New to Watercooling, few Q's

Putting the cards in parallel means the flow through each one will be lower, but the pump acts on a lower peripheral resistance so the flow rate overall in the loop will be higher. The cpu block in particular will appreciate the higher flow. However if the parallel sections aren't identical it gets much more difficult, as one side will tend to starve the other of flow. This is why the sketch shows flow into the bottom card and out the top rather than both in and out on one block.

If the fill and drain pipe are connected to one point, you just have one line which you can use for either. When draining you need a means of letting air in really, if there's only one tube you have to give it a different route in. The multioption has four holes, best performance will be bottom hole going to pump and top hole coming from loop, fill and drain on the other two. Easiest will be using the top for fill line, base for drain, but performance will suffer a bit.

If you put the drain line elsewhere using a T, say near the pump at the base (the lower the better for a drain line really) then you block up one of the holes in the res. Alternatively you could mount it horizontally so a drain line on the side still lets all the liquid out but when working the flow is still straight through. These all seemed too bewildering when I set out, which is a large part of why I chose the combined pump/res and put the drain line elsewhere.

Description of reservoir arrangements above was crap, have a picture
21chx4x.jpg


To muddy the water further, you don't need a reservoir at all. A T line before the pump with a longish tube running up to a fill port essentially is a reservoir, you just have to pour water in more frequently while filling it. From this perspective the horizontal reservoir is just a bulge in the tube which happens to be where the T line goes.
 
Last edited:
I wouldn't do the cards in parallel, not seen it done many times and I expect its because you couldn't guarantee the water was going to both cards.

Like I said, they don't react that strongly to temps because water takes them so far below where they normally operate.

I normally see watercooling CPUs as clocking and watercooling GPUs as making the damn things STFU as reference coolers, especially ATI ones are so damn noisy.
 
^I've not seen graphics cards in parallel particularly often either, but I have absolute faith in the above. I'll do it as soon as I find the funds for another card. The resistance in each arm is identical, as it's one block + one sli connector in each case. Water flowing from bottom upwards ensures air can't become trapped in either, if water flow was downwards there would be a risk of the top block trapping air.

If it didn't work, the issue would manifest itself in the top card running hotter than the bottom (though still cooler than stock), and you'd resolve it by removing one of the sli connectors which I'd be happy to buy off you. Parallel loops are more complicated, but they allow you to optimse the loop beyond what series permits. And this is a very short parallel section with identical arms, much easier than trying to match flow through different blocks.
 
Last edited:
I don't think thats really what he was getting at, you're still effectively pushing water to one block and then the other, the way I read what he said was like having a y splitter feeding each block equally.
 
matt100 said:
you're still effectively pushing water to one block and then the other
Click to expand...

Its not, its exactly parallel and I'm sure that's what Jon has in mind. Each gets half the flow. Less overall restriction and equal card temps.

Follow each circuit, one block and one VID connector per circuit. Making sure the outlet is in the right place lol.

fparalell.jpg
 
Last edited:
I see how you mean now, I'll get 2 connectors then for that :p.

As for the res, you've cleared that up a bit now. I take it you can use either one of the pairs, i.e. Top and Bottom, and 2 sides, for either fill/drain & in & out flow.
Think the route I'd go for would be in/out on the side, fill connected to the top, and drain connected lower close to the pump, keep it out the way. Also for the drain, would you just go for a plug, or some type of valve which you can open/close easier? I dunno how often I'd end up draining it, so don't know if I necessarily need the convenience of a valve heh.
 
matt100 said:
I don't think thats really what he was getting at, you're still effectively pushing water to one block and then the other, the way I read what he said was like having a y splitter feeding each block equally.
Click to expand...

I don't think that would work as if you had a Y splitter directly below and fed both holes you wouldn't create any flow though the block as they'd simply flow though the fittings and out the other side? Unless how you mean is have a Y splitter, then feed each card individually?
With the one entry point, it should split naturally to create equal pressure and run though the cards? Correct me if I'm wrong, still trying to understand, obviously lol.

