Topic: ANTMINER S5: 1155GH(+OverClock Potential), In Stock $0.319/GH & 0.51W/GH - page 93. (Read 451266 times)

as soon as I get the new pcb put the MX4 and I keep you updated on the temperature

On the front of the machine there should be a factory defaults reset switch just to the right of the red and green status lights where the fan side is. You power up the miner and hold it down for say a minute, maybe 30 secs. Then shut the miner off and find the ip of the miner on your router. Give it a shot.

Keep us posted the temp difference between the one done right with the Arctic paste vs. god knows what paste. Remember you don't have to glob the hell out of it. The best way to do it is put a drop on each of the chips then put the heat sink down on the chips. Then extremely carefully move the heatsink in a gentle very small circle then take the heatsink off and clean up any extra that spills on the PCB then put the heatsink gently back down and tighten in a crossways pattern gently increasing tension so that there are no gaps and the heat sink sits evenly. That extra minute of work makes all the difference. Run it at stock or even under for a minimum of 24-48 hours before trying any overclocking. Check temperatures regularily in case a chip didn't get enough paste.

Today i receive my last order from bitmaintech and i saw that the s5 have some aluminium heatsink's over the chips, my previous ants dont have.


ps : Is there somebody tried to fix antminer s5 controller board with sd card ? One of my machines just stopped working, the miner cant get IP and i cant found/access the miner. I try to reset it several times but no success at all.....

I bought the thermal paste, Arctic MX-4, now I'm waiting for that I get the new card, and put the dough only one miner, and see the difference

Well I'm an 1h 30m in and it's ok but I've noticed something else. The "bad" board has now cooled down a couple of degrees and the "good" board jumped a couple. This leads me to believe that sloppy construction/thermal paste application is probably at fault. When you burn in a CPU for the first few days you will see temperature variations until the paste has properly spread out and filled the gaps between the heat sink and CPU. Eventually it will even out but you can minimize this by not globbing the fuck out of the CPU with heatsink paste and use real paste. It could be that batch 5 has changed the paste/moved to a new factory or whatever. Maybe the people assembling them got drunk that day and globbed the shit out of the ASICs with heatsink paste, I don't know. I remember a couple of nasty pics of the S3+'s boards taken apart where they put so much paste on there it was actually spilling all over the PCB's.(holy fuck!) My best advice is to run it stock or even underclock it a little to keep the ASIC chips just hot enough to burn through the crap and hopefully the paste will properly spread eventually on its own. After the warranty period I'm going to take these apart and do it properly but until then I won't touch it. I'll just deal with it.

I am having the same problems, after 1 hour lhash falls from 1.2 terash, falls to 496/520 gh / s, only that my problem and also the errors HW that I find about 22 000 errors, wrapped well 56 000, this just in one miner, others are fine, even though they lose power belonging hash, plus I had a faulty pcb and the bitmain is replacing with a new one (thanks bitmain) I believe as well that the problem lies in the firmware, because my miners do not exceed 50 ° C, and estena temperature about 19 ° C.
I'm using the pool and Antpool, I tried nicehash, ghash.io, f2pool, but the result and the same, after about 1 hour wrapped two hours, and the power dies, wrapped remains so until you restart the miner, and much annoying in the cellar and restart at least two, three times a day, I hope the engineers of bitmain bring out a very good firmware for S5, it seems that the batch 5 has some problem.

Yeah one of the boards is toast or sloppily assembled. Getting alot worse. If I'm lucky I get half an hour out of it before going tits up. How painful is the exchange process? Ironically both hashing lights on both boards stay on and it's still reporting "ooooo" on all asic chips. Maybe a bad controller? I'll try another few things first but as soon as it goes tits up the HW errors skyrocket like crazy. Is it just easier to take the boards out and redo the thermal paste with something real like Arctic Silver? Bummer oh well at least I got a few days out of it. I'll narrow down the bad board it's chain "1". Maybe they can just send me a replacement board instead. My guess is the board itself is probably fine, someone probably just screwed up the paste or didn't tighten down the heat sink screws enough. Maybe I'll try rolling back the firmware for fun. Keep in mind this is NOT overclocked I haven't completed my weekly burn in.

