Topic: ANTMINER S5: 1155GH(+OverClock Potential), In Stock $0.319/GH & 0.51W/GH - page 16. (Read 451319 times)

If I loose some Hash on my S5 I have usually found that just doing a Save & Apply on the Frequency setting, without changing the frequency, get's things going again. Failing that a soft reboot seems to work, have not ever found a power down being necessary.

Not much use to you.... but just my experience.

Rich

Every now and then, I will lose half or all of my hashing power on the S5. Powering off and on the unit will fix it but simply rebooting from the command line or web interface (both are working fine) will not. The status line for the blades shows all 'o's. Is there a way to bring it back to life remotely or am I going to have to work out some kind of remote power control??

Edit: Dropping the clock and restarting brought half of the hashing back. The other board is all Xs now so at least that's getting somewhere.

Edit2: Another reboot and all 'o's again but my hashrate is still around 1/2 what it should be. Think I'll have to rig up an R-Pi for remote power switching.

OK good news, there must be some version(s) of the S5 firmware that dropped or messed up 4 board support. I have been very happy with the latest rease with fan control, so will be sticking with that. Have also tried the Nicehash release but had some problems, not multiboard related.

The 4 Board system was my most efficient overall, but because of the current I could take from my 2 Pairs of 5V PSU's they were running on, 250MHZ was as high as I could go without running out of Power. I will return to the setup but wanted to split them to allow running of one while experimenting with the other.

Rich

 I once read this:
 "One other item to consider in making a "4-blade S5 Extreme" miner. The most recent firmware that Bitmain has released will NOT support four blades on a single controller. If that doesn't bother you then feel free to proceed. I personally would keep it as two separate miners. I think Ethernet switches are pretty cheap. A used one couldn't even cost $10, could it?"

And there was link to that fw that I first tried. It worked, but not so well  .

Edit: after 20h average hashrate is 1.73TH. So far so good. And temps are much lower too.

That's good to know, means I was just lucky that the latest firmware supports 4 boards, had not realised that they all did not do this?

Rich

Yep, I flashed that same fw and it works perfect. Thanks for Tupsu for mailing link to me.

I am running the SD-S5-20150715-fan_ctrl.tar.gz firmware on both my S5 aand have had no problems including the 4 board test. It is just possible that it is frequency related and for some reason it cannot cope with three boards at higher frequencies, although my total Hash Rate was higher than yours so I would have thought it was ok? Might be worth taking the clock down a bit just to see if you do then get the correct hash rate?


Rich

I usually try to solve problems myself but this one goes over my head. Is there any other fw than 20150107 that supports more than two boards. And I'll swap that board to another s5 and see what the hashrate then will be.

That is odd I have run 4 Hash Boards on one controller and got the correct Hash Rate. I was undervolting & Underclocking the Boards at the time at 10 Volts & 250MHz but am uncertain if that would be the reason it was ok? My total Hash rate was 1650GH and 615W at the Wall. Not running things that way at the moment so can't easily go and check anything.

Rich

Well, I have a minor problem. I added an extra hashboard to one controller, totalling 3 hashboards/controller. All 3 chains are found,no dead chips,no hw errors. But total hashrate is only 1.33 TH. I expected 1.6 TH. All blades are good and with separate controller I get hashrate right. So what could be the cause of the problem? I power it with dps-2000 server psu, so it gets enough juice. And yes power consumption is about 910W from wall, I measured it.

I have 2 from the 1st batch that are still hashing with zero problems since I got them.

S5 getting on a bit now in Miner Years & people don't tend to post that theirs is fit and well... 


Rich

This thread should be re-named to "broken / fix / repair S5" thread or something. Just went through 12 pages of the same thing.

The S5 Hashboard is a simple design, but tricky to troubleshoot & repair because of it's string design often gives a "One Down All Down" result. I will at some point do a Trouble Shooting & Repair thread, but although my understanding of the Board is improving all the time I do not feel qualified to do that yet. I also only have one failed board to play with which is slowing things.

Back to your question it is really too complicated to cover in a few lines, however the first step is to check the voltage across each of the 15 nodes (The capacitor you have just replaced) These should all be about 0.8V, a "faulty" node will be more or less than this, however this is complicated in practice because a fault driven increase or decrease on a node will also result in the converse change spread over the remaining nodes.

Next thing to check is the 14V from the Upconverter at the Top Right of the board, and then check that the 25MHz oscillator Y1 is running.

Bypassing nodes is possible, but only practical if it is one of the upper nodes that has failed, and even then you would need to reduce the 12V supply accordingly.

Bottom line is almost certainly an ASIC or two has failed (The Node voltages will help confirm this) and the only real course of action is to remove & replace. The replacement would have to be taken from another board as that is the only source...


Rich

Thank you for the reply!  So, I replaced the capacitor, but the board still does not hash     All ASIC's show 'oooooooo' in the status, but the hashrate is zero, I would have expected to see oooo--oo  or ooooxxoo or whatnot...

Is there a way to bypass what may be the dead ASIC's without overly affecting the rest of the chips / voltages on the boards?

The cap is fried, however most likely it has fried because of a failure of one or both of the BM1384 ASIC chips on the Node. Dependant on the exact mode of failure of the BM1384 the 2V capacitor will have been overvolted causing failure. Measuring across the capacitor / resistor should read about 4.7 Ohms (Do allow for resistance of the meter leads by connecting them together before measuring)

So you can try replacing the capacitor but almost certainly one or both of the BM1384 chips will also have failed. Changing them is much more difficult.

Rich

Has anyone managed to effect a repair on hashboards where this part has burned?

I have in front of me an S5 which obviously burned in this manner, on each hash board the third (R to L on top row) 2v 330uf cap is burned to white powder on top, and on one of the boards the second one has started to go as well.  I'm also measuring low resistance on the 4.99 Ohm Chip resistor above each of the badly burned caps but the resistor on the half burned capacitor is testing nominally.

Any other components to check before replacing the caps / resistors?  It seems that if a 2V cap fried this badly, there surely must be damage to the ASIC pair around it, but perhaps these are miracle boards...?

Anyone who has worked on these please chime in with any thoughts or advice. 

Cheers!

P.S. if anyone has digi or mouser part #'s that would be great.  I saw J4bberwock posted a 4v part which should be fine, but if anyone has the exact #'s it is appreciated!

need advice..
i modified my s5 with watercooling, each bb connected with 4 blades, each 2 blade connected with 1 psu.
last night i got one of psu stop working, the consequently is the water pump stopped (i connect water pump with psu).
the other psu (with bb power connected) still running and still hashing until i found it with temp is 106C and hash 16ghs
i turn it off, replace the psu (that connect with pump) and power it on again...
violaa the previous blade with 106C degree is not working, and only show ------
[spoiler=blade]

[/spoiler]
i calculate the average temp is (106 + 106 + 0 + 0)/4 = 53, thats why the miner did not stop working, the average is below 80 degrees.

now i need advice, how to repair my blade with ------ caused by high temp?

If you have defective S5 boards in EU, without any chance to get them exchanged by Bitmain, I'm interested to buy them if your price is fair.
And I'm also interested in 12 to 20 defective S3 boards, just in case someone have a few laying around.

If you want to just replace U101 then the part to look for is  RT8537 - Current mode Boost-type LED driver

However not sure how easy they are to find?


Rich