Provisional Watercooling Loop

[Re-Invented]

Yeah thats the right way but as i said i found it a absolute nightmare, u not considered a danger den res there plain and simple but i found they do a brilliant job.

Found better temps with danger den res aswell, its been a long time since i played with watercooling as a lot had changed since then the bay res might be better.

Hope this helps

Lee

My Rig used to be:

Res >> Pump >> Rad >> CPU >> N/B >> GPU with Peltier >> Res

[[Grevane]]

One quick question for people in respect to W/C'ing and a setup where you have a GPU, memory (and/or northbridge) and a CPU being cooled...

In respect to the diagram above, what are peoples opinions on having the GPU and the CPU on the same circuit? Surely which ever way round you send the flow of coolant, due to the greater heat put out by a GPU, wouldn't you either end up with the GPU "theoretically" heating up the CPU or the pre-heated coolant not cooling the GPU as much as it should if the coolant flows to the GPU after the CPU...

Wouldn't it be better to split the circuit into two and have one circuit with as for example...

Pump -> Split -> RAM -> CPU -> Join -> Rad -> Pump again

and

Pump -> Split -> GPU -> Join -> Rad -> Pump again

(Sorry, not very arty)

And potentially have a couple of flow meters to check both sides of the circuit to check they are both flowing equally.

Would that not be a preferable solution to either the CPU reducing the cooling on the GPU or visaversa, the GPU heat reducing the efficiency of the CPU cooling...??

[Re-Invented]

It seems plausable but i find a better flow rate helps that will just end up bottlenecking back at the rad, but as said before been along time since i dabbled fully with w/c

Lee

[[Grevane]]

It seems plausable but i find a better flow rate helps that will just end up bottlenecking back at the rad, but as said before been along time since i dabbled fully with w/c

Lee

heh hehe, when i said "the diagram above" I didn't directly mean yours... sorry, wasn't particularly picking on you... started typing the reply when it was on page 1... by the time i had posted... well you know... 2nd page already... was the OP's diagram I meant

But yeah, would be interesting to read any experienced W/C'd OC'ers opinions on single or dual circuit WC systems...

[Million]

It's a good idea, but i'm not sure if it works - this is NOT an experienced water-cooler talking by the way!
I would've thought if it did work you'd see it done more often, but I don't think i've ever seen it done that way. Also I think the extra piping would increase the resistance and decrease the flow, decreasing overall cooling. My CPU rarely gets over 40*, and my GPU can allegedly get up to 90* (on air, stock coolers), so I think as long as GPU is after CPU it will work. Don't know about chipset temps, but like I said, I don't overclock.

Oh, and all the reviews I've read about the XSPC single bay res have been extremely positive - a lot of the reviewers started off skeptical of bay reservoirs and by the end were converted

[Million]

Right, ordered. Watercoolingshop.com did it with free barbs of my choice included. I went for the black, as it fits in with the P180 colours - oh, and 'if it ain't black, it's whack', apparently.

[herulach]

The one in my sig/project log (if its big enough to warrant that title) is the only set up either ever done, however, I have done some fairly advanced fluid dynamics courses, and youd basically end up with the water in the most restrictive side of the loop slowing down dramatically. Depending what blocks youre using this will change slightly (you might not have as severe a problem as me with my very restrictive cpu block).

Given your set up I imagine the single rad may struggle to get it as cold asyoure expecting without some serious fan assistance. Youd be well rewarded by a transition to a bigger case and a 2x120/3x120 rad.

[[Grevane]]

The one in my sig/project log (if its big enough to warrant that title) is the only set up either ever done, however, I have done some fairly advanced fluid dynamics courses, and youd basically end up with the water in the most restrictive side of the loop slowing down dramatically. Depending what blocks youre using this will change slightly (you might not have as severe a problem as me with my very restrictive cpu block).

Given your set up I imagine the single rad may struggle to get it as cold asyoure expecting without some serious fan assistance. Youd be well rewarded by a transition to a bigger case and a 2x120/3x120 rad.

Not saying i disagree with you. I've certainly not done the fluid dynamics but I would imagine that surely, even if one side was slower than the other, that the movement of the faster side would effectively "suck" the fluid up through the slower side of the circuit but the sheer movement of the faster side's fluid bast the point where the 2 circuits rejoin?

And also, even if one side of the circuit was slower than the other, you could use flow meters etc... in an attempt to balance it out.

And lastly, surely in terms of the thermal transfer from the components to the coolant, the total heat transfer from each device has got to be better when the coolant isn't, for example, firstly getting heated up by the CPU, then the NB and then all that preheated "coolant" hits the already hotest part of the kit in the GPU...

I just think this is probably a really interesting line of thought and one that should be given more investigation. I honestly find it hard to believe that there isn't a way to make the heat transfer of a dual circuit far superior to a single circuit so that the "hot" coolant only combines the heat once the circuit rejoins just before it goes into the radiator...