I've also noticed that chain 1(the bad board) which used to have to coolest temperature is now the hottest used to be around 48-49 is now 53-54. It could be possible that when the board was assembled at the factory they did it "right" but didn't tighten down the heat sink screws enough, this tiny gap was filled in with whatever Ho Chi Mihn cheap heat sink paste they used so for the first few days before the thermal paste could melt and spread properly, as it should should, there was no gap between the heat sink and the chips. After a few days of burn in the paste got thinner and spread out more causing gaps between heatsinks and the Asic chips causing a spike in temperatures which will cause some chips to overheat and crap out. I've seen this alot with CPU's whereby the crazy bastard who put it together decided to use the shit paste that comes with CPU's instead of proper paste and sloppily put the heat sink on and then wonder why the computer keeps crashing. Unfortunately for us utilizing the proper heat sink paste at the proper amount and carefully assembling heat sink to spread it evenly is not condusive to pumping out the shit as fast as you can on an assembly line. I can't really blame them as if I was stuck on the assembly line building this stuff day in and day out, not only would I probably get a little sloppy, but I would be hitting those Apple suicide nets every night. Now the good news is if I can keep the puppy running long enough eventually the heat sink paste may even out properly on it's own and fill in the gaps properly. Who knows maybe they just globbed on a whole ton of it so it's having a hell of a time spreading properly I don't know, I would have to take the boards apart and have a look. I'm going to watch it for a few days and see if I can get it to settle down I think before ringing up Bitmain. Keep you posted.

Well my temps barely get into the 50's with one board in the high 40's. Considering some people are running high 50's/low 60's. Heat is not the problem. Either A. I have a bad board or B. The controller is bad. Or C. the latest firmware is buggy or D the ribbon cable is weird or needs reseating or positioning on the controller board. Since there are 4 plugs for the ribbon cables I'll experiment around. I've moved the ribbon cable to the inner most position on the controller board and see what happens. Seems ok so far but it's only been an hour. I'm using the stock fan and with the cold air from outside the fan is only running around 3480 rpm it's that cool. Right now I have one board at 50 the other at 47. Gauging from other people's status screenshots, that's extra ordinarily good so heat should not be the problem. Also HW errors are almost non-existant, if I'm lucky I may get a few over a day. Put it this way if I had to run the miner at 250 Mhz to make it work I'd want my money back or ask for $150 for the reduced hashrate. I paid for a minimum of 1.15 Ths miner not a 800 Ghs miner. I was hoping I could get a lot more out of it. All my S3+'s run at 500 Ghs no problem. Like I said I have it on the other port on the controller board, let's see what happens. I'm using the stock fan as noise is not a problem where the miner sits.

I had an S3+ do the same thing and what fixed it was rolling back the firmware 1 version. Ironically the board with the coolest temp is the one that cuts out. I'll check which one cuts out via the hashing light. But as I said a reboot ALWAYS fixes it. Could be I just got a lemon or I have another theory which I'm testing now and will post the results as other people are having the same problem.

Does the red light on one of the hash boards also go out?

I've replaced the stock fan with some quieter but less powerful 2-wire fans from old PSUs. I had the same issue of a hash board going offline until I dropped the clock to 256.25M and it seemed happy at 50/55 deg. My guess is that spikes in heat trigger the board to go offline. BW is also suggesting heat as the cause.

If it does turn out to be a problem with your miner chances are it's a few chips that don't have enough thermal paste. Try dropping the clock a few notches and see if the lower temperatures resolve the problem.

