Topic: BITMAIN S9 BURNT TRACE ON BOARD (Read 192 times)

all that and you totally missed my point or the reason why I said anything I said. So, I am just not going to reply any more. You have your opinion and I have mine - nuff said.

So your issues are with Slushpool which is separate to Braiins OS. I doubt a majority of slushpool users even know about bOS, people using it are a minority regardless of pool. There are a few advantages but most important to break from Bitmain control.

Try to discuss Slushpool things in the pools section, just because they are the same people doesn't mean one thing equals the other. Even the Stratum v2 proposal is separate, it is an optional feature i would like to see in action, Judging a priory is unscientific at best. If it doesn't work no one will use it, it doesn't change both bOS and bosminer, or slushpool for that matter.

I'm glad they brought us this (free and open source) firmware AND a cgminer replacement. bOS and their pool are separate things, funny you already think it will "screw their own pool" without waiting to see it in action. More importantly is how it would perform with other pools that decide to implement v2, as those that don't would work with v1 fine. I don't understand how it even occurs to you that this somehow might affect anything beyond the miner and the pool it connects to, at best it will just improve latency from farm to pool, then its up to the pool how fast it can broadcast and this is already outside of what this covers.

They probably identified a bottleneck and are trying to improve things. Rather than sit and brag all day how others always do things wrong, they try, they code and they deliver. The other pool operators are way too busy collecting fees to even bother writing software anymore...

---

Yes i have written about it in the bOS thread. To tune you have to start lowering the voltage until it stops hashing properly. This will get you the optimal per hashboard. Then you can try higher or lower speeds, each time you have to find again the best (lowest) voltage value. Underclock tends to lower temperatures and increase efficiency, you would do well to measure power usage. This takes time and patience. Overclock requires more voltage and thermal dissipation.

But it is OK to use the default settings, just not optimal. I have an acquaintance that likes to default to 8.7v and then work from there.

yea i like brains os, i really like the  constant updated GUI

with ant miner stuff you always have to refresh the browser to see the actual stats updated

and ssh is nice and handy
 
Artemis3

is it ok to just run it stock cgminer settings, using the freq scale option,

i do use the freq scale setting,

is there a basic how to tune with brains .

like what you were saying , un check the voltage setting and lower it
what would one be looking for,  lower temps ? no hw errors obviously , not sure what to exactly look for
i know when seeing lots of hw errors in over clocking brains bringing up the voltage sometimes helps   <-----however i am a bit clueless in this

Thanks FM

also i run brains on all my s9's it does work and get paid , seems to run 1200 watt range on stock board freq settings

I didnt say that they dont find blocks - I simply said I find it interesting that around they are pushing braiins os as a better firmware yet their pool is not doing so good - I am sure a large bit of that hash rate is theirs and is running their fw - and when they are on PPS or PPLNS pools it really doesn't matter if they hit they hit the block or not - all that matters is that the pool does. I wish all pools were transparent enough to show who actually hit the block. That is why I said I would like to know - I never said I knew anything definitively.

my issue with slush is a much simpler one - they are ripping off every miner plain and simple. They have gone almost 2 years without paying the merged mining and if I recall when they stopped NMC was at around $10-$13 a coin - almost a year ago they said it would be paid within a few weeks, still has not happened. They are liars and thieves, plain and simple and I have not used them since.

I dont know enough to comment on their "stratum 2.0" I just hope it doesn't make it worse for miners and hopefully only screws with their own pool and not the entire network.

It is a bit late, but when you install BraiinsOS, the first thing you should do is uncheck the voltage setting and start lowering until you find the optimal value. BraiinsOS is manual, there is no self tuning, and those presets are not optimal. But maybe it was too late and the board was already damaged...


BraiinsOS is one thing, using Slushpool is another. First I don't think that many bOS users use Slushpool, it is not even set as default, unlike Bitmain's that is set to Antpool nor there is a hidden dev fee like with other closed source firmware mods. There are large miners running bOS, with large bills, saying it doesn't find blocks is silly to say the least. But those miners like Chinese pools, what can i say?, Convince the kanopool users to use bOS when kano doesn't even like it, so he can tell you if they actually find blocks or not? You would need to ask Viabtc and the others to know if they really do find blocks or not.., or convince a large miner to temporarily switch pools and conduct the experiment, but they can't "risk" switching to a PPLNS pool for a month or so, catch 22... But, its actual cgminer in there, check the code...