[dmuz]

If I understand correctly you guys are talking about a kind of figure 8 with one side going thru the CPU and one thru the GPU but using one pump/rad at the crossover point? This wouldn't make any difference to the temps as you have not changed the total heat input into the system or the heat transfer out of the system. It might acutally make it worse as you are increasing the pressure loss (longer pipe run) which would lower the flow rates (unless you use a beefier pump). To reduce the temps you would either need to use 2 separate loops (with 2 pumps/rads), a bigger rad, take some heat out of the system (eg northbridge) or swap the rad for a heat exchanger with a secondary cooling loop.

[dangel]

Oh, and dangel, I can understand the hollywood factor of putting it all on show, but the P180 is minimalist and kinda austere, so I don't think it'd work. Plus I don't wanna be cutting holes in soundproofed pannels. I actually have one of those bog showy XSPC finned Alu reservoirs, as it was part of the set, but i'm gonna sell it.

Oh don't get me wrong I did it like that because my case isn't big enough to fit it inside (plus it does heat the case and gets the coldest air possible). I didn't cut holes either - I fed the lines through the back of the case Anyway, glad you got it sorted

[[Grevane]]

If I understand correctly you guys are talking about a kind of figure 8 with one side going thru the CPU and one thru the GPU but using one pump/rad at the crossover point? This wouldn't make any difference to the temps as you have not changed the total heat input into the system or the heat transfer out of the system. It might acutally make it worse as you are increasing the pressure loss (longer pipe run) which would lower the flow rates (unless you use a beefier pump). To reduce the temps you would either need to use 2 separate loops (with 2 pumps/rads), a bigger rad, take some heat out of the system (eg northbridge) or swap the rad for a heat exchanger with a secondary cooling loop.

Not sure if we're talking about the same setup, couldn't quite visualise what you meant... anyway, i "knocked up" and quick MSPaint diagram for ya... just to show what I meant and save any confusion or talking at "cross purposes"...

EDIT:Hmm... I suppose it does make a figure of 8 shape... just imagined you meant it the flow was in a figure of 8... gah confusing...lol

[Clunk]

That wont work well if the cpu block is a high restriction one, the water will follow the path of least restriction.

Better to either do it in a normal loop with a half decent pump and an adequate radiator/s, or do two seperate loops if there is room.

[[Grevane]]

Fair enough... and with a normal loop in mind, what's the preferable order of components since the GPU is by far the hotest component in the circuit and essentially I would imagine if the GPU is 1st, the coolant could already be warm enough to add heat to the NB rather than remove it...

something like this maybe?

-> RAM -> NB -> CPU -> GPU -> ?

[EDIT: and this is where I would be slightly concerned about the effectiveness of the cooling on the GPU since there are already 3 component adding heat to the system even before it reaches the GPU...]

[dangel]

I'd agree - keep it simple - single loop will be fine. It's not like the water boils in each block - it moves through too quickly. Is cooling RAM actually worth it? I'd go CPU->NB->GPU.

[Clunk]

Fair enough... and with a normal loop in mind, what's the preferable order of components since the GPU is by far the hotest component in the circuit and essentially I would imagine if the GPU is 1st, the coolant could already be warm enough to add heat to the NB rather than remove it...

something like this maybe?

-> RAM -> NB -> CPU -> GPU -> ?

[EDIT: and this is where I would be slightly concerned about the effectiveness of the cooling on the GPU since there are already 3 component adding heat to the system even before it reaches the GPU...]

Order doesnt matter, once it is pumping round, the water stays a similar temp throughout the loop.

The only thing that can be a pain, is if you have the res after the pump.

This is why, when you are cooling an overclocked cppu, NB, and GP (and whatever else), you need to get a radiator and fans that match, or ideally exceed the heatload of your system.

You can get an idea of you system heatload by using a PSU calculator.

[[Grevane]]

I'd agree - keep it simple - single loop will be fine. It's not like the water boils in each block - it moves through too quickly. Is cooling RAM actually worth it? I'd go CPU->NB->GPU.

Order doesnt matter, once it is pumping round, the water stays a similar temp throughout the loop.

The only thing that can be a pain, is if you have the res after the pump.

This is why, when you are cooling an overclocked cppu, NB, and GP (and whatever else), you need to get a radiator and fans that match, or ideally exceed the heatload of your system.

You can get an idea of you system heatload by using a PSU calculator.

Thank you both, that was pretty much all I was trying to establish.
(And oops, apologies to the OP for kinda hijacking his thread )

And OCZ do some quite nice WC RAM... and basically the same price as non-WC RAM but with a different surround on it... The OCZ WC RAM can even be used non-WC so can be bought as part of a system that would be converted to WC at a later date.