Either way I'll most likely end up going the parallel route heh.
 
The benefit to setting the cards up like this is higher flow rate at the slight cost of graphics card temps.

If the physically highest port is connected to the pump inlet it'll go a bit wrong, otherwise it'll work. Intake and outlet directly opposite will perform marginally better, but connecting fill/drain lines will be less convenient. I'd definitely go for horizontal mounting if it fits anywhere. Worth considering the ddc + xspc top combination too.

Ocuk sell a ball valve which has both ends threaded to take barbs, I use one of these with a barb one side and a blanking plate the other. This way were the valve to fail or me to open it by mistake, it still won't leak. The benefit to the valve is positioning it over a bottle then opening it with a finger instead of a spanner to let the water out. Probably worth it for £4
 
Ah suppose using two makes sense, better to be safe than sorry! hah.

I take it you mean this one too for the DCC+XSPC? May think about that actually. Depends on the look I suppose, I'll take a look at some WC threads around the internet, see what I fancy :).
 
Yeah. My main worry was forgetting which way around the valve was "closed" while the loop had water in it, much harder for distaster to strike if you only take the blanking cap off with the end of the line outside the case. A fillport is basically just a blanking cap, though it has a minature funnel built in. I suppose this could be used with a valve too, but I'm not very worried about it leaking as it is only part full of water and sticking up, rather than completely full and hanging down.

The xspc top is rather good really. It outperformed the various other aftermarket tops in testing and does simplify filling considerably. The top holes are threaded G1/4" and M20, the M20 one has a plastic blanking cap in which I thought was crap as mine leaked. However you can get an M20 to G1/4" adapter for about a pound which lets you use it with barbs, so attaching a fill line to it. If the fill line is half full of water then the reservoir is completely full, and you'll get performance if anything better than with the straight-through multioption reservoir. Of course the other approach is to screw the ek reservoir onto the top of a ddc with ek/other top using a G1/4" male to male adapter.

EK do a top which is more expensive and performs a bit worse as the reservoir inlet isn't in line with the pump inlet, but their design looks better and is compatible with the various pieces from the multi-option range.

Apologies for the lengthy replies, I tend to think of more things to write partway through replying and end up with essays.
 
No need to worry about long replies mate, I prefer them, better than a single line trying to explain stuff lol.

I take it M20 are metric measurements of screw threads?
And how come a completely full res is better performance? Just the amount of available water to be cooled, or the overall pressure in the loop?

Sorry for all the bone questions, want to understand more of the concept behind water cooling, rather than simply getting a system working heh.

EDIT:

Also a quick question about fans, what's the best thing to go for? Higher Static Pressure, or Higher Airflow? I'm guessing higher static pressure for high density rads? Only because I've not long bought myself a pair of Noctua NF-S12B fans, and I really like them, quiet with good airflow(100.6 m3/h and 1.31 mm H2O @ 1200RPM), will this be suitable? Or should I look for anything else more specifically?
 
Last edited:
Cheers :)

M20 is 20mm nominal diameter, can't remeber the pitch of hand but would guess it's 1.25mm. It tends to be sold as "M20 xspc reservoir adapter", a bit awkward that they use this but at least they chose a standard thread. I think G1/4" is so universal because it happens to be identical/compatible to one of the metric threads.

A full reservoir prevents cavitation, which is the pump dragging air bubbles into the intake and a risk with powerful pumps. It also reduces turbulence as there is no air space for water to slosh about in. I think the result is performance almost as good as a T line, though I balk at the equations requred to prove this.

I agree fully with the sentiment that it's better to understand than to bang it together and hope. Unfortunately fully understanding it features thermodynamics and fluid mechanics with resultant differential equations, which is difficult. The main justification for me starting watercooling was practical experience with this, as I'm studying engineering at present. I'm definitely learning from my mistakes. Feel free to ask as many questions as you like, most of them have been significantly thought provoking and a distinct improvement on "spec me" :D
 
And also makes intereseting reading for others to, so thanks for asking stuff, and thanks to jon and the others for answering.

