Cooling WC problems

I had my w/c system running perfectly for one year now. I Decided to extend it and so I ordered a second radiator and a GPU cooler. This is my current setup:




So as you can see from the diagram the water flow goes like this:

Pump-->Rad1-->CPU cooler-->Resevoir-->Rad2-->GPU cooler-->Pump

I am having to use 6mm tubing instead of 10mm in some cases because the second radiator can only connect to 6mm tubing. This system was extremely hard to fill up with water. I simply can't get all of the air out ! The resevoir isn't helping to keep the air out because of some reason. Also the 6mm tubing seems to slow down the setup to a very large extent. This is quite weird, as I am using an Eheim 1250 with 1200 liters per hour. The diagram shows the exact locations of the various components and unfortunately I cannot alter these (had to cram everything into and onto a Coolermaster ATC-210). How can I improve this setup to get a decent flow rate an get rid of the air in the system? I can only change the tubing between the components and not their location!

 

With that setup, you will be lucky of you are getting over 200lph, and that is being optimistic.

The reasoning for this is that the 6mm tubing is adding more restriction to flow. Also, the airplex evo will be a very restrictive design.

You have paired an extremely restrictive system with a pump that likes big and open.

Might I suggest running the two radiators in parallel.

This would reduce the overall resistance to flow in the system.

8-ball

 

Have you tryed it with just the one radiator? Seems like overkill to me unless your going for a veryvery high overclock...GPUs put out praticly nothing compared to a cpu...

 

A 1250 with 6mm tubing? Slight overkill, don't you think?

Why not change the 6mm tubing for 10mm?
After all, the APEs can take up to 13mm ID tubing straight off.



For the air - It helps a lot to orient the case so that the reservoir is at the highest point for bleeding the air out of the system.
Once it's gone, you're ok to run with the res wherever.
If you don't get the air out, you may be harming the performance more than any other factor would (although res before pump is an oft written rule...).

If you have a problem bleeding it inside the case, can't you fill it outside and then mount the components?


This is, of course, assuming your pump (which is overkill, IMHO - a 1048 would do) isn't cavitating and inducing vapour lock that way or drawing air through, out of the res itself (which is possible, I s'pose).
As 8-ball suggested, this may be resolved by running the rads in parallel but with the rads not being identical, you may even harm performance this way and besides, it seems you may still be stuck with a restricted inlet anyway.

(Obviously I disagree with him that the rad is the restrictive factor but what can you say to someone who has, to my knowledge, only ever seen pictures of the components he's commenting on, rather than had hands on experience with them?) *shrug*
To quote my own personal experience to back me up, I have used the AirPlex Evo 240 in a serial configuration to cool a CPU, GPU, NB, two HDDS and a PSU on 6mm ID tube with a 1046.
I'll happily add a 1048 and an APE 120 to prove a point, even if it involved upping the tubing to 10mm to do so.


The serial twin rad approach can be more effective, even with a lower flow rate than a single rad system with a high flow rate.
It's amazing how many people don't realise this fact.

After all, the more coolant surface area you are effectively cooling and the heat dissipation from the extra fannage is going to have a bigger effect on temps than how fast the water gets from the block and back again.

CFM and/or radiator surface area can easily count for as much, if not more than GPM.

- If we take the Papst 120mm 4412 FGML as an example, we see that a single fan at 32dB, moving 67.1CFM can dissipate a maximum of 385Watts.
So lets use the power of three instead of just one or two!
(although Haddy's right, either rad would probably suffice - but two rads will still give better temps)


To sum up, I would say that you should increase your 6mm tube to 10mm if the 1250 is your preferred pump and redo the loop so that the res is immediately before the pump.
So, have the loop go -
Res -> pump -> Cpu -> rad1 -> Gpu -> Rad2 -> Res
and lose that air!

____________________________________

On a related note (and to prove the point further), Aqua Computer recently gained a commision to find a cooling solution for a 42 CPU supercomputer.
Some people might think they would just go and get a humongous pump but no, instead, they release the Airplex Evo 1800!
I'm sure you can work out how it gets its name (hint - the number 120 has a lot to do with the model range )

 

Pug, I never said that gpm was more important than radiator surface.

My point is that for a given area of radiator and a given waterblock, the next method of improvement is "optimising" the flow rates for each of the components.

My point being that if your block performs most efficiently at high flow rates, the ocpc atlantis being a perfect example, and you have a couple of rads which would prefer half the flow rate (the optimum flow rates for max efficiency "tends" to be lower in rads than waterblocks), then running the two rads in parallel, would, provided your pump is also matched, allow all the components to perform at their max efficiency.

Obviously, more rad surface area is better, but their comes a point when bigger isn't possible, and it is necessary to make the best of what you've got.

8-ball

 

oh no here comes the pug v.s. 8-ball battle again to see who can write the longer post

 

So would you care to elaborate on what exactly you were implying when you said

?
You just know I'm going to bite on an inflammatory statement like that.

Exactly, and he has TWO radiators which makes more surface area both bigger and possible.


