Topic: Bitmain's Released Antminer S9, World's First 16nm Miner Ready to Order - page 287. (Read 531168 times)

they use daisy chain i think

No magic here.  You need 10 PCIe power plugs to power up the each miner.  The controller does not draw as much power and the asic boards, I suppose people are "cheating" and using a splitter on one of their plugs, but you risk overloading that plug doing so.  The combined 3000 watts of (3) 1000 watt power supplies is barely adequate as is - you will be running those supplies at max.

rm1000 have 6 PCIe too and they using 3 here to power 2 miner
how they do that ?

Yet another power supply.  You need 18 PCIe just for the boards.  Note that this will likely need to be spread across 3 power circuits, unless you have 30amp 220V available.

how to use 3 PSU EVGA 1000GQ 80+ GOLD to power 2 antminer S9
each EVGA have only 6 PCIe ( power 2 boards )
how to power the controller on both miner ?

when you buy one they send an email to you so
find the tx id.  and reply in an email to them.
see below

People need to stop running their S5 and S7 miners as soon as possible, they're a fire hazard. This summer we will hear lots of stories of people's houses being burnt down because their miners burnt starting fire.

Hi,

I made an order for two Antminer S9 B5 than I made a direct bitcoin payment but almost 72 hours after that transaction is stil unconfirmed. Bitmain said "The payment is on the way.
We don't receive it yet. " Did anybody seen something like that?

Let it dry for a day or two.

Is it hashing and would the soap and detergents corrode the board just dont keep doing it. POOF!!!!

damn, now atoms comes out. where are the aliens & flying saucers ?

B5 still stuck at clearance agency at PRC ... grrrrr

btw, i took a risk & "cleaned" my boards !

https://youtu.be/OajjZMC5vKo

Enjoy

Every time you insert and remove the plug from its socket, the internal metal contacts get looser and looser until resistance builds because of poor contact. That will then produce micro arcing ( you really wont see it at first). Thermally the connection builds up heat, melting and or starting a fire.

Older power supplies are notorious for this because the molex connections have been inserted and disconnected many times, Especially so if you use the molex connectors to shut down, turn on or reset your miners.

Invest in a non-contact thermometer and ALWAYS keep an eye on the connector temps!  If you find one that is warm, REPLACE IT!

Physics disagrees with you, sir.

Thats why you got a burnt up miner. if there is not enough ac power at the input of the PSU you will not get enough DC output power to the miners and when this happens electrons flow through the wire really fast bouncing off atoms causing friction making the wire get hotter and hotter until meltdown like that miner you have.



 

You might want to consider swapping the data cables around and see if you possibly have a bad data port on the controller. While I haven't heard much about that class of problem (i.e. data port/cable), it's possible for them to happen. You might also want to try swapping the PCIe power cables between boards, although the PSU swap should have covered that.

What chains do you have identify on the display? You have mentioned Chain 1 & 4. What's the third one called? Do your identifiers match something on the control board? Do you get any readings (e.g. temp) from the 3rd board?

None of what I have said makes much sense if it was working OK, and then just quit.

Just something to consider......

That might be accurate for some miners, but in the case of the S7's they have 3x PCIe connectors per blade (which I was using all 9 + 1 for the controller board) and each blade was drawing 375W DC (125W per PCIe jack) according to B1 specs.  Even if 1 of the PCIe connectors completely lost connection, that would split the load up between the other 2 and resulting DC load would still only be 187.5W per PCIe connector, or 5.2A per 16awg 12V+ circuit.  Well within the spec and capability of these cables, both in theory/rating and in practice.  

Therefore, unless 2 full PCIe connectors from each of the 3 blades developed loose connections simultaneously, I don't see this as a possibility.  Also, looking at the burnt PCIe jacks/cables, none appeared to have been under more thermal load than the others when comparing eachother within a blade.  

The fact that the boards appear to have melted from the middle outwards, suggests to me that it may have been some kind of controller-related failure causing all of them to over-heat and burn, similar to an internet shortage.  One would think that if there was a fan failure or something else non-controller-related that caused the temperature of the chips to rise to that critical of a temperature, the over-temperature would have kicked in and shut the miner down.  Unfortunately I have no information on any of that, and was told to send the controller along with the boards to Bitmain.  They did not give me any explanation other than "Sorry ,burnt board are not cover in warranty ."

More pictures and details in this thread: https://bitcointalksearch.org/topic/beware-bitmain-s7s9-failures-and-lack-of-warranty-1540037

It takes one loose connector. Current initially flows through the other two until they burn out and their resistance increases above that of the first loose connector. Then the loose connector starts taking the load and it quickly burns out too. It happened to me more than once with Bitfury gear.

I know its probably a tier one comment, but I have to ask:

What is line voltage at the back of the wall plug on your units when running full throttle with all gear?
Ive ran into issues with hardware frying because line voltage dropped below 100....   And I had PLENTY of overhead as far as supply wiring and such... it was a failing generator.

But.. it sounds like the hashboard has a failure of some sort.   No physical signs of burning/darkening anywhere?

It's kinda known the BM's warranty is like taking a BM.... drop it in the toilet and flush that garbage away if you expect things to happen 'right now' a failure happens.  Its always a gamble with new products with them; who knows if it will be good out of the box, or else you will need to be without it for usually weeks to a month or more... If they were a product in a store;  that would be different.  Sadly, its their strategy to protect their ass from incompetent people or ones whom dont provide what it needs to operate correctly..

Fixed that for you, otherwise spot on as usual.

As to the comment about AC wall voltage, read my case again and you will see that A) I had 9x S7's and 12x S5's hosted using the same AC power and B) that specific 4000W PSU was powering 1 other S7's and an S5, neither of which suffered from spontaneous self-combustion.

And loose connections? Like, all of them? Lol...

Hey everyone.

I have a Batch 4 S9 (I have 8 S9 units running from batch 3 and 4) that arrived on Friday and ran all weekend perfectly until this morning. Seems an entire blade is down.

Chain 1 and 4 are hashing fine, together at about 8,600, but chain 3 shows all 0's but is not hashing at all.

I have tried multiple hard and soft reboots, resetting the frequency to 575 and 550, and finally I swapped out the EVGA 1600 PSU with my spare 1600 PSU unit new out of the box and nothing seems to want to make that board come back up. Is there anything else I should try before I report a defective board to BitMain under warranty?

Thanks!

Bob

How exactly do AC source limits cause a problem at the DC load? Seems more likely inadequate AC wall power sent to the PSU would cause an issue with the PSU, either in drawing too much current from mains to supply the DC load requirement or simply turning off.

Also, I think his complaint is as much "Bitmain will deny warranty for anything they build which burns down your house due to poor manufacturing" as much as "Bitmain manufactures things that will burn down your house". Accidents happen and sometimes electronics fail, but that's kinda exactly what warranties are for.