Topic: ANTMINER S2 Discussion and Support Thread - page 144. (Read 355730 times)

I am turning mine off with the powerdown command via CLI. Tested it 4 times and it comes back up with no issues for me. I have a B2 S2.

The stock gold psu is awful
Mine dead after 1 day

I've dropped about 3 post about OC-ing. Nobody responded....

At 225Mhz more then 4% HW.

I couldn't figure out how to OC these... I think there is a bottleneck on the voltages to the chips.

If you take the power away from a psu under full load the psu fan will stop instantly.
This way you overheat the psu so you will have to wait for it to cool down.

Any one testing these yet overclocked?

Am just starting a 12 hour test @ 200 chip frequency up from 196.

Initially it appears to made very little difference. about 30-50GHZ increase over my first hour, maybe a 2% increase in HW errors. Temperature OK. Power supply OK. I'm thinking to overclock these effectively we are going to need to increase chip voltage a little, increase chip frequency much more and thus another power supply to take the extra load?

Anyone dabbling with this yet?

 

Several people seem to be concluding that their power supplies are "dead" or malfunctioning based on them refusing to turn on for several minutes after turning the S2 off.  FYI, all three of my batch 1 S2's act this way - when you turn one of them off, you have to wait at least 10 minutes (haven't timed it, but it's more than 5 minutes and less than 20 minutes...) before they will agree to turn back on again.  disconnecting power cord, swearing, etc have all proven ineffective. 

I don't think this is a malfunction - i think it must be a design "feature" of this particular power supply.  Or perhaps it's the way they are powering off... typically, you would not shut off a server by flipping the switch on it's power supply.  you'd issue a shutdown command in the OS and have it gracefully shut the system down.  I don't think that is possible with the S2 as the ATX motherboard cable is not utilized.  it's really unfortunate that the BBB board was not powered by the ATX motherboard cable - that would have gotten rid of the "paperclip" mod and perhaps made it possible to gracefully power on and off with a soft button the front of the case. 

But the way it's designed, when you flip the power supply off while the case is open, i see the system power off then, after about .5 sec, it powers on for just a fraction of a second.  a little blip. 

Perhaps this blip is causing two problems - corrupts the SD card, and trips something in the power supply that stops it from powering back on for several minutes.

I'm very interested in hearing if replacing the power supply solves the corrupt SD card issue.  that would be great!

Guys start putting technical discussion in here. This thread was never really meant for it and findings don't get amalgamated into the OP.

user: root
pass: root

Which is the default username and password?

Root?
Root?

so running with 1 blade less works fine?

S2 is nothing compared to S1 .

No problems for me. Setting up couple of S2 batch 2's was pretty much plug and play. Now hashing at around 1025 GH/s each in my living room. The noise of the fans speeding up and slowing down is quite annoying. I will move them to a DC this weekend.

Sorry you guys have so many problems with the S2. S1 looks like a breeze compared to this.

Any info on batch 3?

my avg hashing speed at poll is 930 gh

why?

what is the avg?

Edgar, Batch 3??

And seems like you are fighting a bad PSU, and maybe a bad blade or slot. I would test the blade in another slot just to be sure (and maybe opposite of BBB) , or in another good S2 if possible.

And I think that with the 9 blades you will eventually settle down to around 900GH/s. And keeping the case lid on does seem to allow the fans to cool the unit better.

So it doesn't work with 1l1l1l1l1l's SD image (unmodified)... that worries me.  That image worked for me (and obviously 1l1l1l1l1l).  Must be a piece of faulty hardware because we know that software image works.

A couple of shots in the dark:
1) Try running 1l1l1l1l1l's SD image on another new SD card (just to rule out the chance of a faulty new SD card).
2) Try running it with just one hashing card, to see if you can get anything hashing at all.  If that works, build it up slowly one card at a time, testing each time that each additional card works.  If one card doesn't work, try a different card and a different slot until you can get one card hashing.  If you can't get even one card to hash then something central must be the issue: BBB, daughter-board that BBB sits on, backplane at the bottom...

Found my culprit with the faulty antminer.
Seems one of my slots doesn't report any temperatures when a module is running in it.
This makes the fans go haywire and probably trip a security thing in the antminer it self or the PSU

  This probably won't help but, I lived overseas and remember that not only did most electronics have a 110/220 volt selector but some had an internal 50/60 hrz cycle switch. Things like clockradio's would do weird things if the cycle switch was in the wrong position but they would still try to work. I doubt the PSU's have this 50/60 setting but it could explain some of the issues if the supplier screwed up.

looks like Bitmain should've listened to all the numerous shouts made here about properly sizing the psu, or atleast gave the option of no psu and/or better ventilated the unit.

For those looking to turn off their units gracefully, use the powerdown command from the CLI, ive tested it a few times and it seems to work.