My WC system to be expanded...critique requested
I came to you guys because SLIzone is dead recently
OK, so I have water cooled just my CPU for now:
- D-TEK FuZion (beast) CPU block
- Thermochill PA120.2 (beast) radiator (must...make...shiny) http://www.overclockingwiki.org/foru...lies/wink2.gif
- Scythe Ultra KAZE (beast) fans http://www.overclockingwiki.org/foru...s/blushing.gif loud on full, silent on low
- Alphacool AP1510 (beast) pump http://www.overclockingwiki.org/foru...es/001_tt1.gif
Below are links to big pics of the rig...sorry for the messy wires; when spring brake comes, the dremel shall meet the motherboard tray (MWUHAHA)! The last pic is my rig on air.
As of a few months ago, my eVGA 680i board fried (5th time) and I have been using an ASUS P5N32-E SLI board, and by god is it dumb. Doesn't overclock for some strange reasonhttp://www.overclockingwiki.org/foru...es/cursing.gif! I reached 3.6GHz on the quad core with the eVGA board; I can't stand running @ 2.4 http://www.overclockingwiki.org/foru...ies/crying.gif
Thanks in Advance
PLANS in PDF:
As is now, looking mostly good. Probably sometime around the year 3000, whenever D-Tek finally makes some more of those Pro mounting kits, I'd love to get one myself, because I love how they open up the junk around the socket. That sharp bend between the pump and the rad isn't ideal, but certainly not design breaking. As for your plans, there are a few things that I'd be concerned about.
1) You're stacking two of your radiators. See THIS THREAD for a quick explanation about why this is a bad thing. Best case scenario, it'll simply be less inefficient. Worst case, perfect-storm scenario, the heat soaking and increased flow restriction might actually manage to make your overall system perform slightly worse.
2) Your connection between the two GPU blocks will leave the top one dry. That particular arrangement won't let the water actually flow through the block itself, since it won't have anywhere to exit. Maybe a drawing snafu, but just a heads up. Switching the inlet and outlet on the bottom block will fix that.
3) I'm assuming that your loop order is designed to minimize the amount of tubing you need to run, but priming that system is going to be an absolute :censored:. Getting water to feed from the reservoir all the way down to the pump... well, just make sure you stretch first, okay?:biggrin: I don't know if you can run an AP1510 dry, but many pumps can't be. Ideally, you want the reservoir to gravity-feed into the pump.
4) You've got your chipset blocks on the same loop as your main blocks. You'll probably find you're hurting your CPU temps a lot more than your helping your chipset temps. Perhaps using that rear rad for the chipset and get a little DB-1 or MCP350 pump to run that? And save the Thermochills and AP1510 for the CPU/GPU's?
That's all that comes to mind for the moment.
Should I just add another pair of fans to the rads? Is that all? If it is, than that's no problem :punk:. In case you didn't notice, I must point out that I will be using a shroud between the radiators...
BTW, should I go for smaller tubing with the chipset loop...making the whole thing less complex and painful to look at?
Shrouding never hurts, but the documented benefits tend to measure in the low, single-digit percentage realm. More important is to seal the fans against the rad, be it with a full shroud or just regular foam tape. Extra fans... if you want. As rads go, Thermochills rads have below-average air resistance, so it's a little less important. If you have room and don't mind the extra noise, feel free.
It boils down to basic physics. In the process of passing through the first rad, the air will become warmer. Thus, the second rad will be trying to cool the water down using this same (warmer) air. The warmer the air, the smaller the temperature difference between it and the water in the second rad. The smaller the temp difference, the less heat will be removed from the loop. The slower the air moves, the more drastic this effect will be, but it'll never go actually go away, even with high-powered fans.
3/8" tubing is still plenty for chipset cooling and can seriously help reduce clutter around the socket area. If possible, making the NB block the first after the radiator will ensure that it gets the coolest possible water, since it usually benefits the most from cooler temps (But it's not critical).
About the separate chipset loop...would it really matter?
Out of the first loop it goes straight to the CPU and not the chipsets...so I would guess that would not affect CPU temps...am I wrong?
Thanks In Advance
It wont really make a difference, unless your chipset blocks are restrictive. That will reduce the flow in the whole loop, therefore raising your CPU temps a bit, and the overall efficiency of the full loop. I myself have everything in the same loop, with 2 RADs and a powerful pump, and I am very happy with my temps. If I had a little more room to work with, I would however be doing seperate loops. Only because it is extreme, and to get every last ounce of cooling potential. It is definitley not needed though.
According to MartinsLiquidLab flow calculator, I get 1.08 GPM with the pump @ 12V and 1.61 GPM with the pump @ 24V with 2x PA120.2s, a dual 92mm radiator, a D-TEK FuZion, 2 EK MOSFET, the DD MCP-680i NB bloc, and 2x MCW60s with no 90 degree adapters. How does this sound to you guys?
Should I consider upping the flow rates?
Thanks in Advance.
That sounds pretty complete to me..... All those blocks are quite free flowing, however I dont know much about the 680i NB block. Your CPU temp looks great. I wouldnt be too worried at all.
Thanks for the posts guys!
They really reinforced my decision to water cool my baby-beast!
BTW, The air coming out of my radiator with just the CPU and Pump in the loop is at 26C (ambient 23C~24C). How does that sound?
Thanks to all the dudes who posted:
Just a few more teensy-weensie questions:
- How do the above mentioned flow rates look (they are with the pump @ 12V)?
- Could I wrap the pump in sound insulation without any adverse results (wanna run @ 24V)?
- Will it be possible for me to bend tubing from MOSFET to MOSFET?
Thanks in Advance
P.S. Guess why I couldn't overclock w/ the ASUS.
I FORGOT TO TURN CPU SPEED SPECTRUM ON!
LOL I feel like a duphos.
1 gpm is a popular goal for flow rates. Slower can still work just fine, but if you look at thermal capacities of waterblock and radiators, you can see that they loose effectiveness very quickly below 0.5 gpm, while beyond 2gpm starts to give you much smaller rates of return. Try your pump at different speeds, and see how the temps differ - a high pump speed will increase the pump's heat dump, remember.
I wouldn't wrap the pump - while most of the pump's heat gets dumped into the water, there is still a significant portion that needs to dissipate into the air. I've heard figures quoted anywhere from 10% to 30%, but it's generally agreed that having absolutely NO airflow around your pump is a bad thing, and can contribute to pump failure. Even if you don't have a fan actually going across it, simple convection can do the trick - as long as you haven't boxed the pump in.
If the tubing doesn't kink, going from mosfet to mosfet is fine. This is an area where flexible tubing can really shine, if you're trying to keep things neat. If the tubing isn't as flexible, then you'll just have to use bigger loops.
|All times are GMT -7. The time now is 09:55 PM.|