Bitcoin Forum

One of my Antminer S5s, powered with a Corsair 750m PSU, sometimes has almost 2% hardware errors and either chain #1 or #2 powers down (as it is >10°C cooler than the other chain). Why is this? It's just this one S5 that has this issue.

2% is still acceptable. But since you mention problem on chain 1 & 2, probably dead chips or bad connection.

No dead chips are reported.

Sometimes chain #1 fails, and other times it's chain #2. I haven't seen both fail at the same time though. That's what's peculiar about it.

My average temperature is 52 °C, and the temp never exceeds about 68 °C. (Where the miner is isn't in a very constant temperature environment.)

Maybe near dead chips, but you can check the ribbon cable and tighten them.

Yes, I do have a Rosewill 750W Capstone PSU I could try. It's an 80+ Gold, which technically should make it better than the Corsair 750m (which is a Bronze), even though the Rosewill was less expensive, right?

Yes, that's correct.everything normal: 350 clock rate, 1.1 Thash/s hashing on average

I updated to the new, April 15 firmware. Let's see if that fixes things.

So, I'm using another PSU with it, and one of the two chains is still a much cooler (~20°C cooler) than the other, and it's only hashing at 526 Ghash/second (@350 clock rate). Clearly, one chain is dead.

So, that was using P2Pool. Switching to Eligius, things are hashing at normal rates, both chains are operating, etc. I really wonder why that would be.

Why do I only have 4 operational ASICs on Chain #2? Did the other chips short out or something?

Now my whole Chain #2 died. It's somewhat intermittent:
https://i.stack.imgur.com/Eyzta.png

Why is my AntMiner S5 chain only intermittently operational? What could cause this, how could I diagnose it, and what is the solution? Will BitmainTech replace it for me? How do I know if it's under warranty? Is there a warranty?

On warrenty did you buy from bitmain?  And has it been under 90 days.   That is warrenty and you must buy it from them for it.

And first look through - https://bitmain.zendesk.com/hc/en-us/articles/204875015-How-to-diagnose-S5-hashing-PCB

Then if needed try - https://bitmain.zendesk.com/hc/en-us/articles/203559969-My-S5-hashes-at-about-600GH-S

Yes, it came from Bitmain, but it's been more than 90 days…I'll check those out; thank you

Oddly, I unplugged the "Power Cable to Control" (with it still running), which cut the power to it, and plugged it back in. I also sprayed the Chain #2 board with some electrical contact cleaner, and that helped. Now the little red light on that board is on! Thank you & may God bless you!

Update: But the it's no longer on now! ☹ It just shows: "------ -------- -------- -------- " in the "ASIC status". Hmm…

If the problem was on one hashing board, I'd say its the S5, but because the problem is moving it becomes 50x more likely that its the shared resource, being the PSU. Test with your second PSU and see where the problem migrates, if it remains.

It has the same problem regardless the power supply. I have a Corsair and Rosewill and have tried both.

Yay! The boards are dated March 3, 2015! But, I bought it via Craigslist. Am I out of luck? I know the person I bought it from bought it from Bitmain directly. Should I ask him for the order number?

I would contact him to get order number.   You will need that to do a warranty RMA. Also would contact them soon as 90 day on that most likely is quick approaching.

I'm sorry to possibly pop bubble I don't know how strict they are but it's possible they could consider it not warranty with it being bought from another party and not directly them :(  - https://bitmain.zendesk.com/hc/en-us/articles/202661989-Limited-90-Day-Warranty-Terms: This warranty is valid only in the United States and Canada and grants specific legal rights, it does not extend to owners of the product other than the original owner.

But you might ask them nicely and see if they are willing to help.  Please let us know if they do give you work under warrenty or not.

Best of luck

They actually know I'm not the original owner, but they don't care. They just said they'd need the order number. :)

I had this happen with my miner. What I found out is that the connection between the hash board and the central board was not good. Tightening the cable connections and rebooting the miner fixed it.

Thats good news from you!   I was not sure how close they would follow the term.  I was hoping they did not follow it and would help out.

They really are helpful and a good asset to use.   Great support team.

Yes, Bitmain's support is much better than Coinbase's (https://bitcointalk.org/index.php?topic=1015360) (so is Circle's).

I sprayed some electrical contact cleaner on the pins, and it changed things a bit:

Under the "ASIC#" column, http://192.168.1.xxx/cgi-bin/minerStatus.cgi reports 30, so it's seeing all the ASICs, but in the "ASIC status" column, it displays "------ -------- -------- --------", not "oooooo oooooooo oooooooo oooooooo". The chain is certainly drawing watts (as my watt meter reports about 600W), but it must not be communicating with the central board properly.

The connection is good enough that it knows the temp, etc., but apparently not good enough to transfer more data?

The temp is 5°C higher than the other chain. Maybe I do need to put new thermal paste? I checked it, and it didn't look bad.

Wow, Bitmain is awesome! They received on Friday my fried board I sent them, and already today I received an email from them:

:) they really do have good support.  Always happy to see stories like this.

Even better: Now that I put more thermal paste on it, I'm getting 10°C cooler temps (https://bitcointalk.org/index.php?topic=1046270.msg11598266#msg11598266)!