High failure rates with S19K PRO

Hey folks, We've been noticing a pretty high failure rate with the S19K PRO miners. The hash boards are starting to fail, like losing chips, after just 3-4 months. Currently, we're seeing around a 6% failure rate in under 6 months of use... they’ve been repaired under warranty, but some of them have failed again. Honestly, the S19J wasn’t the best, but these K models seem to be way worse. We got these units new from several batches (though they were pretty close together). Does anyone have real-life experience with these miners? Thanks!

Same experience, the most recent batches being the worst that are all aluminum substrate hashboards. Bitmain regressed back to the S17 nightmare design. Even after repairing them the poor quality solder can cause problems like intermittency from cold solder joints all over the board that may be barely disconnected. The initial release batches that are fiberglass backed still run well in immersion

Curious… which variant of the 19K Pro are we talking about here? I’ve had about 60 Of the 115T variant miners come through my shop - 30 for personal use, 25 for resale, 5 to tear down to make single board setups. Now granted I’ve not overclocked any of them (actually underclocked and undervolted them with ePIC controllers slightly to get them closer to 20.5-21 j/TH - but I only had 1 fail which was DOA (middle hashboard only detected 44 chips during initialization and the machine would just power itself back down). All of the other ones have been fine. More recently I’ve been getting the 120T variant which has a larger hashboard and different cooling solution outfitted to the boards (much larger, single heatsink that is bolted on as opposed to the pasted on single chip sinks found on the 115T boards). Is it one model or the other you guys have experienced high failure rates with? Or all of them?

Mine are mix of 115T and 120T both are failing slowly under 12m warranty period. All are the huge one heatsink on the hash board versions.

Interesting.... the large heatsink on the 276mm board is the 120T. I've only had 6-7 of those particular variants and actually yes - 1 of them DID ship with 2 faulty hashboards (only detecting 35 and 49 chips of the 77) rendering the machine inoperable. I only had 1 of DOZENS of the 115T (individually pasted heatsinks on the 266mm board) that shipped DOA.

I'll jump in here to say I am also seeing this issue with 115T and 120T k pro models - . Still working through the list but definitely worse than S19, J Pro. etc

I have about 7 units S19KPro 120TH units offline at the moment all but 1 of which has a cvitek control boards. Seems to me that the rate of failure is now much higher as the outside temps get colder. My guess would be there maybe an OEM firmware issue that happens when the units are coming out of curtailment and are cold. The units just don't seem heat up they way they should resulting in them crapping out. Gonna be switching the control boards out for Amlogic and using either LuxOS (1st choice) or Braiins to see if that sorts things out a bit.

For us colder weather (not cold, just colder, as we are in TX) actually improved situation. It seems that around 20C they like a lot. The double whammy: Bitmain repair centers starting to not fixing the boards, but reporting unit as "scrapped". So Bitmain still sucks a lot.

When you have little to no competition, you get away with shit behaviour like this

Yeah Bitmain seems to be pumping out quantity not great quality. Even the S21's are requiring repasting already which is just silly...

just had all 3 hash boards pack up and go home within  2 weeks,  one after the other...  I switched over to Braiins to see if I can maybe  still go a bit on the 2 that still worked after the first one crapped out,  but to no avail  they lasted not 20 minutes before they too joined their buddy...  this time they were able to get the planned obsolescence to actually kick in just after the 6 month warranty...

Too bad there are still no good (or any) alternatives. So we keep giving bitmain a fortune for questionable service + product. I don't have high hopes that it will change soon...

There are at least 2 other alternatives: Canaan and Whatsminers. While both are slightly less efficient, both are far more reliable.

My experience with Whatsminers (M50 line) is that they just don't recover from curtailment in a timely manner. As a result I have stayed away since discovering this the hard way. And yes I have tried fast boot etc in their firmware. Can't comment on the "Pro" side of Canaan miners. The S19K's flaw seems to pertain to expansion and contraction of the soldering between the chips when there are extreme temp changes. Additionally the S17 heatsink design flaw fiasco was carried over to many of the S19K models which is utter crap.

Canaan also doesn't work for hot climates as they only specs up to 35C. As you write the reason of death for S19s is heat it could be that Bitmain just wrote a larger number even when it causes issues. [/quote] My experience with Whatsminers (M50 line) is that they just don't recover from curtailment in a timely manner. As a result I have stayed away since discovering this the hard way. And yes I have tried fast boot etc in their firmware. Can't comment on the "Pro" side of Canaan miners. The S19K's flaw seems to pertain to expansion and contraction of the soldering between the chips when there are extreme temp changes. Additionally the S17 heatsink design flaw fiasco was carried over to many of the S19K models which is utter crap. [/quote]