T21 - hashboard issues / missing ASICS

[danieleither]

I have a T21 which was working perfectly on all 3 hashboards. I had to power down for maintenance (not on the miner itself) however when I powered back up, this miner now has missing ASICS as follows:

[2025/04/21 19:02:27] WARN: chain#1 - 107 of 108 chips detected, attempt 1
[2025/04/21 19:02:27] WARN: chain#2 - 15 of 108 chips detected, attempt 1
[2025/04/21 19:02:27] WARN: chain#3 - 0 of 108 chips detected, attempt 1

I'm no stranger to mining, or missing chips (at least on my Bitmain junk) however I find it strange this was working perfectly then has such an issue after a simple power recycle. I have tried the obvious and re-seated hashboard and control board connections to no avail.

So my question, is it possible this could be a control board issue rather than dead hashboards?

[philipma1957]

I have a T21 which was working perfectly on all 3 hashboards. I had to power down for maintenance (not on the miner itself) however when I powered back up, this miner now has missing ASICS as follows:

[2025/04/21 19:02:27] WARN: chain#1 - 107 of 108 chips detected, attempt 1
[2025/04/21 19:02:27] WARN: chain#2 - 15 of 108 chips detected, attempt 1
[2025/04/21 19:02:27] WARN: chain#3 - 0 of 108 chips detected, attempt 1

I'm no stranger to mining, or missing chips (at least on my Bitmain junk) however I find it strange this was working perfectly then has such an issue after a simple power recycle. I have tried the obvious and re-seated hashboard and control board connections to no avail.

So my question, is it possible this could be a control board issue rather than dead hashboards?

well three hashboards from good to bad with a controlled turn off is odd.

did you try swapping three white wires from controller to hashboard?

i am running four t21s but on six cases two boards per case and epic controllers.

i have had one bad epic controller that gave me false readings.

i used about 17 epic controllers.

and i get a dead board read on and off  with one t21 board.

So it is a definite maybe the controller or the wires and less likely the hashboards.

[danieleither]

I have a T21 which was working perfectly on all 3 hashboards. I had to power down for maintenance (not on the miner itself) however when I powered back up, this miner now has missing ASICS as follows:

[2025/04/21 19:02:27] WARN: chain#1 - 107 of 108 chips detected, attempt 1
[2025/04/21 19:02:27] WARN: chain#2 - 15 of 108 chips detected, attempt 1
[2025/04/21 19:02:27] WARN: chain#3 - 0 of 108 chips detected, attempt 1

I'm no stranger to mining, or missing chips (at least on my Bitmain junk) however I find it strange this was working perfectly then has such an issue after a simple power recycle. I have tried the obvious and re-seated hashboard and control board connections to no avail.

So my question, is it possible this could be a control board issue rather than dead hashboards?

well three hashboards from good to bad with a controlled turn off is odd.

did you try swapping three white wires from controller to hashboard?

i am running four t21s but on six cases two boards per case and epic controllers.

i have had one bad epic controller that gave me false readings.

i used about 17 epic controllers.

and i get a dead board read on and off  with one t21 board.

So it is a definite maybe the controller or the wires and less likely the hashboards.

I didn't. I was on a tight schedule with no time for further investigation as I was heading off on holiday. I am now back home, however the site is very remote (3 hour drive) so just trying to get some ideas before I drive down there

[BitMaxz]

What exact procedure of maintenance do you do? Did you just use compressed air to clean them up?

How about the temp and fan speed?

If it worked before, I believe this undetectable ASIC chips has poor contact with the heatsink.
If you can disassemble the unit again, I recommend cleaning it up, removing the old thermal paste, and replacing it.

Also, test them one by one because sometimes other hashboard affects the other hashboard to send false signals.

[philipma1957]

What exact procedure of maintenance do you do? Did you just use compressed air to clean them up?

How about the temp and fan speed?

If it worked before, I believe this undetectable ASIC chips has poor contact with the heatsink.
If you can disassemble the unit again, I recommend cleaning it up, removing the old thermal paste, and replacing it.

Also, test them one by one because sometimes other hashboard affects the other hashboard to send false signals.

I did a repaste on some t21 boards and they do work now.

You could be correct.

[danieleither]

What exact procedure of maintenance do you do? Did you just use compressed air to clean them up?

How about the temp and fan speed?

If it worked before, I believe this undetectable ASIC chips has poor contact with the heatsink.
If you can disassemble the unit again, I recommend cleaning it up, removing the old thermal paste, and replacing it.

