That looks incredible, hope you get it to turn out like that in the end.
Printable View
That looks incredible, hope you get it to turn out like that in the end.
Looking at the WC layout i think going from the CPU to the small Rad to the GPU then to the dual would be better, that way your GPU will get better temps, which would also allow for a sweet OC on the CPU. and GPU :D But that works looks stunning.
I've been thinking about that.. But I can't do a loop that does that... Here is why:
There are too many components in a too small of a place.. I chose the best straight lines between fittings, so as not to bend the tubing too much.. but..
I could manage to squeeze this:
But there is one portion of tubing that is quite bent.. I'm not sure if it would hold in real life.. probably need some springs/coils around it. Which I don't have... Also, there is no way I could do this with 1/2" tubing.. That bend is too sharp.. But with some special fittings, could be achievable.. Still, more complexity could mean more leaks.. more flow loss..
Gotta thank you for your suggestion, though.. This is the good kind of peer pressure, as I had disqualified the second loop from starters, but it appears do-able now. 10ks :)
There is no telling on the influence of the cooled liquid on the system and I need to study more on what could happen condensation-wise. Also, if I give too much power to the element, I could freeze the water or end up chilling my room ;) One thing that got me thinking when reading about pumps and liquid dynamics (not PC related, but applies) is the fact that having several blocks to slow down liquid will mean the liquid will spend more time in the radiators and get more cooling time vs getting more heating time.. so, the difference between the 1st and second loop should be minimal.. should be :| ? I'm not that good with physics and mathematics to calculate this, but real life testing is better :)
That does look do able, personally I would leave out the South bridge water block and just get a heatsink with active cooling, would end up cheaper and you would have slightly cooler temps on everything else. Great sketches by the way ;)
10ks :) As for leaving the northbridge out of the loop.. hmm.. I'm not sure.. could be an option.. Right now there is a HR-05 on top of it.. one of the best available chipset cooling devices, IMO... it is fully passive and if it has just a lil air flow around it keeps the chipset cool enough to the touch... So, adding the block there isn't necessary, but I would like it :) If I don't get all my stuff together, only the cpu block will remain in the loop anyway.. for a while.. I may order the other blocks this moth, but as for the video.. There is a choice to be made as I have money for only one thing at a time...
If I use the lowest possible pressure drop block there, it serves more as a means to an end, because I can't seem to find a way to route the tubing the way I want in any other way.. unless I look more thoroughly ;)
I'm looking at either a simple labyrinth block or the mcw30, which I don't know what has inside (can anyone help?).. But I like it.. So, does anyone have any recommendations regarding a good chipset block? also.. cheap :)
I said leave tho south bridge off, not the north bridge but that Is also an option. The north bridge gets a lot hotter than the south bridge. As for a north bridge block I would recommend the Swiftech MCW-30, it's made the same way as the Apogee GTX and that cools really well, it's not that expensive either. If you really wanted you could use one on the north and south bridge because it's compatible with both. If you have a look on the website you can see that it fits on both the eVGA 680i chips with plenty of room, you should be able to use that as reference.
10ks :) But I didn't have enough money to buy a board with a northbridge AND a southbridge ;) j/k.. The gigabyte board I have has an integrated graphics core, no southbridge and ultra small chip controlling everything.. it's like 5mm x 5mm.. AND has 10 integrated usb ports.. So, all I have is a "northbridge".. I think the mcw is the way to go too.. Especially if is made like the apogee.
I was messing with some spare tubes I have and measured them: they are quite small.. less than the advertised 10mm.. And quite thin.. I don't think I like them.. They work, but I'm gonna give more thought to the 1/2" tubing.. to change my stuff I would need like 170$ (incl. shipping) for the 2 swiftech blocks, new fittings all around and new tubing.. Which I would fork out immediately if I wasn't more keen on a new video card.. Which is gonna be more expensive in the end.. I also need a video block.. and that's gonna add some more $ :(
I think I'm gonna look for some copper and acrylic.. I had a nice idea how to easily make something.. close to the apogee architecture, but different.. Still needs refining and reverse engineering.. I wonder if it is moral to take something, improve(or not) it by changing it a lot, but keeping the good idea at its core.. I would call it progress, but I hope I don't get in trouble :p..
I would go with the MCW-30 for the north bridge. As for the GPU I'm going to be using an MCW-60 on my 8800GTX, it's a good block and can fit almost any card, only downside is it's $100AUS plus the adapter plate and extra heatsinks.
