Afternoon folks. Yes, unfortunately it's another of these pesky watercooling threads popping up. However, I've not got the usual questions. I'm actually feeling pretty focused about what I want, and what I need to do it successfully - That's not to say I don't have my qualms. My system is as in my signature, and I plan to add another 580 and upgrade to a 30" monitor prior to BBC3 coming out in October for some 2560 * 1600 goodness. I will be cooling the CPU and GPU(s) only, both which will be overclocked as far as is possible (the CPU is currently on a mediocre clock/voltage as the H60, even with 2 3000rpm gentle typhoons isn't quite up to it!). The plan is a single loop with a quad 120mm radiator doing the cooling; from what I've read this should be adequate if a good quality radiator is used in a push/pull configuration. For those interested, the planned layout is: Reservoir -> Pump -> Radiator -> CPU -> GPU(s) -> Reservoir. I'm pretty sure I'm all set re: components and think I've got everything largely picked out in my mind already. So, where's my question/problem you ask? Here's the kicker - I've no space for a radiator of any sort inside my case, so need to mount the radiator externally. I've sourced some external radiator mounts at specialtech, and I plan to mount it (physically screw it into the wood) at the rear of my desk either underneath, or on-top if it looks okay (aesthetically) after purchase, which is of some additional merit; having a cool/lukewarm breeze blown over me while sitting at my desk in summer! So, by now you're frothing at the mouth: All this babble and assertiveness, and no sign of impending problem; TheStockBroker's got it all figured out. You came to this thread because you wanted to help but there's nothing to say or do, other than dish out negative rep for wasting your time. But I ask you this: How do I control, or even power these eight crazy fans sitting on my desk metres of cable away from my chassis front panel for a integrated 5.25" fan controller, and feet from my PSU's molex outputs (not that I would want them running full tilt 24/7 in this manner anyway)? Also, will a standard D5 cope with traversing the miles of tubing needed to accomplish this feat (and therein how much liquid will be needed)? Any suggestions appreciated whatsoever, and any corrections re: anything I've said also appreciated. TSB
I would get a 12v psu from maplins, and use that to power a fancontroller, which in turn would power the fans. Lemme see if i can get a link from maplins website. edit: something like this http://cgi.ebay.co.uk/HDD-POWER-SUP...-MOLEX-EU_W0QQitemZ320705061192QQcmdZViewItem
Not a bad idea, if a little inelegant! Although, that would appear to supply 34 watts as an absolute maximum (if that's 2A on both rails simultaneously), and I would expect fans and controller to draw a probable 140 watt minimum - will look into this further - in the meantime, keep the suggestions coming! TSB
This is one of the few occasions the a cheap, unbranded, PSU should be ok. Combined with this(or 2). The D5 should have enough grunt for it, only one way to find out really... You can always buy a second if needed. And lots of fluid will be needed, in my loop (in sig, but I'm sure you've read my thread ) I used about 1L of fluid, so I'd say 3 times that...
Although I don't have a watercooling setup, I would like to comment on the NZXT Sentry product linked by Pete. I tried the Sentry LX once upon a time. It now lives in my spares box and I've gone back to an analogue fan controller (Zalman MFC1+). I found analogue controllers to be more fine tunable and more reliable for me.
Not sure why you would need 8 fans... Am using 4x Scythe Ultra Kaze 2000 rpms on the external rad in my proper system which 'should' be creating a little more heat than you're trying to deal with d.t. the different CPU & o/c... & temps are really good even under 100% load testing. As to powering it, i use a Phobya Radiator Stand -> attach each fan molex connector to the next & tuck them between the rad & the bottom of the stand -> & then use a molex extention cable running from inside the case - through the back where a top psu 'could' be fitted - & to the first of the 4 fans. [NB i actually do the same thing with the internet machine only using a 360 rad & different fans.] Oh, & i also don't bother screwing the stand to anything - instead, it balances quite happily at a jaunty angle resting on the bottom edge of the fans & the stand. As to the location, i think you're being a little ambitious having it that far away from the case - re tube lengths... i have the stands/rads on a shelf above each machine & that works perfectly well... ...okay, it's then not going to cool me down with a nice breeze, but you'd be better buying a cheap desk fan for that than arsing about trying to get the rad fans to do that as well.
I have to agree with pickets. I would just use an extension cable for the fans. Specialtech do a good 4 to 1 3pin cable. You can then connect this to a 7v cable and do away with fan control or connect it to a fan controller. Also check these out for a good way to root pipes http://www.aquatuning.de/product_info.php/info/p7438_Alphacool-HF-38-Slotblende-Panama.html
if it were me (and providing you have space on or next to your desk), i would mount the rad(s), pump, and fan controller, along with a dedicated el-cheapo PSU all in a small enclosure of some type. mayhaps an ITX mini tower or the like. this means the only "stuff" you have to run between your case and the rad(s) is 2 tubes for water and maybe a tach signal if you feel so inclined. everything is properly tucked away in a case where its protected and hidden from view.
