAM joint, or is that more likely to be install debris in your experience? And does anyone know of issues with Hekatos specifically during eMMC backup that could throw a boot error like this without actually damaging the eMMC? Cheers.
