I was thinking a system more similar to the asicminer system, but i beleive that involves a sealed tank, and still requires either circulating the liquid through a radiator or having a peltier cooler for the gas state

I updated the firmware, and there is no change to the hardware error rate. (Bitmain, is there a way to see which board or chip(s) are producing errors?)

stat                 unit 1                     unit 2
GH/s avg           200                        204
rejected            7                            8
HW                   23,500                    3,800
DiffR                 896                        1,024

also, i tried the 425Mhz code, and hashrate seemed to drop to around 187GH with more errors, so i guess temperature is a factor

anyone have immersion cooling experience? Looking at the success of asicminer's efforts, it seems as though immersion cooling of antminer boards would not be difficult. A 12"x12"x10" tank could easily house 8-12 boards.

Maybe this will sound a bit naive, but would it be excessive to offer the V1.2 cards for cheaper, simply as a 'thank you' to everyone who bought an august starter kit and then had virtually no opportunities to add cards to it? (october preorders or a small pile of seemingly overpriced cards in november)

There are a lot of better options available, and to see these prices SUCKS

wow thats a massive aluminum heatsink lol. I was thinking along the lines of immersion cooling (single or dual phase, though single phase would obviously be cheaper and simpler to build and seal)

if it would work with the heatsinks removed, you could put 4 antminer blades (total 400GH in aircooling, ~900W) into a tank as small as [12"L x 6"W x 8"H] and have space for liquid to circulate. With the heatsinks, the width would be closer to 12".

they really are. I am strongly considering replacing the other stock PWM fan with one. They look AWESOME and are very quiet, but unfortunately have lower airflow. (box says around 75CFM, while the stock fans must push over 100CFM).

BTW: anyone have good liquid cooling ideas or methods? It would be pretty sweet if the antminer boards could be placed into immersion cooling, but i know that they would require a radiator capable of >400W for each antminer which i imagine is costly and semi-complex to install

Here is my rig:
1x Antminer V1 with the secondary board being a V2 replacement (bitmain absolutely nailed it on the RMA process and got me a replacement within only a few business days)
1x Antminer V2
1x Bitfury August Starter Kit w/ 1 H-board
1x 730W modular RAIDMAX
1x 600W Modular Corsair

Total hashrate of about 435GHash/s. The units are on a small table underneath another table. The antminer with an orange fan is to replace the obnoxiously loud fan that came on the unit. the second antminer did not suffer the same noise issues

you could use that 8K to buy two antminer systems = 400GHash and the power consmption is only a little over 2W/GH

first, follow the instructions for increasing to 400Mhz (200GHash) since your temps are surprisingly low

your error rates look okay, about the same as my newer unit is performing

the pillow is a cooling pillow. look it up, one that size is capable of dissipating about 700W, which is more then enough for a jupiter with the fans unplugged 

correct. yellow 12V lines go to the red terminals

I will try the update. as of 14.5hrs operation for both units (Unit 1 has a v1 board and the v2 replacement board, unit 2 is v2) shows:

stat                 unit 1                     unit 2
GH/s avg           199                        204
rejected            218                         13
HW                   68,500                     10,800
DiffR                 28,000                     1,700

the rest of the stats were similar enough to be omitted. It seems fairly clear from the results at 14.5hrs of operation that there is roughly a 5GH/s difference in the units, probably a result of the different firmware. Temps on unit 2 are 49/50 while unit 1 is 51/54 - I will update the firmware shortly and may try pushing unit 2 to 425MHz

At the current BTC price of under $1000, these boards make no sense based on the instantaneous usd/btc rate. It would make far more sense to price hardware in BTC, with instantaenous conversion rate to USD (like the opposite of bitpay). These boards will struggle to return more than 0.8BTC each in favorable conditions, and the current price is roughly 1.5BTC

ps: it would be nice to see the v1.2 boards at a slightly lower price, simply as a favor to those that bought starter kits, and had only 1 small opportunity in october to fill them.

I will take a look later, i dont know if theres any different version marked on the PCB, but in my case the units from the first auction had the green power terminal while the unit from sushi's groupbuy had the red/black terminal and unoccupied capacitor pads - see the post above

PPS: looking at my stats so far, it appears the newer unit has a much lower number of stales/24hrs but slightly more low difficulty shares. I havent been monitoring stats very long yet to see if its random or a trend yet, but has anyone else noticed something like this?