Iv learnt a few things from this thread myslef.
 
Thermodynamics and fluid mechanics.. piece of cake mate ;) lol.

But yeah, I understand about the turbulence thinking about it, I suppose not having a steady flow would affect performance. Didn't realise the pumps would be so powerful they'd drag air in mind you! So yeah, having a fill tube 1/2 filled would make sense.

Just received and put together my 800D too, and good christ to god it's big, and heavy, especially walking around 1KM with it across base, wasn't the most fun experience lol. The extractor fan and intake fan on the base of the case are actually 140mm fans, so I'd have no issue in putting two extra 120 rads in there what so ever. Also think I could move into the case if ever the need arise haha.

Just asking this again, as you were most likely typing out your response when I did my edit. Just a quick question about fans, what's the best thing to go for? Higher Static Pressure, or Higher Airflow? I'm guessing higher static pressure for high density rads? Only because I've not long bought myself a pair of Noctua NF-S12B fans, and I really like them, quiet with good airflow(100.6 m3/h and 1.31 mm H2O @ 1200RPM), will this be suitable? Or should I look for anything else more specifically?

And no problem bifday2k, glad my thread has been able to help others :).
 
Just to add to JonJ excelent posts - a D5 and Typhoon 3 res as another option.
That way to get two loops off one pump and res.
It's drive bay mounted and alows for removeable drain/fill lines on the front.
Should tidy up the loop/case and give higher flow to both loops (or just a single loop)

If it didn't increase the noise level of the D5 slightly (still well below fan noise) it would be perfect.
 
Last edited:
Yeah, had seen them, had actually wondered how it would work exactly, but I think I've got it figured out. I think I'm going to go with the single loop option using the EK Multioption Reservoir 250 Rev 2, just something about it I like, think it'll look quite tidy :D.

Unfortunately, after looking at my credit card statement, and my payday, and after a bit of calculation, I won't be getting this in about 3 months as planned :(.
Totally going to end up getting it at the end of this month, or very beginning of February, my calculations came out better than expected haha, can't wait :D.
 
Ah, I always forget the typhoon 3. Cheers ss.

It's amazing how heavy cases can get. I can still lift mine onto my desk, but I wouldn't want to walk any distance with it. It was much more reasonable before mass loading the sides and filling it with water though, so I think your one 800D will be a lot heavier in the near future.

140mm may be a very good thing, thermochill are starting to release radiators that take 140mm fans though I think it's only the triple so far. That said, there are few good 140mm fans available so 120 still makes more sense at present. As you suspect, static pressure is what matters. Airflow is quoted as a figure into zero resistance, while static pressure reflects how much of this airflow it maintains through resistance. A radiator, even a 60mm thick one, imposes significant resistance.

The Noctua 120S is an excellent case fan, but a woeful radiator one. Certainly you'll find uses for them, I think you'll find one aimed at your motherboard chipset shortly after putting a block on the cpu, and it will be better than whatever fans corsair have shipped with their case. The fan I most recommend for a radiator is the 38mm san ace which are difficult to get hold of. It's just far better than the alternatives. Out of the 25mm fans, the scythe gentle typhoons are the most highly recommended. The higher speed the better, as static pressure scales quadratically with rpm and you can always undervolt fans to slow then down. So if san ace cannot be found, I suggest the 1850rpm gentle typhoons. Bang for buck is the 38mm yate loons, followed by the 25mm yate loons, but I don't believe these are highly considered as far as noise goes.

Past these the standard "high pressure" fans will do, the noctua P12 (sorry :( ), scythe s-flex, akasa apache. I'm using one san ace and two s-flex, will get another two san ace when I can.

Entirely with you on the multioption, I want to change from xspc res top to one of these. I can't justify it in better terms than that I want to though.
 
I have the 150EK res (I use mine upside down) fillport and inlet at top, outlet to D5 at bottom and it's a really nice bit of kit.
But if using a DDC - the XSPC res top give the biggest performance gains
Worth having a look a case central - a few 800D builds linky
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom