Topic: DRILLBIT SYSTEM Miners - Avalon range available - Now $104.99! - page 30. (Read 273879 times)

I just had one die on me and I am thinking there is a power issue. Are there any micro fuses, resistors or caps that I should be checking with my volt meter?

I worry the miniplane may have been to blame in some way. I saw very inconsistent performance and errors with it. I also know that molex power can backfeed through the miniplane. This is somewhat troubling to me, as I have seen this cause issue in other products. I would highly recommend against using the miniplanes at this time, from personal experience. I await the developer's analysis of the situation but, I would not use the miniplane and molex pins at this time. You can make a stack using only three of the holes on the boards and the included nylon risers, as I have done, to temporarily run without the miniplane and have them all stacked neatly.

Do you happen to have a 'history' output from the different --drillbit-options you attempted? That could be very useful, particularly the last working settings.

All four boards were plugged into a miniplane and powered by PCI 6-pin connectors from a Corsair RM550 power supply (mfgr. data sheet with specs - http://www.corsair.com/en/media/cms/manual/corsair-psu-spec-table-091813.pdf). No other devices were connected to the PSU. The miniplane was firmly connected to a Spotswood open style case and a 120mm fan was blowing directly across the heatsinks, the fan was powered by the on-board connector. Each board showed green when initially powered on, all hashed without immediate error when cgminer_drillbit.exe was run without any --drillbit-options.

Each board continued to respond, and hashed without error for approximately five to ten minutes each before being stopped, when individually started with the --usb command option to determine which USB device ID was an Eight and which was a Thumb. After all USB device IDs were tried, an instance of cgminer_drillbit was started with the --usb option so only the Eight were hashing. After approximately ten minutes, the hash rate was lower and more variable than I had hoped for, so I was going to start testing each board individually with a variety of --drillbit-options to find the optimal settings for each board.

At this point two of the four boards failed to respond when tried with or without --drillbit-options and "cgminer_drillbit.exe -n" failed to list two of the boards. On investigating physically, two of the four (the nonresponsive two) were found to have no green LED on.

I then shut everything down, removed the boards from the miniplane, plugged them into 4-pin molex connectors, and powered everything on. The two LED-less boards remained green ghosts. I then returned everything to the miniplane, powered it on, and have been consistently and stably hashing on the two responsive boards (at ~19gh/s per board with default settings) for the past 16 or so hours.

Sounds like you may have ran the board at suboptimal power by having the other computer components attached. Those could easily draw enough power to have a problem for the boards. I suggest listening to nexus99 and detaching everything from the PSU and trying again. These boards seem to really like having clean power, based on my testing. I got MUCH better results after buying a new powersupply to replace the aging 400watt PSU I was using. I went from 55gh/s over four boards to 70gh/s.

I'm waiting to hear back from Barntech's team in another thread where I asked for the various points they recommend for testing voltages across the boards. I feel that we may learn quite a bit from collective testing.

Thank you for posting such detailed information about your setup. It is very helpful for those of us trying to get an overall opinion of what is causing board failures.

I would pull everything off the PSU except the MB and 1 Eight. Use a working one so you know the PSU and cable is good... then turn everything off and replace the good one with one having problems.. If no green light then you know for sure its the Eight having a problem.

550 watt psu,no backplane used,used 2-4pin molex lines-2 per line(each molex cable had 3 connectors,used 2).

No other devices on PSU except mobo/cpu(45 watt dual core)/hdd/dvd rom & NVidia 8100 non powered vid card(no external connectors).

No OC'in was done,ran @ stock CGminer settings,got CGminer from drillbit forum:

https://github.com/drillbitsystems/cgminer/releases/tag/3.8.1_drillbit

No green light is the problem on 2 boards.

I miss anything 

This sounds like really really good advice right now.

I read about the miniplane issue that others reported and only two of my boards were powered by the miniplane.  The others were stand alone with the Molex connectors and the 'standoffs' for support.

Everyone's boards seem to be dying left and right.  Some run for minutes, hours and possibly a day.

It might be better if people with boards working longer than 24 hours were to post their details.   

Myself, I was using the power supply from a Dell T5400 (876w, +12V  at 18 amps), Dell T7400 (980w, 12v @ 18 amps), an old RadMax (450w, +12v @21 amps).

One of the boards was being powered off the T7400 (6pin connector), 4 boards off the RadMax (4 molex connectors on two rails), and one board off the T5400 (molex connector).

Strangely enough, the power supply in the T5400 appears to have died over the last two hours.  I think that explains why I was loosing video on several occasions yesterday.  That T5400 ran fine for years and had no problems with a 7970 graphics card for months.  I have not been running any bitcoin mining on the GPU in a couple of weeks.  Maybe it just got tired and died.

What more 'useful' information would you like to see posted?

We need a comprehensive report of the states in which people's boards are going bad. There are a ton of reports of bad boards, but little useable information.

I think that it would be helpful to Barntech's team if we could list how boards were hooked up, i.e. miniplane or not, pci-e power or not, etc. I would also think that powersupply type, with as many details as possible would be helpful. Reports of failed boards don't tell anyone much, other than failure rates.

Are you using a mini-plane for power? Or, are they stand alone? I am debating if I should even try connecting them again until there is a fix. 

Is it mostly with people with multiple 8 boards or single boards as well? I am supposed to be receiving mine on Monday and not sure if I want to plug it in unless there is a fix like a firmware update. I have seen issues where my ASIC Eruptors didn't work on certain computers I tested but worked fine on my main one. Hoping this gets resolved without replacing hardware.

During the last two hours I lost another of my Eight boards so I'm down to 3 out of 8 working now.

Well,I'm a sad panda 

2 out of 4 are "no green light" specials after 12 hours 

Hope I can fix em or trade em for working ones.I'll wait for info before trying any "fix".

I have faith Barntech will work this out or find a fix 

My single Drillbit 8 board runs also with 2 fans.  It had no issues running overnight with int:52:2:950 settings (22.6 GH/sec average low R & HW rates, WU:316).  The heatsink is warm to the touch, but not too hot.  It has been running continuously for about 16 hours since the last config change.

Cheers

w00t, the mailman was good to me today, got my four thumbs a few minutes ago (had to sign for the package, make sure you are home if you haven't gotten them yet!).  Everything was well packed - lots of bubble wrap!  Now I'm reflashing my Pi with the forum version of MinePeon.  Will post results once I get everything up and running again.

Ok, thats obviously worrying. Anyways thanks for the info!
I think I'll get a fan anyways, it doesn't hurt to be careful with temps.

Still no reply to the PM I sent to Barntech on 11-15.  Should I use the Drillbit forums?

Mine were run stock with 2 fans on them. 

It's not a heat issue. I had all four of my boards running in a cold ambient temperature room with a 120mm fan blowing directly across the heatsinks. The heatsinks were barely perceptibly warm to the touch, and two of my four boards went "green ghost" while running on default settings (i.e. no --drillbit-options specified at all). The remaining two are running on default settings and have been stable now for about 12 hours and the heatsinks are still barely perceptibly warm to the touch.

It's not heat. It's faulty hardware or serious software bugs.

So is there a heat issue with these boards?
Did you guys overclock, and did any of you use a fan?