Speaking of that, they are attempting to improve on stratum, once they finish replacing cgminer with theirs, you can then test if their idea is worth or not, either directly against their pool, or via (their) proxy against classic stratum pools. One thing they seem to be working with is the idea of a more compact, (proxyfied?) data stream, perhaps helping reduce the latency to large miners. These things will become more obvious next year when they release bosminer officially.

While some people keep whining, at least the Slush guys are coding and sharing. Less words and more action.

there are 2 boards with this trace exposed , either doped in shipping and when powered up they connected together via the case and somewhere on the board there is either a broke trace or failed part , just dont know where to start, do i try flashing it with serial device , or connecting it and see if it responds to serial device ,


when powered up without controller i see the red light flicker and light up , no heat or other,

but i have seen some posts about flashing it , its been a  while since i was flashing epproms of speed controllers for my raido controlled multi rotor drones,

That is a weird spot to have a burn, that should be the +12 and ground backplanes. I wonder if the chip next to it shorted out, burned the board, and then opened up.

C

the only flashing you can do - as far as I know is to the controller - by "some life" do you mean some chips work and some do not? are some showing as 'X' or not as as all instead of as an '0' in the ui?

it seems as the broken hash board has some life , but im not sure if i need to flash it with something  ?

well my opinion on Braiins is this - slush promotes it, if slush users are starting to use it why is slush luck so bad? their firmware is so great and better should they not be doing better?

i would like to see if any Braiins users have actually found a block. most pools wont tell you which of their users found the block though - even though they could do so very easily. Also there are pools that will ban users for using non-official firmware, also not worth the risk in my opinion.

as for the stock firmware - it works, there is no stealing of hash rate or dev fee with manufacturer fw - yes they do block ssh and yea that pisses me off too - i used to use awesome miner to do a lot more with my miners - now I have to be on site.

if ssh is your main reason, then i understand.

no braiins ?  im running it on 11 s9's    love it ,   have several cgminer files and use cron tabs to call them up and restart cg miner,  starting and stopping cgminer by cron tab ( set day time )  and also switches the conf out for the heat on certain days  i.e  from 12:45 goes to 75% under clocked and up to 1.03 at 5pm and up to 1.10 at 11 pm to 6 am

i under clock for heat & time of day, and over clock once sun is down and things cool off , and further over clock at near midnight

have herd lots of bad things about the stock firmware,

possible hash rate stolen - not sure if its true but like a hidden dev fee

not secure - able to be hacked into

No ssh - this is crucial for me as i scp in and change cgminer files and add different configs , and cron tab info

ok yea that should not be connected to anything - but getting knocked around in shipping may have caused it to make contact and short out. i would recommend putting original fw - braiins os is not worth using in my opinion.

it is furthest away from the pci connectors on the (adjacent corner) and on the side opposite of the surface mounted pci female connectors,

its not supposed to touch the ground, i feel that when it was shipped from NYC to CALI it got dropped several times causing the hash boards to come into contact with the case,

as 2 boards connected to the case in the same spot , i feel as if the boards were shorted together, but it wasn't a magic smoke moment ,  i do recall smelling faintly the smell of

overworked electronics / possible overheating , but caused by these points becoming shorted between two boards , im no expert but do have some experience in circuits, seems like 9-12 volts shorted through the boards and the smallest trace acted as a fuse, or messed up some transistor or step/down/up ect

the other 2 boards work great , even have 668 factory freq  in fact all of the boards were above the 630 freq standard,

just not sure if the board logic side got nuked or flashing the hex/eprom stuff might bring it back to life , it does have the red light come on but when it trys to start

cg miner in braiins os cant start hardware 

that area in the first two pics - is that the corner furthest away from where the 6pin connectors are? looks like some sort of grounding connection. but I do not recall anything connecting to the hashboards outside of the psu and the wires to the controller.

so just bought a s9 from Fleebay and it worked for about 30 min before smelling like burnt electronics,
using ibm server core power supply on 220 it does 236+ amp to the 12 volt rail, so i doubt the power supply is the problem
flashed braiins os , and set it to default settings , all looked good, checked it 30 min later and the middle board was missing
thought i smelled burnt electronics , could be wrong as it was subtle.

so i have found only this damage



these are 2 diffrent boards, and the 3rd board seems to have different traces in that location

the one with less damage still hashes , but the one i scraped off the burt coating will not hash

and will not reconise the board at all

any help with this would be nice

i see in the case the boards must have grounded out