17 Avalon photos

[Felipeo]

Very nice photos
Thank you for the opportunity to look inside Avalon!

[1715222491]

1715222491

[1715222491]

1715222491

[1715222491]

1715222491

[cypherdoc]

NEMA 5 series is rated for 125V maximum.  I'm sure you *could* do it but I wouldn't, personally.  FWIW, one Avalon should mine enough in a day at the current difficulty/exg. rate to pay for the whole 6-20 setup...  

let me just make sure i understand you correctly.

it seems you're saying that Nema 6-20 power cords (connecting the Avalon to the surge protector) should be used together with the Nema 6-20 surge protector you linked to, correct?

[SebastianJu]

Did anyone try to add little heatsinks to each of the chips and did it lead to better performance somehow? Maybe more stable overclocking?

Dr. Zhang recommend NOT to add any heatsinks on these chips and saying that will reduce the surface airflow for PCB board

I see the point in this but i wonder if you would put it on following the airflow and not blocking it if the effect wouldnt be the opposite. The air could flow free but has more time to take on the hot temperatures.

I asked this already but didnt get an answer... how is the capability of an avalon for being overclocked? I only read someone experiencing a wall at 300(something) but no info if this is a wall by design or if it could be fixed with better cooling. For example, if one would try, watercooling the single chips.
And is it somehow possible to change clockrates, volts and so on? Only theoretical for the moment because i dont have an avalon yet. But i mean... only 10% more hashingpower would have a big impact. And such a number is easily achievable with normal GPUs too. So im wondering if something is possible there. Of course one would have to be ready to lose the miner when playing with such things.

[johnyj]

Did anyone try to add little heatsinks to each of the chips and did it lead to better performance somehow? Maybe more stable overclocking?

Dr. Zhang recommend NOT to add any heatsinks on these chips and saying that will reduce the surface airflow for PCB board

I see the point in this but i wonder if you would put it on following the airflow and not blocking it if the effect wouldnt be the opposite. The air could flow free but has more time to take on the hot temperatures.

I asked this already but didnt get an answer... how is the capability of an avalon for being overclocked? I only read someone experiencing a wall at 300(something) but no info if this is a wall by design or if it could be fixed with better cooling. For example, if one would try, watercooling the single chips.
And is it somehow possible to change clockrates, volts and so on? Only theoretical for the moment because i dont have an avalon yet. But i mean... only 10% more hashingpower would have a big impact. And such a number is easily achievable with normal GPUs too. So im wondering if something is possible there. Of course one would have to be ready to lose the miner when playing with such things.

My past experience from GPU mining tell me: Do not overclock too much, the long term stability of the mining rig is far more important than a small gain in hashing performance, especially if you want one rig to run for a long time (Now we see that BFL is also power hungry, so avalon would at least be around for a year or two)

I had a pair of 5970 overclocked quite high (just 20Mhz below the crash point), undervolted, installed custom cooling device and the best thermal compound I can find (liquid metal), although the GPU temp never goes above 55 degree, all 4 GPUs on both card still died one after another in 1 year. I guess some of the components on the board are just not designed to endure that kind of added load for a sustained long period. For a gaming rig that run once a while it is fine, but for a machine running 24x7 like a server, the stability is the highest priority

I'm considering to put better thermal compound on avalon when summer arrives, but that is a lot of work, when I do it I will post some photos of the other side of the PCB

By the way, my avalon running firmware 0225 shows nice stability, cgminer has been running uninterrupted for days, really a robust and solid product

[SebastianJu]

Did anyone try to add little heatsinks to each of the chips and did it lead to better performance somehow? Maybe more stable overclocking?

Dr. Zhang recommend NOT to add any heatsinks on these chips and saying that will reduce the surface airflow for PCB board

I see the point in this but i wonder if you would put it on following the airflow and not blocking it if the effect wouldnt be the opposite. The air could flow free but has more time to take on the hot temperatures.

I asked this already but didnt get an answer... how is the capability of an avalon for being overclocked? I only read someone experiencing a wall at 300(something) but no info if this is a wall by design or if it could be fixed with better cooling. For example, if one would try, watercooling the single chips.
And is it somehow possible to change clockrates, volts and so on? Only theoretical for the moment because i dont have an avalon yet. But i mean... only 10% more hashingpower would have a big impact. And such a number is easily achievable with normal GPUs too. So im wondering if something is possible there. Of course one would have to be ready to lose the miner when playing with such things.

My past experience from GPU mining tell me: Do not overclock too much, the long term stability of the mining rig is far more important than a small gain in hashing performance, especially if you want one rig to run for a long time (Now we see that BFL is also power hungry, so avalon would at least be around for a year or two)

I had a pair of 5970 overclocked quite high (just 20Mhz below the crash point), undervolted, installed custom cooling device and the best thermal compound I can find (liquid metal), although the GPU temp never goes above 55 degree, all 4 GPUs on both card still died one after another in 1 year. I guess some of the components on the board are just not designed to endure that kind of added load for a sustained long period. For a gaming rig that run once a while it is fine, but for a machine running 24x7 like a server, the stability is the highest priority

I'm considering to put better thermal compound on avalon when summer arrives, but that is a lot of work, when I do it I will post some photos of the other side of the PCB

By the way, my avalon running firmware 0225 shows nice stability, cgminer has been running uninterrupted for days, really a robust and solid product

Do you use a custom PSU, if so which?

I doubt a bit that your cards died because of the overclocking. Im not a pro but at least i believe i learned that overclocking itself only has  very slight negative effect on lifetime. And that only comes from higher temperature. The real threat is overvolting it. Its something with leaking electrons that destroy the circuits or so and make them age. So i always downvolted and overclocked my gpus and they never died. But i didnt use them 24/7 of course. I tend to think that you didnt do anything wrong with undervolting and overclocking and it was because of the fullpower-use all the time.
But like i said im no pro in it...

I really would like to have an avalon hashing now too...