I had the same issue. I tried different pools and a firmware update. After the latter I noticed the one of boards was heating a bit more and started losing hashrate earlier than others.
The seller suggested to send it to a service center.
But then I decided to check hashboards (like it is suggested in the paper instruction manual). And I found the problem.
One of hashboards fell out of the bottom channel and got stuck completely in a wrong position during delivery.
Here's how it looked like:
I detached all other parts (including all main fans) from the main aluminium body. Used a blowdryer to warm up the body a bit and pushed the hashboard with both hands on the radiator closer to the bottom channel. Little by little it returned to the bottom channel.
After I assembled it back it has been working like a charm.
The seller suggested to send it to a service center.
But then I decided to check hashboards (like it is suggested in the paper instruction manual). And I found the problem.
One of hashboards fell out of the bottom channel and got stuck completely in a wrong position during delivery.
Here's how it looked like:
I detached all other parts (including all main fans) from the main aluminium body. Used a blowdryer to warm up the body a bit and pushed the hashboard with both hands on the radiator closer to the bottom channel. Little by little it returned to the bottom channel.
After I assembled it back it has been working like a charm.
I encountered the same thing today. Same middle hashboard slammed inside, stuck. But it is still defective. Retightened the screws of the heatsink etc. its a RMA for me. We encountered over 10% faulty S21+ in our batches. Mostly defective hashboards.