Q - What will happen if he gets stagnant flow in one radiator because the water takes the path of least resistance through the other?
A - He will lose the performance that he would have otherwise gained had he run in serial.

I will agree that in theory, what you are saying seems sound and it may even work in practise too but until you prove the point, you may be suggesting a worse course of action.

 

Ding! Ding!
Round 3...

Knockout!!!

 

Just keep it with the facts or ill tko you both

 

Ok... I'll try moving the resevoir infront of the pump, however the change to 10mm tubing will be more of a challenge. I was sure, that this tubing was handicapping the flowrate, yet it is not as easy as it seems to change it. The problem is, that I am using the external mounting brackets for the airplex evo. These only support plug and cool connectors, which are 6mm . I'll see if I can somehow mod them to 10mm connectors. The Eheim 1250, does seem to be an overkill, however it has always served me well in my setup with all 10mm tubing. It was just 15€ more expensive than the 1048, so I thought, why not pay that amount for double the maximum flowrate...

 

Pug,

If you look at my last post, you will find that I agree that more radiator surface area is better. But once you have the maximum radiator surface area, the only remaing ways to improve performance are to tune the flow rate to the radiator, or increase the cfm.

Since a large number of people watercool for silence as well as performance, increasing the cfm is not always possible.

As fot the airplex evo being restrictive.

compare a radiator with a tube folded bacwards and forwards down the length of the rad, such as the airplex evo, with say a thermochill of the same size.

The airplex is a multipass design, where the effectively has to get through a very long narrow pipe. The thermochill is a dual pass, ie once up and once back down, with several channels which the water can flow down.

That if I am not mistaken is a less restrictive design.

My point was that depending on the waterblock used, you may be better off with the thermochill.

An example, the ocpc atlantis.

If you look at BillA's testing, (here), you can see that this block needs quite a lot of flow to perform to the best of its abilities. (look at c/w vs flow rate.) The gradient is quite steep, implying you get quite a benefit for pushing a lot of flow through it, and it actually performs quite well with A LOT of flow. Around 10lpm would be good.

Compare this with the white water, which has a better efficiency than the ocpc atlantis, even with only 2lpm flowing through it.

My point here is that if you want to use an atlantis with a tube and fin style radiator (generally more restrictive), then you would do better running two in parallel, since the optimum flow rate for most radiators is somewhere between 4 and 6 lpm. If you had this running through each rad in parallel, hooked up to an atlantis, then both the rads AND the waterblocks will be operating with their most efficient flow rate.

I also agree that if you have two dissimilar rads, and run them in parallel, then the flow will not be the same through each of them, but if you run them in series, then you will end up affecting the efficiency of the waterblock.

Hence I always encourage people to choose components which are well suited to one another.

8-ball

 

8-ball - Read the thread again.
He has neither an Atlantis nor a WW.

What you are saying is completely irrelevant.

Unless you are recommending that he goes and changes both his block and radiators, your post has no usable input.

Just because you believe that water may flow better through another particular design (similar to the older Airplex) does not make this design "very restrictive" nor does it harm the cooling ability of the system as a whole.
I will probably be stocking the Thermochill range of radiators soon myself anyway but that doesn't help the original poster, does it?
It's just another unfounded attack from you on this company's products. Perhaps I should start keeping track of how many times I have actually seen you do this.

As it happens, to my knowledge, the ThermoChill series only really make a single pass, as they only use half of the radiator per transition (unlike the older Airplex which makes a full pass each time - thus being even less restrictive still by your reckoning) but obviously you know this, having had first hand experience with all the components you comment on, rather than just working from pictures and dreaming up supposition based on what you think you see.
I've been in touch with ThermoChill for a while now and am waiting for BillA's results with their products (Data will be published on overclockers.com shortly, further technical data will be published on thermochill.com shortly after.) before deciding to stock them (subject to my own findings, of course).

You should stop working from theory and just post based on your personal experience - sure, your post count would drop dramatically but we'll all appreciate the input more and won't have to waste time debunking your theories or wading through your homage to BillA.

You would then find that a system is more than the sum of its parts not just the block.
You are constantly taking information provided for optimal waterblock design instead of complete system optimisation.

The best way to optimise flow for a waterblock is to do away with the radiator altogether.
Why don't you think on the repercussions of that action for a while?

 

Pug,

I agree that I can be a little theoretical, but I plan to change that very soon. (Soon as I've got my finals out of the way. 13 days to go!!! )

The thermochill rads are dual pass, same as the more recent black ice.

Single pass, involves water travelling once through the core, using the full width of the rad. These rads have the inlet and outlet at opposite ends.

I'd quite like to do a full radiator roundup sometime this summer, autumn, depending on when I can get the kit together.

Would like to compare the various dual 120mm rads. There's quite a few out now, so it would be quite interesting.

Black Ice extreme 2.
Black Ice 2
Thermochill 120.2
Air plex 240
Air plex evo 240

and many more.

I wouldn't do it unless I could do full pressure drop comparisons at varying flow rates, and the resulting efficiencies, but it would be quite nice to get an idea of how the flow rate affects these rads.

Also nice to get an idea of relative performance.