Agreed; from my own personal experience I prefer the theoretical precision analogue offered over the digital 'steps' in RPM. That said though, I can never get two fans to match speed in analogue, and that really sets my OCD off, not having exactly identical RPMS showing! Well, this will be my first time watercooling; so maybe I don't need 8, but I felt from what I read (wrongly or rightly) that 2 GTX 580s, a physx card, and a CPU on a single loop was asking a lot for a single radiator - so I should try to get as much heat out from it as possible. I'm not totally against straight molexing the fans if it saves me a headache otherwise - but I just can't stand anything more than 1.6kish from the Gentle Typhoons I have on my H60 already - they're so noisy! - and that's with them inside the case, under my desk, away from my delicate ears. Surely not screwing the thing down means it rattles/vibrates on the desk, or is the radiator just that heavy? (In which case I wouldn't bother myself) I felt I had to attach it to the desk, as I couldn't leave it on the floor, and there's absolutely nowhere to attach it to the my case itself - so the underside of my desk, with a few cm clearance seemed sensible. The breeze was just a mental bonus, as I tried to convince myself it wouldn't be a major annoyance to have whiny fans blowing papers about my desk all the time! Not a bad idea, I do actually need a mini pc to run 24/7 and automate some things for me at the moment already, so perhaps I could combine the two! *goes to check lili for the enclosure now* Cheers guys. TSB
Must admit i misread slightly, as i didn't notice that you were adding an extra 580 - sorry about that. This being the case, imho you really do need to look at having a shorter loop - &, i'll consider that you 'may' need to look at 8 fans; although i'm not 100% convinced it'll be necessary if you get 4 with a decent cfm... ...well, adding a 2nd 580 wouldn't, for example, double the temps with my 480 loop - at most something around a 10-20C increase using this post as a very loose guide vs my temps. Well, testing what o/c the 580 would do on one the odd particularly hot days recently, i couldn't get it over 43C even with a large increase in voltage & both it & the (4.8 o/c) 2600K at 99%+ (at a stable o/c then it fluctuated between 41 & 42C)... ...so there's a lot of headroom if things are set up properly. Otherwise, i personally don't find the 2000 rpm Ultra Kazes to be *that* noisy - & the proper machine gets used for a variety of a/v editing (plus some occasional gaming) where i need to be able to hear fairly clearly what's going on to time stuff & they've never been an issue... ...& i've got a variety of headphones kicking around so it wouldn't be difficult for me to improve the situation if i actually needed to, but haven't thus far... ...though my ability to shut out extraneous noise & focus on what i'm trying to do could always be unusually good, & the computers have their own room though so i'm not attempting to sleep with the sound or anything. & finally, with either of my kits (obviously the 360 is slightly lighter), there's no noticable vibration at all - the rad/stand/fans stay perfectly still & they're resting on 7-8mm chipboard (cheap industrial shelving that's strong enough to hold computer cases & DASes & whatnot) rather than some magical vibration absorbing stuff... ...although i do use neoprene gaskets on either side of the rads, though that's because provides a better seal for the fans &, as they're hardly all the money in the world, then it seemed churlish to not buy 2nd ones for the stand side of the rads.
what does the length of the loop have to do with anything? if you size your pump accordingly, you could run the tubing around the outside of your house if you wanted and it shouldnt effect performance.
