Topic: Constant Dropping of one Board Antminer S5! (ALL of THEM) (Read 3116 times)

The problem could be a firmware issue that has never been fixed from the first firmware iteration. Otherwise it would be a serial controller defect which i kind of doubt. It could be easily tested if someone had a unit that does this often and switch boards, see if its the same board or the other board that now has the problem.

Ok i Think the FW has nothing to do with the drops i have tried Following FW:

Jan from Bitmain.
April from Bitmain
Graphs Nicehash
Smiths FW
July from Bitmain

I have tried them for periods of two weeks and still happens.. Im really getting frustated of not founding the problem.

Correct, it can happen from 0 to 3 times a day, last long enough to register a dip on hash graph on external website;


Basically, if i wasn't watching my Antminer S5 and noticing one blade dropping, i would probably just think its normal variance. Right now looking at the 1 week graph, i can only see 2. The one i notice last generally 30seconds to 2 minutes. But that might just be because i don't notice the smaller once since i don't have the monitor refreshing every few seconds.

I think there may be more than one problem here? The drop I am seeing is very short, a second or so, whereas looking at your graph I think it is longer? The OP has also mentioned that their Internet is not the best.

Mine has not done it again since I started watching closely, so I will retrace my steps with the other PSU. If I can get it to repeat I will also try going back a firmware version as it's quite recent that I upgraded.

Rich

But i have no internet issue. My ping is very small and does not spike. Still i am waiting to hear from you again, maybe you can figure it out.

I had never seen this, but Today was testing some PSU's and a board dropped out briefly, red light off, hashing stopped and power dropped, came back on almost immediately, red light back on. Thought it was the PSU so changed back to the normal one and still had a couple of drops.

I have a theory which is that is is an Internet problem. I am having a very slow speeds at the moment with high latency. Of course since I then stated closely monitoring it has not done it again.

Also running on the latest firmware with the fan %.

Rich

I've noticed this myself with s5 miners using newer firmware. Back flashed several to the April 2015 version and the dropouts from them seem to have stopped. Hmm.

I just got two more S5's and one of them seems to have the exact issue you describe.  One hashing boards drops out, temp drops off to idle on that board, and then it starts working again after a short time.  It is the right side board as viewed from the front.  The other new one is not doing it so far and neither of my older ones do this (although one of the older ones does something similar but worse as I pointed out above).  It is worth noting that the newer ones also both have the newest firmware while my older ones have the Wed Apr 15 15:17:03 CST 2015 firmware.

I mean M's Miner Monitor;
https://bitcointalksearch.org/topic/ms-miner-antsp-monitor-v52-alertsautomassscheduled-rebootmobile-miner-596178

MMinerMonitor.exe or MMM as i like to call it.

Pretty sure they means M's Pool Monitor http://96.44.166.190/MPoolMonitor42.zip

It appears to be "SD-S5-20150107.tar.gz".  I also had some heat happening during it.  I ended up making up new mining area with a lot of CFM.

It is a batch 1 for what it's worth.  It also has not had the problem where network drop kills it.  So i could just be lucky and and it's a pretty good unit.  When I did have the problem for a few day's I had to manually turn off PSU and turn on PSU.

What firmware are you using? I am using the latest from Bitmain. "SD-S5-20150715-fan_ctrl.tar.gz". I have the few minutes dropping for always the top board problem and i have not tried other firmware.

It all depend on the miner.  The S5's really do have a personality.  Some are much better then others.  Mine did it a few times for a few day's, then stopped and it's bee great for months.

I did do the wire check and do the firmware install.  So i'm guessing my firmware update was what really did it.  But in no way do these fix evey S5 with dropping a blade.

Also which one is MMMonitor?  I could not find it by name.

I see, this is not the issue i have. At least, the board only drop for a few minutes at most, then get re-picked up. I hope my issue won't devolve in your issue :S.

I mine in Kano and i use MMMonitor.

I have 2 and one of them does this every so often.  It requires a power cycle to make it start working with both boards again.  It only happens maybe twice a month or so, so not a huge deal but certainly annoying.  I absolutely do check hashrate on my miners every morning and then several times throughout the day.  I wish there was a way to make it correct remotely without a power cycle but I have not been able to.

I'll just pipe in and say that out of 15 S5's I have 3 have been pulled from the main farm and put on the Group-W bench for misbehaving. With mine it started withing the past 3 months.

They also have a habit of dropping the same card every few days. With these sometimes a soft boot will cure it, most often take a hard boot. Only difference on my rapscallions is that looking at the non-fan end on 2 of them it is the card on the left and on 1 it is the right side card. The misbehaving ones range from batch -1 to batch-5. However, for each of the 3 it is always the same card.

What do you mean in common? I certainly do not have Venezuela low end internet. And i can't possibly see a way how a board dropping has anything to do with your ISP.

The only thing we all have in common is the internet provider.. I am starting to think is has something to do with the internet.. We are using best of the worst of Internet in Venezuela

It's a very intermittent thing, and lasts for a very short time.  Overall the average rate is stable even looking at my own S5's and to specs.  Not saying they won't take that 200 GH drop for no reason what-so-ever, but it recovers pretty quick.

As long as the chips are running fine and you're not getting HW errors out the back side I think it's something that's just a design flaw.

If you're looking for two more used Bitmain S5s and are in the US let me know.  I'll toss em your way at a loss to what I purchased them for, but you pay for shipping.  Have the original packaging so they would be securely shipped.  Would rather put that money toward another S7.

Yes, i think so too.

I am also under the impression it happens to many people without them being aware of it. Many have the mentality; "It works for me so you're doing something wrong." without really understanding the issue.