All are good. Power supply is overkill. I have cold air from outside pouring on the unit the temps never reach above 52 celcius. One board runs at 48 celcius. Tried btcguild and it does the same stuff, few hours(random) then halves the hash rates. Other people are having this problem as well so it's not an isolated incident. I'll change the ribbon cables to the other plug positions on the controller board for fun just to see if that helps. I'll let you know how it goes. Possible bad batch? As a computer tech I'm guessing the latest firmware may be buggy for some of these batches. I have the latest batch which I think is 5. Reboot fixes it every time which leads me to believe it's software related. If it was hardware it wouldn't be fixed with a reboot. In addition the boards always say "OOOOOOO" not 1 "XXX". Like I said I'll try the other cable position on the controller board and see what happens. Hardware errors are almost non-existant.

I will say this. When it goes tits up it seems the left most hand side of the units board goes cold so maybe I have a bad board? Software report nothing out of the ordinary, nor a big drop in temperature it's all normal, just the hashing rate is halved until rebooted. I'm trying the other controller plug now and it's running fine so we'll see what happens. Maybe the ribbon cable just needed re-seating. If it is dead do you just send me a replacement board then I send the bad one back? Clock is stock, maybe I'll experiment with over/underclocking if it doesn't fix it.

This this is the case, please check the room ambient temperature, firmware version, PCI-E power to all 4 power port.

Ok but my 3 S3+'s and my 2 sp20's were working perfectly fine. But just to rule anything out I'll try another pool for fun. Put it this way your pool is fine now and it went a total of 52 mins before halving the hashrate. I just turned it off completely for a few mins and powered it up. Let's see if that helps. Longest I've seen it run is about 2 days before doing the hashrate nosedive of death. It doesn't cut out, just halves the hashrate. And I doubt it's hardware because after 2 days of running fine I had only 4 HW errors. Next time it dies I'll let it run for a bit and see if it recovers. Thanks for your help.

Antpool was down on and off for the last 36 hours.

If you can to rule out the pool node hanging, please try KANO.IS or BTCGUILD.COM and see how it goes for at least 4-5 days.

Thanks!

Yours, antpool. I have tried a few but I've only got it up for a week or so. I'll switch to btcguild for a few days for fun. Nicehash works too with extronance no problem but it's been paying out so shitty lately I just parked it on antpool. I'll switch to btcguild tonight for fun and see what happens. It seems to be a few of these puppies doing this. I'm guessing firmware is the culprit as one of my S3+'s did this until I downgraded the firmware.

temps are good 52 and 47 degrees so it's not overheating. In fact I've never seen it go beyond 52 degrees even overclocked at 375.

What pool do you use?

I have this too. Sometimes runs days, sometimes half an hour. No XXXXXXX's or anything it's like one of the boards just doesn't want to hash, however the temps are the same. I had an S3+ that did this and it was so long ago I can't  remember what I did to fix it but I believe I downgraded the firmware. THis unit did it from day one at stock speeds so I downloaded the latest firmware and it obviously isn't fixed yet. Reboot fixes it everytime.

3 pin fan controller,using 5 volt only:

http://www.amazon.com/Manual-Variable-Speed-controller-connector/dp/B002PNFEM4/ref=sr_1_3?ie=UTF8&qid=1430181863&sr=8-3&keywords=4+pin+fan+controller

I think this is 3 pin only,pic is hard to tell,12 volt:

http://www.xoxide.com/sunbeam-rheosmart-pci-fan-controller.html

why dont you just make it?
the red and black go on molex the yellow and blue on the fan header.

but that woudlnt really fix the issue, the fan would still shut down if the board told it to..

what you want is a 2nd computer controlled fan pwm board that has temp sensors.. i dont know if such a thing exists.. im sure you can build one, but i bet it would cost more then its worth.

I am not sure that this is even possible as you described it. S5 controller is what uses PWM feature.
If you power the fan directly from PSU, you are bypassing the controller, hence a constant (maximum) speed.
The only thing that can mimic PWM is a voltage controller, whereas you supply less than 12 v to the fan, like the one below or something similar:
http://www.amazon.com/Zalman-Fan-Speed-Controller-FANMATE-2/dp/B000292DO0/ref=sr_1_4?ie=UTF8&qid=1430179287&sr=8-4&keywords=fan+voltage+controller.