PocketDemon, cheers - I won't quote you for sake of thread length. But, as the second 580 will be added after initially setting up the loop - which will require me to rebuild it when I receive it, I'll see how I get along at first with just four fans and my CPU and GPU overclocked further than they are currently, and make a decision when purchasing the second card. May just leave the 550/560 on air for physx. I had already planned to use your mentioned neoprene gaskets, as well as some shrouds to maximise performance and minimise noise (apparently) so hopefully those will help somewhat in both respects. Kaze's, eh? I may very well just research fans a little better before buying this time round, check the quoted db figures... Appreciated, I didn't really think about the physics of it - I just wrongly jumped to the conclusion of more tubing equating to more work needing to be done by the pump Thanks guys, I think I'm pretty set for now! TSB
The phone went whilst i was typing this so this was supposed to be before the OP's post... Other notes... i've never bothered with shrouds, but they apparently can give decent results - fan grills are a must though as it's very easy to forget (since you don't move the rad often) & attack your fingers inadvertently. i based my fan & rad choice on a round-up of the common 360 rads - though my results are probably going to nearer to the undervolted to 2300rpm U Kazes as they're 2000rpm (balancing cooling with noise as the quoted dB for the 3000 on the manufacturer's site looked a bit too excessive), but it's a 480 rad. As you can see from the tests, different rads are optimium at different cpm rates - my GTX ones are designed for higher cpms -> greater cooling ability at the expense of noise. Okay, if the OP wanted to have numerous pumps (as a single one with both huge head & flow rate is likely to destroy the loop from the inside) then of course they 'could' run it where ever they want... (though, beyond a certain point, i believe they'd need to add extra rads periodically) ...but, on the basis that pumps aren't free (or particularly cheap), assuming that they're looking at something like either - (a) a single D5 (what lots of people use / claims to be quieter than the DCC1 / more variable in speed which is handy for less restrictive loops) (b) or a single DDC1+ Ultra (my pump of choice (with the very unrestrictive XSPC res top) / lower max flow rate but higher head pressure than the D5# / tested better as a single pump on more restrictive loops than the D5 back in the day) [NB # flow rate is the rate that water can be shifted in an unimpeded state, whereas head pressure (whilst literally the height vertically that a pump will raise water to) is important when looking at a 'loaded' (ie restricted by going through water blocks, rads, reses & extended lengths of tubing) - having either being too small would make for a shonky pump, so it's about balancing for your needs.] - having a ridiculous amount of extra tubing to place the rad (& or a box with rad/pump/whatever, as it's been suggested as an idea in this thread) a significant distance away wouldn't be a great idea without (again) extra pump(s) & increased cost. Obviously, externally mounting a pump will lead to some increase in tube length anyway (vs internally mounting) but, when this is done within reason, the disadvantage of the tube length is surmounted by the decrease in temps simply by the external mount - partially by considerably lower case temps & partially through the air that's being pushed &/or pulled through the rad not being pre-heated by the components in the case... [oh, & it's also much quicker to get the air out of the loop as you can rotate the rad periodically around all its axises during the leak testing phase - rather than trying to tilt the whole case about.] ...but once you increase the tube length beyond the minimum amount necessary to sensibly place the rad (you need a bit of flexibility to get to the back of the case & whatnot), it will gradually increase the resistance as the tubing adds turbulence.
Haha! The thought alone of trying to do that with the fully loaded FT02 (even without the water etc) is enough to make my back seize up (and I'm only 22!) TSB
Yeah... i've done it with both of mine up to ~11 months ago (& i'm 15 years older than you) & it's not the most convenient of things to do... Though the weight's not the only factor... it's also when you suddenly realise you've not screwed the top of the res on...
while increased tube hose lengths do add more pumping resistance to the loop, the effect of a half dozen or so feet is pretty trivial. for instance, im running a single loop for just my 5870. since this is a temporary setup for me, the pump and 240 radiator are mounted outside the case, on desk. my pump is a tiny EK 2.2. add to that the fact that im running the pump and both fans on only 5v and you can imagine my flow is pretty low. but my GPU still stays under 40C at full synthetic load. the OP is looking to run a much more restrictive loop than i am, but hes also looking at a much stronger pump and will probably be running it at 12v. even with quite a bit of space between his computer and rad, he should be OK with that setup.
A couple of feet to the total tube length perhaps, but the OP wrote "metres of cable away from my chassis front panel" - which therefore means the back of the case will be at least the same approximate distance away... ...though obviously cable can be routed much more sharply than tubing so it's likely that there will be a more considerable extra length. So being a plural (though undefined) length, i imagine we're looking at at least 2-3 metres of extra tubing each way (but this 'may' be much higher - sensibly, based on common room sizes, this isn't likely to be more than 4 metres each way, but we don't have the info so it 'could' be higher), & at that length there will be a discernable decrease in flow rate -> a decrease in heat removal/higher temps. Now, as you say, the OP is looking at a much more restrictive loop - 3 blocks, twice the rad size &, without moving the rad to the other side of the room, a more lengthy loop anyway - plus (obviously) ~3(+) times more heat (if i can estimate properly at a quarter to 4 in the morning) being produced by the cpu, gfx cards & the pump itself... ...& so i'm not convinced that you can reasonably draw conclusions from your one - whilst i'm a gfx card block down on the OPs, at least everything else is something like similar. Although admittedly, on esp hot days (far above the low-mid 20s we've had so far in Sheffield), i can reduce the ambient temp significantly as i've got a snazzy Australian water cooled air con unit in the room - so i'm working at a significant advantage to the OP. it's then perhaps important to note that there are a significant no of people who, faced with the OPs cooling requirements, would either look at two loops or a 2nd pump in the same loop - & that's without adding 4-8(?) metres of tubing. it is, however, not my money that's being spent so imho it's better to advise a shorter loop where a single pump is more likely to be man enough for the task (& being far less likely to lead to disappointing results for the OP), rather than suggesting that a more considerable sum is added for at least a 2nd pump; not forgetting the cost saving on tubing & fluids. Still, that's just my considered opinion.