Version 1 Unit:                  Version 2 Unit:
13m 8s                               12m 54s
192                                    203
200                                     210
0                                        0
27                                      27
265                                     311
12                                       0
767                                     143
20                                       24
49                                       52
0                                          0
44,333                                 47,171
2,798                                  2,940
33,920                                 39,808
1,536                                   0
0                                         0
417,709                                36, 023

any idea whats causing the difference? I can repost the stats in a few hours once they are larger numbers, but the version1 unit still produces a higher number of HW and rejected

MY BITMAIN RMA EXPERIENCE

About a week ago, I received my first antminer unit from the first-round auction for a reasonable 4.75BTC. it ran well at 180GHash but had an extremely loud fan that reported its speed in the range of 4700-5500. I mentioned this in the forum, and bitmain quickly [and nicely] advised me to buy a replacement PWM fan that they would cover the costs of, as the noise and high speed appeared to be some sort of individual error. I installed a replacement fan with slightly lower CFM and virtually silent, and was quite happy. however, somewhere in this span of time the second board became unresponsive and would not show up or hash.

I tested the board as best i could - using a multimeter checked almost every accessible capacitor or point of voltage, and the issue appears to be within, or related to, the TI TPS53355 DC/DC buck converter, as the first one on the board was not producing any 1.10V output, whereas the other 3 were doing it successfully. Unfortunately, without the first chip working, the entire chain will not operate, and I requested an RMA/replacement as finding a local engineer to determine the issue and if necessary order the $>10 regulator would likely cost more and take nearly as much time as to be shipped a replacement.

BITMAIN shipped a replacement board alongside my second unit (from the group buy) and it arrived on thursday. It did not come attached to a heatsink (I was told it should, and am very happy i had not yet shipped back my bad board+heatsink!), but replacing it was straight-forward with the removal of ~16 screws. On removal, i noticed a slight issue with the application of thermal paste:

here is the backside of the dead board i removed, you can see that the paste was not under sufficient pressure in some areas.


Once the new board was put in place (I reused the thermal paste, but may need to buy a new tube and redo it as the temps are 2-4 degrees higher than the primary board), the unit booted up and returned to hashing perfectly.

The second unit installed like a breeze, with only one small exception - It came with the ip address 192.168.2.049. This may have been an obvious misprint to someone more involved with networks, but in my case it took almost 10 minutes of confusion before i realised it is actually 192.168.2.49 without the extra zero.

Now, both units are running well at 200GHash each (thanks goxed!) and i have installed secondary fans for both to impove airflow, and also put some thick paper over the top where possible to ensure the air flows across the entire length of heatsink. I still have yet to submit receipts to bitmain for promised compensation of the RMA shipping but have  no concerns about them renegging. BITMAIN has done an awesome job of producing a stable miner with a strong frame and it has the least hashrate varience of any of my hardware. Add a dedicated user profile and email contact, and I have absolutely no regrets over my purchase!


PS: the new boards are a little different, namely [better] screw terminals and [inexplicably] unused pads for 4 largish capacitors. heres a picture for those interested:

my impresion was that the discolouration is just that the chips were poorly stamped/printed with the logo, and not an effect of operational heat - but i could be wrong

I noticed that when I removed the secondary board from my auction batch1 unit, the thermal paste is obviously machine-applied, with a dot of paste placed every cm or so. However, it was obvious that only some of this paste was in contact (flattened properly) and a few areas the dots of paste were clearly under little pressure as the paste had not spread out. It may be a good idea for everyone who receives a unit to check the tension of all the hex screws, and if necessary give them an extra twist to ensure full contact to the heatsink

Yes it's the thermal paste, it must've been under pressure and leaked out the screw holes in shipping.

I received my unit as well as a replacement part I requested. Bitmain really pulled through on the issue and I plan to put a full review later.

However, I noticed that when I removed the secondary board from my auction batch1 unit, the thermal paste is obviously machine-applied, with a dot of paste placed every cm or so. However, it was obvious that only some of this paste was in contact (flattened properly) and a few areas the dots of paste were clearly under little pressure as the paste had not spread out. It may be a good idea for everyone who receives a unit to check the tension of all the hex screws, and if necessary give them an extra twist to ensure full contact to the heatsink

check your power cables and fan/cooling status, and reboot. that might help