[RANT]Well, after visiting the post office to get my small peltier, I was curious enough to ask about taxes applied to.. stuff I order online. Apparently it would cost me 150 euros to get me a video card (be it a quadro or an 8800) thru customs.. errrm.. nope :| won't do that. I asked the guy there what can I do to, well, not pay that much money and he kindly told me that if it is like that, then I probably don't need it.. I won't comment on this anymore. On the water cooling side, I am happy that most of the stuff I want to order isn't on their list, except the water blocks.. they go to aluminum or copper radiators and cost me 14 euros.. a piece.. again.. that is weird.. So, I might end up ordering tubing, fittings and other stuff, specifically what isn't on their list :twisted: [/RANT]
That being said, I'm playing with the peltier right now. I glued 2 small heatsinks to find out which part is the cold side and which the hot.. and I did, but I can only give the unit 5V.. at 12V, the small heat sink can't do much, it gets hot, then the cold heat sink gets hot too... I am keeping it at 5V to see how cool it gets, but no miracles there.. can't get it to work at 7V, cause the psu stops (the peltier should have a resistance of 1 ohms, so that is close to a short circuit.. enter over-current protection). Well, on 5V it is enough to cool down a hot beverage.. I think the temp I'm getting is a subjective 10-15 degrees.. No condensation, no freezing.. I don't have a big heatsink to test the unit with :( Unless I strip on from a working PC.. nope..
Some pics:
While thinking about how expensive water blocks are, long time ago, I started "developing" my own line of cooling elements :alien: lol, right? well, a further understanding of mechanical processes and some limitations vs price and time, I came up with these little puppies:
..meet the exclusive, not available anywhere, AM2 water block, version "who knows what number", revision 2. I won't go into much detail, but the block is a parallel type, combined with a pin matrix type, with a little jet impingement.. or something..
here it is naked.. well, in wire frame.. I personally find a wire frame model sometimes can better relate the inner structure and workings of something ;)
I am learning the ways of autocad in order to make detailed plans, with dimensions and angles and stuff, so I can print them or hand them over to someone when the time comes to mill the blocks.. I found some copper plates, but they don't ship where I live.. and if they did, I've seen a list for construction materials at the damned post office.. again, taxes.. No luck at the local scrap metal yards.. But there is a local industry with non-ferrous metals, so I know I can score some copper sheets around.. I may end with a big bar, just like other DIY on the net.. :)
An the quick render. This shows more the aesthetic part of the block. But it was designed to work, not look good. 2 LED holes on each side should make it glow, and a clear acryl cover should allow the unobstructed view of the insides.
this is how it should look in the end.. I'm mostly playing with ideas, so..
..a night shot too.. still an idea..
And here is the idea behind the renders: how to get temperature as low as possible?
*Well, one way to do it is have the water as close to the chip as possible.. So, in the groove areas, there is only 1mm of copper between water and heat source.
*Another way to cool things down is by getting lots of water in contact with the heat source.. That's where the parallel design comes in, allowing for lots of water to move fast thru the block.
*Turbulence is needed to avoid that dreaded layer of "dead" water that forms on the surface of the water-block and hinders further thermal transfer.. so there is the pin matrix solution..
*To equalize the water path, I made some small cambers at the intake and exhaust ports, to allow a nice and even distribution of water.. Also, I think I positioned the ports in such a way to have a good even water-pressure distribution, and long water path.
*The block is as big as I can possibly make it as to not interfere with other components on the mother board, and the north bridge block has a notch so as to allow a big air cooler on the video board, if needed..
*The holes are threaded G 1/4" to allow for flexibility, but I would go for 1/2" fittings.
I would like to hear some comments on my ideas.. Maybe there is someone here that has some training/experience with water systems, not just installing them I mean. Any comment is appreciated.
Reality time:
..this is a pic of the edges of the non-filed edge of the flames on the side of the case.. I just though it is interesting, as a comparison with the cheap-ass case I use for the tensa project.. I couldn't even capture the edge with my phone in that case :p
..so, the small, 40mmx40mm peltier.. I think it would be good as a north-bridge super-cooler part, but I won't be using it for that..
..the testing rig.. 2 north bridge heat sinks.. small ones..
..mm.. peltier sandwich..
..some power.. a 420W cheap PSU.. prior use: "AC" power unit in the dorm
room, glued on the wall.. current use: tests..
..not knowing how hot it could get, I wasn't planning on holding this in my hand.. so I used a piece of wood for support..
No active cooling picture, cause I don't have thermal sensor on my phone (that would be a nice mod, wouldn't it?).. Like I said, it gets cold to the touch, but nothing more.. Well, it will be a little tough giving the small fry 15V at 10A, but 12V would work.. in the future, of course, when something cools the hot side better..
Well, that about covers stuff.. I'm gonna be out of time for the next 2 days.. so.. bye
I thought of getting a peltier and installing it under the water block, then i saw the price lol (for a good one). I like you water block design, base looks pretty much like a Swiftech Apogee GTX of MCW-60/30. This is good because you already know that cools well, you can only make it better ;)