Asicminer Tube - Some minor newbie help required

[k3nnyx]

Hello all!

Please accept my apologies in advance if my mistakes seem pretty noob-ish. My knowledge level is quite rudimentary at this point but im a willing and quick learner! 

I recently find myself the proud owner of 12 Asicminer Tubes.

I've set them up according to the instructions i should be following however it seems that i may have gone wrong somewhere. Would like to seek some advice on where i couldve gone wrong.

How i connected them was as follows.

Power cables - i only used 1 power cable per board, hence 4 power cables per miner. I used the outer-most power port. The one on the right hand side if you are facing the fan of the machine and i used this right hand side port in every instance.


I connected the miner's as follows

Miner 1 - Long cable connected to Block Erupter - UART1
Miner 2 - Long cable connected to Miner 1, Board 3 port (The board on the left hand side if you are facing the miner's fan on the front)
Miner 3 - Long cable connected to Miner 2, Board 3 port
Miner 4 - Long cable connected to Miner 3, Board 3 port

All miners have unique dip-switch identification. No duplicate if im not mistaken but i will double check this.

The miner's all power up properly, and they begin hashing properly however....

Out of the 2 set's ive set up, a total of 8 miners, the results i get is as follows

Set 1 : 1.6 T/H
Set 2 : 2.3 T/H

Clock speed at 300 (overclocked speed)

When i go into test results, i get the following results.

http://s29.postimg.org/uj26gbuur/Screen_Shot_2015_05_20_at_11_07_51_AM.jpg
http://s18.postimg.org/pidpsbx1h/Screen_Shot_2015_05_20_at_11_07_59_AM.jpg

Im given to understand that there should be 16 boards in total. Im not sure if i made a mistaken somewhere along the way.

Any help would be much much appreciated!

Thanks again in advance

[dogie]

1) The PCI-E ports on the boards are not specific, they both connect to the same places.
2) If you have the reach, the closer PCI-E on the cable is preferred to shorten cable length.
3) You have 9 boards on the 1st picture and 11 in the second, so that is 7 and 5 missing boards. That could either be due to non unique dip switches in a chain, or somewhere a board is dead and dropping part of the chain.
4) Go through the boards again and set them up in numerical order rather than just unique, then you can trace exactly which is missing: https://i.imgur.com/RWsscqw.png
5) Set them all at 240mhz just for now so we know its not PSU related
6) You can recheck through setup instructions here, but it looks like you've got everything for now.

Good luck, let us know how it goes.

[sidehack]

He bought these from one of my hosting customers, and they were all tested and ran overnight before packing for ship so they went into the boxes working. I hope there are no dead boards.

Though I didn't think any of them were set on 300MHz; stock is 270MHz for the Tube. Maybe he changed that? Taking them back to stock (or lower as suggested) would be good to rule out some possibilities while getting them going.

Sequential addressing will also help locate problems.

[k3nnyx]

Hello Matt!

Yes, it seems the dip switches were already set correctly according to the diagram that dogie posted. Sorry for wasting time on that. It was already set correctly for 4 miners in a set.

Ive reset the clock to 270mhz. Decided to get greedy yesterday and went to 300mhz.

Ive also rechecked all the connections on the daisy chain wires and powered them back up and lo-and-behold, all 16 boards are now functioning!

Im going to keep them at 270mhz for the moment... 

I did however notice this under statistics.

http://s22.postimg.org/ct897axt9/Screen_Shot_2015_05_20_at_12_56_32_PM.jpg
http://s21.postimg.org/4vgerr06b/Screen_Shot_2015_05_20_at_12_56_11_PM.jpg

On one set of miners, board 6 is showing something different. What does that mean?

Beyond that, it looks like these babies are ready to go. I will put together the 3rd set today.

Thanks again in advance for the help and input!

[k3nnyx]

A quick question...

On the gekkoscience breakout boards, there is a small black knob. Can i understand abit more about what that knob does? Its turned all the way to the right hand side now until maximum.

Thanks again

[sidehack]

Board 6 looks to have a few chips out. If you reset or power-cycle, they might come back up. It could be a heat issue residual from them being overclocked. Glad things are going for you.

The black knob on the boards is fan speed for the PSU fans. If the fans were plugged into the 4-pin headers (near that knob on the boards) you could adjust the speeds up or down. If you're running them in a potentially toasty environment where noise isn't a concern, running the fans at full speed is probably a good idea, but if you want them slower/quieter that's how you do it.

[dogie]

Board 6 looks to have a few chips out. If you reset or power-cycle, they might come back up. It could be a heat issue residual from them being overclocked. Glad things are going for you.

I really doubt its heat damaged from just 300mhz in a normal ambient, they're crazy good with heat. When I first got one of these I tried to fry it with the fan off. Entire board was at 110C but nope, still mining fine for hours.

If we're dropping entire boards when OC'ed and some chips still when at stock it could be a power issue. The boards themselves again should be fine at 300 no problem, so looking towards power. How many tubes / boards per PSU?

[sidehack]

If it's set up the same way it was here, he's got 8 boards per DPS2000 PSU, which ran fine on the shelf for about seven months.

[dogie]

If it's set up the same way it was here, he's got 8 boards per DPS2000 PSU, which ran fine on the shelf for about seven months.

With the nice chunky 16AWG? Power is the only thing I can think of that would cause an otherwise working board to entirely drop rather than just throw out junky shares. At 300 they'd very nearly tip over the 2KW per 8 boards which is close to max capacity but acceptable. I do remember that the Tubes really desired 8 PCI-E even if the cables could take it so maybe its power but on the miner's side?

Could test that theory by putting 16 cables on one PSU and trying it on 8 boards again. If that works then we at least know the cause.

[sidehack]

I had those PSUs running two Prismas each for a couple months without issue, but that was on two cables each. 16AWG, yes, I wouldn't knowingly sell someone smaller wires than that.

Past my bedtime here. k3nnyx, hope you get everything up and going. You're in good hands on this forum.

[Finksy]

A couple things that have happened to me with Tubes that caused boards or chips to drop:

UART cables got pinched when re-arranging miners underneath the aluminum legs, caused all boards to lose communication until cable was replaced/removed.

Capacitors knocked off from man-handling miners

Boards flashed with certain computers/browsers


Yes, these can handle lots of heat, but there have also been pre-mature chip failures due to heat as well (I recall seeing pictures of a chip popped out).  IMO 300 mhz is way too much for these. Even when the difficulty was lower and i added a second pull fan on the rear of the tubes, I kept them no higher than 290 mhz. 2 cables were not needed (16 awg) for that clock speed, even in relatively high ambient temperature (30-35*C)

I still have one tube with a board that absolutely will not connect, for reasons I have not been able to figure out, even though it did months ago.  The tubes were like old carbureted engines in cold weather, you had to know what little tricks work for them to get running, and even then it hasn't always been 100% for me.