Finally.. i realized a lot of people has this "problem" and they don't know!

last week i was helping someone on facebook setting up his cryptoglance.. and i saw one of his miner hashing half of it.. and i said like "hey look,, that's the problem i told you about, it's happening in one of your miners" and he was like.. "shit, i have never seen that before"

im pretty sure it's a common problem. but people are not all day looking at their miners, they just care about the average hashrate

I have the exact same "problem" i decided to just ignore it as overall its 2-3ghs loss;

I posted about this in several places and got a few replies but nothing helpful. I can sum it up to people posting the same 101 troubleshooting suggestions and a handful of others saying they have the same problem.

It still happen once or twice a day. Its the top blade if you look at it in the webserver in miner status.

It look like a sharp dip in the pool's graph everytime;


Typically only last 2-3 minutes.

Edit: My S5 is batch 7 (used from bitmain) on the latest PCB.

That's the one im using in all my miners, tried smiths firmware too. nicehash firmware too.

Use this Firmware


https://bitmaintech.com/files/download/SD-S5-20150715-fan_ctrl.tar.gz

Each of us setup the miners, but out of curiosity.. miners are basicly plug and play, you just connect 4 PCIE, connect RJ45, set up pools.. leaving default frecuency on all miners. All temperatures are bellow 58, where could posibly be room for a wrong setup?

If 22 miners are the same controller board problem, it can only be problem in the software.

That was the answer I was hoping to read here.

I think this must be a tad exaggerated. Just the chances of 22 miners having same problems are slim.   Did the same person setup all of these miners?

I read the PSU's do you have any that are a little bit higher quality?  Those are all ok... none great.  But I suggest posting some networking screens, status screen, etc.   I think 22 miners something chances are something was set wrong.

I have one s5 that behaves the same way. set up a crontab entry with cgminer -restart every 6 hours

Well assuming this is not a wind up. and I only say that because S5 has been round for some time and I have never seen the same problem or seen it reported before. So a couple of questions & ideas.

First carry out all the suggestions above, do not reject ideas because you do not think they will work as these people are very experienced. In particular do try reversing the controller board as described by Tupsu as it is far from certain that the problem will stay with the controller board or swap to the other. I think you should also swap the fan to the other end of the S5 to keep the airflow the same.

I have seen a single hash board drop on my S5, it's a rare occasion, and a lot of the time was when I was playing with reducing the 12V to undervolt them. Usually when doing this one hash board would drop out before the other. But here's an odd thing. On all occasions when I have seen a single board drop out I have never had it recover on it's own. I have always had to reboot, which you do not seem to have to do?

So what other info do you have? What frequency are you tuning at? Worth trying dropping it a notch or two and see if that fixes the problem?
Finally can you think of any common factor between this "group" of S5's that might not be present generally? Mains supply, temperature, location etc?


Rich

1- We use 4 Cables from one psu into 1 S5

2- They are all in different locations

3- Different Pools.. Nicehash, Antpool, Bw

okay lets start right here.

you have s5's correct?

every psu you use is a bare minimum psu.  they all should be able to supply enough power yet they act like they don't.

So first question is on each psu do you :

 use 4 cables from the one psu into 1 s-5  or do you use 2 cables from 1 psu into 1 s-5?

Next question is the gear in different locations with different internet.  Or all in the same location?

Next question do you mine on the same pools or different pools?

You will see the miner hashing bellow 700 GHS, and if you take a closer look to the miner the RED LED is off, after 6 minutes the RED LED turns back on the board temperature rises from 30 C to 50 C again (if you touch the heatsink while the RED LED is off you can feel it's cold

Sounds extremely unlikely. How do you know that the left board turns off for 6 minutes?

Rich

What are the odds that You  and You friends have bought 22 miners with the controller board   problems???

Come on, i would know if the problem it's the controller board im talking that 22 miners from different batch do the same thing! and people here come and say to check the PSUs.. it's like they are not reading what i typed..

What are the odds that me and my friends have bought 22 miners with the left BOARD with problems???

Obviously, this is boards problem.
They are sometime overheated and now we have some chips with poor contact.

My recommendation. Turn the controller to the other side, so that data cables can be exchanged. It is only four screws, at Controller metal frame.
Check , the error is repeated on the same blades or not.

Official ideas:
https://bitmain.zendesk.com/hc/en-us/articles/204875015-How-to-diagnose-S5-hashing-PCB
https://bitmain.zendesk.com/hc/en-us/articles/203559969-My-S5-hashes-at-about-600GH-S

If that does not work try other PSU's as others have said.  Some S5's have a personality and do this.  You will notice it is posted about more then a few times.

The problem is not because of the PSUs, we tried with different PSUs, like i explained

Try switching the power plugs into the hashing boards, and see what happens. If the problem moves, then the issue is the power supply.

I have a couple of S5s each powered by their own EVGA 750 Gold units without issues. These units have been hashing since the S5 was released; I think mine are batch 2.

So! me and my friends got each like 8 antminers S5, different batches, using a Corsair CX750M, HP DSP 800GB and DELL N750P on the miners.. we all use different routers, switchs, firmwares, but we all suffer from the same thing.

One of the boards on the miner will turn off for about 6 minutes, and then turn on again, its always the left board if you look the miner from where it pulls air. it happens in ALL the miners! not even one saves from this problem!

Its random, so in the morning it can happen to 2 of them, and in the afternoon to 1.. it can happen twice in 24 hours to the same miner.

How did i find out about it? one day i was checking the miner Hashrate 5 seconds and realized it was hashing under 600 ghs for 6 minutes and then went back to 1.15th, you can even see the red led on the board turned off!