What causes chain scission of chips?

[space_cat]

Hello everyone, I just diagnosed an Antminer S9 of mine and it has something called "Chain Scission of chips". I couldn't find a lot of information on what causes that to happen. I have a couple other machines and I want to make sure they don't end up with the same problem.

[1714017043]

1714017043

[tim-bc]

This is a (relatively rare) issue that seems to arise from shorted/burnt asic chips. For example, if a heatsink has fallen off of a chip or the thermal conductivity is abnormally low, that chip will burn out (scission = cutting the chain, since the chips are wired in series). Also, if polarity is accidentally reversed to a board, capacitors can blow and/or the board might start to melt.

You could try flashing a stable firmware just in case, but otherwise don't run any possibly burned boards. Send them to bitmain for repair if possible, although marks of burning do void the warranty.

[space_cat]

Okay, that makes sense. Do you know specifically what can cause it? Chips too hot? I also didn't have that machine on a surge protector, but the PSU is fine so I'm curious if that caused it.

[tim-bc]

The problem lies specifically with one of the many chips on the board. You should be able to selectively test your hashboards to find out which one is faulted. If you hear a rattle when you shake the chassis, you have a loose heatsink. Otherwise, you can always take a look at the bad board or check the kernel log for more info.

Not much point in worrying about the issue since you effectively only have one option: sending the bad board back to bitmain for repair.

[space_cat]

Thanks man. Yeah, I tried flashing the firmware and that didn't work so clearly the board broke. I just wanted to make sure that I can prevent it in the future. 

[tim-bc]

Sorry about that, hopefully the machine is still under warranty.

If you want to catch this problem early on in the future, give new miners a shake and listen for loose heatsinks. When you start it for the first time, you can also check the board temps to make sure one isn't wildly larger than the others.

All the best