I had three identical setups of 5x AM Tubes across 2x DPS2K. On each I put a lightweight 120mm fan (12V 0.5A 3-wire) and a sturdy 120mm fan (12V 3.3A 4-wire) to keep them cool. Two of the setups are running fine on probably around 60% speed on the sturdy fan. The third setup, one PSU kept cutting out even with two sturdy fans at 100% so I junked it and swapped in a fresh PSU, which has been performing beautifully.

Used equipment has varying tolerances on what it can handle, I guess.

Looks like you can set a failover, but it doesn't revert. If your first pool goes down it switches to the second, but it doesn't periodically check the first for activity and switch back if it comes online.

I can confirm they work on slush (bitcoin.cz) and mmpool (mmpool.org) as well. I'll see about posting our experimental slush proxy mod (which is still in progress) that I've been using to mine on ozcoin successfully for the last 24 hours.

I was having issues earlier connecting to slush with one, but it's working beautifully through the modified ckpool proxy. Might try again later and see what was up.

Yeah hit the back button and refresh the page manually. Flashing takes maybe 10 seconds.

I'm seeing 7 of 7 good controllers and about 72 of 76 good boards so far.

Open for all miners, or just on the test port?

You're not kidding about the mad scientist, that's his actual job title. He was already coding on it when I came in to work this morning. The proxy was already fixed for slush-based stratum implementations, but eligius and its ilk are still having trouble. I think he was packet-snooping BFG outputs for comparison to see what formatting was changed or repackaged for valid share submissions.

I think warranty is done through the reseller, but I could be wrong. So far I have identified one board for sure to be replaced, one that's behaving erratically and one entire Tube that's nonfunctional but it's probably one board hosing up the serial line. Tomorrow I should have a final count on what does and doesn't work.

I don't really care what Friedcat says, the controller is labeled with 5VDC and I've been running it on 5VDC for two days.
It's got an AOZ1021AI switching controller inside it (same as on the Blades and Cubes for providing 3.3V) with a 35V electrolytic input cap so it's likely pretty flexible. Might run just fine on 12V, I know the chip is good for that high an input voltage so long as all the input parts are. Also good to see what looks like a series input diode for reverse voltage protection in case you wire up your barrel connector backward. That's good thinking.

Cool. I wonder if slush would want to talk to the guy I've got working on it? We've got a slush proxy modified to truncate the extranonce2 length to pool-requested size, but haven't yet determined what issues there are with non-slush pool compatibility.

I've actually only assembled 19 of the 31 so far, been also busy running new circuits and troubleshooting the pool compatibility stuff. I'll start the rest tonight and finish up tomorrow though. Saturday will be a much-needed day off, I think I've worked every day so far this month.

Set each board's DIP switches to a unique value. What could be happening is every board is addressed the same, so when the controller polls the boards it's getting collisions like mad and errors into a reset. The easiest thing with 4 boards on one controller is flip a different switch on each board.

The ethernet controller is the only 5V device. The entire rest of the Tube is 12V.

5VDC, not sure on current but mine isn't asking for more than an amp I think. The barrel is hot-center.

So, a report on pool compatibility issues. I haven't tested bfgminer yet, but ckpool proxy has been working perfectly on ozcoin for a while. ckolivas looked at the issues I was seeing between his proxy and brickpools pools apparently based off Luke-Jr software? He looked it over and looked it over and threw up his hands in disgust after a couple hours. It appears the things are basically hard-coded to ghash.io nonstandard header format, which means that, basically, if ghash changed their header structure these things wouldn't mine on a single pool natively. Which is pretty retarded.

Anyway, ckolivas gave us some ideas but had to get back to working on other stuff so we're gonna try and isolate the format issues and fork a slush's proxy implementation to make these stupid things work with any pool. If we can do it (Novak's already got one working on ozcoin, but that's the easy fix) we should be able to release a Windows build for people to use which will make it quite a bit easier for lazy and/or underteched people to not be stuck with ghash.

If this whole thing becomes moot and ASICMiner releases a firmware update to make them not kick stratum testicles in an aggressively impolite manner, I'll be both shocked and pleased. But in the (probably likely) chance they don't, hopefully we'll make an efficient workable solution anyway.

I'd consider putting up an assembly guide and ckpool instructions but I'm still sorta waist-deep in setup. Maybe in a couple days.



Thought about saying balls-deep but then people would think I was stickin' it in their Tubes and that's not a customer service we provide.

Did you do the firmware update?

8 Tubes? Each Tube is given a 5-bit address, which covers 0-31 so 32 blades at 4 per Tube makes 8 Tubes.

Well, it is paid per KW not per device, so the only real changes on my end are worrying how much of the buffer is being eaten up on my power allocation. I'm gonna have to take out and move all of my own hardware to free up the difference.

As for picking up and everything, can you imagine how bad it'd be if I waited to get everything shipped? People wouldn't be mining until probably Saturday instead of all the machines going up today. And being able to work with other pools besides ghash.io, aside from being a no-brainer basic operating requirement, is also a specific requirement for my hosting facility since I can (and would like to) block ghash.io because they really don't need any more business. I'm seeing to it the pool compatibility issue is solved not just for my own needs, but for the good of the community (and the network) as a whole. Decentralization is not always the easiest path, but it is the essential path.