S19 Damage from Water and Snow

6 replies 287 views
Posts: 3 · Reputation: 31
#1Dec 15, 2021, 03:23 PM
So, I just set up an S19J Pro at home, and I vented it outside to keep it from overheating. I kinda underestimated the suction from the vents, and it ended up sucking in some snow during a storm. I took it apart and dried everything, but now only a third of the hashboards show up on the dashboard, which means I'm only getting about a third of the hashing power I should have. I even tried a factory reset, but that didn’t help. I’m new to this mining thing and not really sure what steps to take next or where to find replacement parts. Any tips would be super helpful.
2 Reply Quote Share
moondev98Newbie
Posts: 5 · Reputation: 0
#2Dec 15, 2021, 05:29 PM
First about venting: If you are talking about anything resembling the unit on a table exhausting its hot air toward an opened window, the unit will wither be too close, too far or both, depending on the weather. There is no sweet spot, unless you can somehow force a positive pressure in the room. Fans from the unit produce airflow that cannot compensate for the negative pressure in the room. If you have only one unit, a properly sized air hose from the exhaust of the unit to the outside will "force" the air to go out, and you have to prevent water from following the hose to the unit, for example, by making the air exit below the unit. Hopefully it is helpful for you next setup, or prevents that mishap to someone else. Unless actively forcing a positive pressure in a building, the wind direction and its environment will decide where air comes in and where it goes out. Now that it happened... WARNING: If some part(s) makes some other part(s) fail, then playing around may damage parts that are not yet damaged. It appears that you may have 2 failed hashboards. Sometimes, somehow, swapping things around fixes stuff for unknown reasons. If you haven't done so already, you should clean the boards. Removing dirt alone may fix it. Then look at the logs. Sharing it here may help. You want to know, for example, if the boards are not-detected or detected but malfunctioning. It may literally tell you what is the problem. After identifying which board is working, you know that its data cable and its power cables are working. You can test each data port by connecting the working board with the working data cable to each port of the control board. If the working board does NOT work when connected to the two other data ports, then the control board is almost certainly the (or a) problem. If the board DOES works in all data ports, then you can test each data cables, then each power cables, etc. to isolate the problem. A bad hashboard, bad power supply, a bad data cable, a bad connection of the data cable and/or a bad control board, or any combination of those may have the very same symptoms.  The physical position of the parts may squeeze or stretch something just a bit in a way that makes the difference between it working and not working. If you are equipped to do so and "have a clue" about it, I would suggest testing the power supply first, then testing for short circuits on the hashbords power inputs. Did you contact Bitmain? It is unlikely covered by the warranty, but they may be able to give you some directions. Best of luck!
1 Reply Quote Share
blockhubMember
Posts: 1033 · Reputation: 43
#3Dec 15, 2021, 10:03 PM
Check those two hashboard not showing on the dashboard physically and maybe there are corrosion and rust after you dried them. Also, if you want us to troubleshoot it for you then post the whole kernel logs here, and don't forget to put it inside the code tag(a # button you can find this while editing your post). But if you decided to hire a tech contact zeusbtc.com for help.
5 Reply Quote Share
Posts: 3 · Reputation: 31
#4Dec 16, 2021, 01:51 AM
Thanks for the help. I believe I have narrowed down the issue to the two hashboards, the good hashboard will show up as 1, 2, or 3 if I swap out the cables on the control board. I tried cleaning some of the visible corrosion off but still only 1/3 boards is registering on the dashboard. Looking through the logs, it appears some of the ASIC chips are not being detected.
1 Reply Quote Share
blockhubMember
Posts: 1033 · Reputation: 43
#5Dec 16, 2021, 05:03 AM
Now I see the problem it only detects a few ASIC chips and the hashboard power off if the number of detected chips is lower than the expected number. Possible few ASIC chips are damaged on that two hashboard or maybe there is a leak/colder joint on a few ASIC chips that need to be reflow with hot air. If you don't have BGA hot air try to use a hair blower and blow hot air to the broken hashboard 3 to 8 minutes and try to run it again. If you can't find the exact hashboad which one is the broken and working one test them one by one just disconnect the other two hashboard PSU cable and ribbon cable from the control board.
4 Reply Quote Share
Posts: 3 · Reputation: 31
#6Dec 18, 2021, 03:06 PM
Update - I was able to send the boards out for repair. The boards themselves were fixed as I received the same serial numbers back and I am now getting the expected hash rate. Dumb question - the boards were returned with stickers on them to show which chips were repaired. Can I leave them on there so I know which chips had issues or will the boards get too hot?
0 Reply Quote Share
Posts: 210 · Reputation: 2
#7Dec 20, 2021, 07:15 AM
I don't think that the stickers block any airflow...
3 Reply Quote Share
?Reply
Sign in to reply to this topic

Related topics