Also, test them one by one because sometimes other hashboard affects the other hashboard to send false signals.

I didn't touch the miner prior to the fault. As I stated in my OP, I shut down power to the miner as I was conducting non-related maintenance at the site.

The miner was literally working perfectly, turned off for 10 minutes or so (I turned off all power at the site while I added a breaker in the distribution panel), then when it was turned back on it had 3 dead hashboards. If I had been working on the miner itself, it would make more sense and point to something I'd done. Also the miner runs from a 12-way Bitmain PDU, and the other 11 on the PDU continue to work fine (therefore nothing to do with me installing a new breaker)

I have since been down to the site and changed control board and PSU, but neither fixed the issue. It just literally went from working perfectly to 3 dead boards after a power recycle. Very strange!

I can only assume some sort of power spike / anomaly from the PSU when power was removed.

[mikeywith]

If it worked before, I believe this undetectable ASIC chips has poor contact with the heatsink.

The miner would run just fine without heatsinks, it will work for a few seconds then it will overheat and will give you a temp-error in the kernel log and goes to sleep or on some models it would just enter a reboot-loop, I think what you are describing is the chip itself is loose in respect to the PCB, and that my friend is most likely the reason.

@danieleither, I can tell you for a fact that in most cases, the bad boards will die during a power recycle, mainly caused by what's known as "thermal shock," the rapid change of temp will crack poor solder joints, the miner is more likely to break when it goes from hot to cold than staying hot, you already had at least one badly soldered chip, when you powered the miner off temp changed very rapidly from 70c or so to near room temp, what happened there is uneven shrinking of different components, the result is a mechanical stress on the chips to the PCB, so now there is a "short" in that hashboard and thus it's reading 0 asics whereby in reality it could be just 1 bad ship that needs resoldering, funny enough, this shit only happens on Antminers, it was very common on their 17 series in general and anything with a T is usually terribly made by Bitshit.

If you are still deep into mining and using Antshit, you may consider getting a fixture tool and some tools to start fixing your own hashboards, if you don't have shaky hands, I believe you are smart enough to figure it out.

[danieleither]

If it worked before, I believe this undetectable ASIC chips has poor contact with the heatsink.

The miner would run just fine without heatsinks, it will work for a few seconds then it will overheat and will give you a temp-error in the kernel log and goes to sleep or on some models it would just enter a reboot-loop, I think what you are describing is the chip itself is loose in respect to the PCB, and that my friend is most likely the reason.

@danieleither, I can tell you for a fact that in most cases, the bad boards will die during a power recycle, mainly caused by what's known as "thermal shock," the rapid change of temp will crack poor solder joints, the miner is more likely to break when it goes from hot to cold than staying hot, you already had at least one badly soldered chip, when you powered the miner off temp changed very rapidly from 70c or so to near room temp, what happened there is uneven shrinking of different components, the result is a mechanical stress on the chips to the PCB, so now there is a "short" in that hashboard and thus it's reading 0 asics whereby in reality it could be just 1 bad ship that needs resoldering, funny enough, this shit only happens on Antminers, it was very common on their 17 series in general and anything with a T is usually terribly made by Bitshit.

If you are still deep into mining and using Antshit, you may consider getting a fixture tool and some tools to start fixing your own hashboards, if you don't have shaky hands, I believe you are smart enough to figure it out.

I hate Antminers. My Whatsminers (which I don't have any more) were far better. The only reason I keep buying Antminers is the availability of Whatsminers is very poor. There never seems to be any available when I am in a position to purchase. Antminers are also easier to sell used, at least here in the UK

[mikeywith]

I hate Antminers. My Whatsminers (which I don't have any more) were far better. The only reason I keep buying Antminers is the availability of Whatsminers is very poor. There never seems to be any available when I am in a position to purchase. Antminers are also easier to sell used, at least here in the UK

That's worldwide, I suppose, mainly for two reasons IMO.

1- Bitmain manufactures a lot more gear than MicroBT
2- MicroBt focuses more on large farms than small miners (Bitmain is slowly following)

Regarding the second point, you can see it in their PSU, a single C19 plug for a massive 3-3.5kw is not available at every home, so many average users avoid them for that reason.

[philipma1957]

my epic controlled 2 board t21 with repasted heatsinks work really well.

[danieleither]

my epic controlled 2 board t21 with repasted heatsinks work really well.

My antminers that work well, work well. lol... but generally they are crap and you know that Phil!