dogie
Legendary
Offline
Activity: 1666
Merit: 1185
dogiecoin.com
|
|
April 29, 2013, 12:24:44 AM |
|
You can cool the oil...
Does anybody know the measurements of the heat sink? I will let myself manufacture some water cooling elements, maybe I also offer them for sale.
You can cool the oil via a heat exchanger, but then what is the point of using the oil to substitute air if you then have to use a heat exchanger to cool it and a dissipation system to move it?
|
|
|
|
el_rlee
Legendary
Offline
Activity: 1600
Merit: 1014
|
|
April 29, 2013, 06:21:49 AM |
|
less noise & energy consumption
|
|
|
|
gyverlb
|
|
April 29, 2013, 10:12:04 AM |
|
You can cool the oil via a heat exchanger, but then what is the point of using the oil to substitute air if you then have to use a heat exchanger to cool it and a dissipation system to move it?
If you have hot spots in your system that are very hard to reach by conventional methods, oil submersion can be a solution. This is quite rare though and oil submersion is mostly done for fun (I'm not sure if the fun lasts when you have to clean things up).
|
|
|
|
allinvain
Legendary
Offline
Activity: 3080
Merit: 1083
|
|
April 30, 2013, 12:19:57 AM |
|
Any updates on this. When will a prototype be ready ? Summer is coming and I have a feeling it's going to be a scorcher even here in the south of Canada.
|
|
|
|
el_rlee
Legendary
Offline
Activity: 1600
Merit: 1014
|
|
May 03, 2013, 06:39:03 AM |
|
Could somebody point me to the measurements of the heat sink? Cannot be a big secret?
|
|
|
|
Malawi
Full Member
Offline
Activity: 224
Merit: 100
One bitcoin to rule them all!
|
|
May 03, 2013, 05:41:34 PM |
|
Not sure about that oil thing. I have read that you easily get hotspots either because of trapped air or because of fluid dynamics where the fluid moves slower the closer it is to the object and this effect easily creates spots where the oil does not move fast enough.
I have no exp wit watercooling myself, but would think that a design where a thick copper-plate is milled with a cnc is the best option. -That is if the copper does not corrode. If so I think a design where copper "bullions" is pushed into an aluminum block. -Then the aluminum block is carved with cnc so that the bullions are as close to the water as possible without having contact. This may create it's own problems if the two metals expands/extracts at different rates. But I have seen many heatsinks in this fashion, so it must be possible.
To me it seems like the part of the block that will have water going tough, should have the intake at an angle and rounded corners so that the water whirls around inside. i also think that a wave-pattern might be best to get the most surface with the least resistance. (Waves that go along the expected waterflow.) Not sure how that will work to get an even distribution of the waterflow at such a large and rectangular surface though.
|
BitCoin is NOT a pyramid - it's a pagoda.
|
|
|
el_rlee
Legendary
Offline
Activity: 1600
Merit: 1014
|
|
May 09, 2013, 07:29:56 AM |
|
Could somebody measure his Avalon and post the results? I would make the construction drawings for my water cooling elements public!
|
|
|
|
|
SolarSilver
Legendary
Offline
Activity: 1112
Merit: 1000
|
|
May 13, 2013, 01:34:34 PM |
|
Did my design now based on the measurements in the Altium Designer files. Don't have my Avalon until now - so no chance to check! Be careful!
If your cooling block replaces the original cooling element, isn't there a risk of damaging/crippling the Avalon module by not being able to get all of the chips to stick correctly to the new cooling element? I have a mixed experiences from the past with FPGA boards (individual chips overheating and shutting down) and those only had 4 chips. The Avalon module has 80 chips... I guess it's not feasible to have a negative of the current cooling element that would slot on top of it with thermal paste, thus not disturbing the original Avalon setup?
|
|
|
|
el_rlee
Legendary
Offline
Activity: 1600
Merit: 1014
|
|
May 13, 2013, 01:41:18 PM |
|
The surface on my drawing has a very nice specification (Ra 1.6), that's very smooth. Also you can see from this picture that they merely used half of the screws https://en.bitcoin.it/w/images/en/4/41/Avalon-modular-4.jpg - who knows why. However I don't believe they used some magical mumbo jumbo when screwing the hashing modules onto the heat sink. One word on the used packaging for the chips: from a heat distribution view I don't know if a worse -or cheaper- choice exists.
|
|
|
|
tom_o
|
|
May 13, 2013, 03:35:20 PM |
|
Not sure about that oil thing. I have read that you easily get hotspots either because of trapped air or because of fluid dynamics where the fluid moves slower the closer it is to the object and this effect easily creates spots where the oil does not move fast enough.
I have no exp wit watercooling myself, but would think that a design where a thick copper-plate is milled with a cnc is the best option. -That is if the copper does not corrode. If so I think a design where copper "bullions" is pushed into an aluminum block. -Then the aluminum block is carved with cnc so that the bullions are as close to the water as possible without having contact. This may create it's own problems if the two metals expands/extracts at different rates. But I have seen many heatsinks in this fashion, so it must be possible.
To me it seems like the part of the block that will have water going tough, should have the intake at an angle and rounded corners so that the water whirls around inside. i also think that a wave-pattern might be best to get the most surface with the least resistance. (Waves that go along the expected waterflow.) Not sure how that will work to get an even distribution of the waterflow at such a large and rectangular surface though.
Mixing metals in a watercooling setup is a big nono in PC circles as it causes breakdowns of the waterblock elements, you wanna stick to a single metal really. xtreme systems have a lot of threads and year of experience on custom coolers you may want to stick your head in over there.
|
|
|
|
Icoin (OP)
|
|
May 24, 2013, 09:10:01 PM Last edit: May 25, 2013, 12:11:06 AM by Icoin |
|
|
|
|
|
AvenG
Full Member
Offline
Activity: 133
Merit: 101
Res Et Non Verbum
|
|
June 25, 2013, 02:33:34 PM |
|
So, when are you starting the manufacture process?
I'll definitely buy some of these..
|
not for sale
|
|
|
aTg
Legendary
Offline
Activity: 1358
Merit: 1000
|
|
June 25, 2013, 03:05:33 PM |
|
So, when are you starting the manufacture process?
I'll definitely buy some of these..
+1
|
|
|
|
CoinHoarder
Legendary
Offline
Activity: 1484
Merit: 1026
In Cryptocoins I Trust
|
|
June 27, 2013, 02:08:56 PM |
|
So, when are you starting the manufacture process?
I'll definitely buy some of these..
+1 +9999 After seeing I could probably OC a 3 module Avalon to 98 Gh using proper cooling, I am definitely interested in picking up some of these to test them out!
|
|
|
|
el_rlee
Legendary
Offline
Activity: 1600
Merit: 1014
|
|
June 27, 2013, 02:57:25 PM |
|
So, when are you starting the manufacture process?
I'll definitely buy some of these..
+1 +9999 After seeing I could probably OC a 3 module Avalon to 98 Gh using proper cooling, I am definitely interested in picking up some of these to test them out! -> https://bitcointalk.org/index.php?topic=211825.0
|
|
|
|
|