"
quote:
--------------------------------------------------------------------------------
Obviously, more rad surface area is better, but their comes a point when bigger isn't possible, and it is necessary to make the best of what you've got.

8-ball
--------------------------------------------------------------------------------

Exactly, and he has TWO radiators which makes more surface area both bigger and possible

"

I understand that, but he asked how he could improve the flow rates, given that setup. I told him.

 

I would say that it's an imperative prerequisite before you go bandying about the kind of statements you've been making.

Did you not stop to think that there are people here who have had hands on experience with some of these things?
Thanks for finally admitting that you've had none though.

Emphasis mine.

2 x 1/2 = 1, last time I checked.
(Although admittedly, this is just being pedantic about semantics )
The distinction here is fairly grey (and not the real problem anyway) and you aren't the only one to see it that way.
My answer to that is, so what? The Zern rad is a single pass example anyway, afaik.

I was merely illustrating that imho, the older Airplex is a truer example of what should be considered a real dual pass example, as it travels both passes through the full width, not half... but we digress.

As did I, without potentially, one would hope, seriously reducing his actual real world system performance.

Do you really think you know where the exact sweet spot lies between combined air flow, water temp, flow rate and Delta T?
Please, share your findings. I'm sure many people would love to know.

I can't believe you can give people a hard time for using the onboard diode for comparison between temperatures attained by different setups on the same motherboard (although air vs water is arguably less accurate this way) but then turn around and justify your own theoretical suppositions with "So you've lowered the water temp a bit more".
Way to be exact, 8-Ball

No, as I implied, if you introduce stagnant flow, you've effectively disabled one of his rads, so you may as well be talking about one rad versus two anyway.

Also, you are thereby effectively disabling the extra fans which can make a much greater difference to the system, introducing a temperature gain, easily measureable by on board diodes -
Unlike the difference achieved by optimising the flow through the block to a degree that you actually need the high class equipment to tell the effing difference!!.


Have you ever experimented with this?
[Off-topic] What rig do you actually run anyway? I don't think I've ever seen you post your own cooling setup.

Here's an easy practical assignment for you, if you're interested.
Get one of those shower assemblies that attaches to a pair of taps.
Remove the end that goes over one of the taps and connect the other to your tap, so that you have a short length of hose on one side of the T and the long length, complete with shower head on the other.
Turn the tap on (no, the one that the hose is connected too! )
See which end all the water runs out?
Notice how NONE actually comes out of the shower head?
Get my point now?
Extreme example, maybe but perfect illustration of what I am trying to get across to you.


It wouldn't take much for me to add a second rad to my loop and test the different configurations in serial and parallel. I even have a 1046, 1048, 1250 and an adjustable 800LPH pump to test your theories on flow rate, together with both 6 & 10mm ID tubing to recreate this setup as far as possible.

Do you know what stops me though (apart from a lack of Zern parts)?

BillAboys - who go around criticising people for not having £20K's worth of surplus test equipment on hand to keep the students entertained.
I could waste a lot of valuable leisure time on this to be discredited by a single dismissive statement.


Actually, I can't believe I'm still sitting here typing this when I have three 2GHz gaming machines with a fresh copy of GTA: VC in the drive of one and the last level of Splinter Cell to complete on another...
I finish work for the day and spend my time writing this?
My beer's run out, I'm off.
________________________________________

Good luck with it, Der Me!$ter.

I'll order some of those brackets in and check out if there is a way of using larger hose with them.
I will see if I can get some modified ones produced.
Thanks for pointing out the problem, I wasn't aware of it.

If not, I may be able to help you with selection of suitable connectors once I've seen it.
You can always get them direct from AC if they do them - may as well take advantage of the free shipping as you're in Germany.

I know that there are convertors from the P&C to 10mm hose available because I've sourced them for someone before for this rig -



He used them to integrate a 3/8" flow meter into the loop, which you can faintly see above the Airplex rad housing.
The connectors may not have been from AC though, iirc but they may have some alternative ones.

Obviously, if we can lessen the amount of 6mm tube as far as we can, it will help matters even if we can't isolate it completely.

 

Looks like a ghostbuster pack...

 

That was so fun to read ... thanks guys ... keep on going.

Cheers

[DE]FreD_S

 

That is the coolest rig i've seen in a LONG time!

 

think it looks very tacky tbh. just looks like they've thrown in lots of bright pretty things all together, thrown in a bit of plexi and hung it on the wall.

 

I don't know, I find it kinda fun, if a bit impractical. Theres a lot of work gone into it, but I would imagine its a bit noisy despite the WC as the HDD is out in the open, plus there's three fans pointing strait at you, which isn't ideal. As as for lighted fans, I'll rather stick to my all-Papst approach.

 

I tried to find another way to connect bigger tubing to the airplex and came up with the idea of connecting 13mm tubes over the plug and cool connectors. Will this do any good ? Even if I could somehow connect 10mm tubing to the radiator, I don't think it would be of much use, as the tubing inside the radiator is around 6mm. Do you think I should try it with the 13mm instead of 6mm and then reduce it to 10mm inside the case?