Circulating Bubbles. Help?

UnusualSuspect · 15 Nov 2007 at 19:28

Hey,

I've got my watercooling setup all ready to go, but I've run into one problem with it. The pump seems to be circulating small air bubbles. If I do the switch on/switch off thing then the bubbles slowly dissipate, but as soon as I leave it running for about 30 seconds or so, the bubbles start to reappear. The reservoir makes weird gurgling sounds and the bubbles start travelling up the first tube again.

For a pump I'm using the Laing DDC with Alphacool top and res bay, which I figure is the problem. As the water churns about from having it thrown at speed into the bay, the pump is sucking down the occasional bits of air and blowing them back into the tubes.

Is there any way of stopping this? I figure if I can't sort this out then I'll put in a T bend, strip out the res and use the Laing purely as a pump. It just seems a bit weird that this wouldn't have been designed out of the whole setup. Maybe the water is just travelling too fast? Is there a way to slow it down?

Any help would be appreciated, I'm at my wits end here. On a side note, however, I put the loop into the computer for a test run, only plugged in the necessary stuff - motherboard, graphics card - and in the BIOS I was getting a rather satisfying 28-32c temperature reading. This frankly amazes me, as previously on air it was sitting a good 10-15c higher. Awesome.

For reference: Res/Pump -> GPU -> CPU -> Rad -> Res/Pump

I get a lot of bubbles coming out of the GPU - EK8800 - they appear like a small stream travelling down the centre line of the tube and into the CPU block.

YZFAndy · 15 Nov 2007 at 19:35

Bubbles will disappear and be purged from the system given time. I disturbed mine two weeks ago and still occasionally here the odd bubble get nudge out of one of the radiators.

One thing I would say is your loop doesn’t seem to be right. I would say it needs to go res/pump > Rad > CPU > GPU > Res/pump.

UnusualSuspect · 15 Nov 2007 at 19:41

YZFAndy said:
Bubbles will disappear and be purged from the system given time. I disturbed mine two weeks ago and still occasionally here the odd bubble get nudge out of one of the radiators.

One thing I would say is your loop doesn’t seem to be right. I would say it needs to go res/pump > Rad > CPU > GPU > Res/pump.

How long roughly? I mean, I have a ton of tiny bubbles taking up residence here.

And I read that the design of the loop doesn't really matter, because the temperature is going to stay the same regardless of which goes where. However, if you think it will make a difference, I can just reverse the tubing in/out of the pump. Read my loop backwards.

Mint_Sauce · 15 Nov 2007 at 20:13

Is it possible you have a tiny leak? Somewhere air is just making its way in to the system?? Don't want to scare you but it might be worth getting some loo roll and wiping around all the joins to see if you pick up any moisture.

UnusualSuspect · 15 Nov 2007 at 20:37

Mint_Sauce said:
Is it possible you have a tiny leak? Somewhere air is just making its way in to the system?? Don't want to scare you but it might be worth getting some loo roll and wiping around all the joins to see if you pick up any moisture.

Bingo! That part I mentioned, where theres a spiral of air bubbles coming out of the graphics card. I found a tiny spot of water there. Probably the barb needs tightening slightly. Will try that next.

YZFAndy · 15 Nov 2007 at 21:11

I would say the order of the loop is important.
First of all although tiny the pump will create some heat so best that goes straight into radiator.
Water straight out of the rad is going to be the coolest so personally I would put that into the CPU cooler. Main reason being that your graphics is going to heat up the water much more than you’re CPU and you don’t want to be trying to cool your CPU with warm water. Your graphics will tolerate the high temperature a lot more than your CPU will.

UnusualSuspect · 16 Nov 2007 at 12:35

So, I changed the loop so its Res/Pump -> Rad -> CPU -> GPU -> Res/Pump.

Still getting the same problem. The amount of air bubbles is insane, especially coming out of the GPU block. No idea why this is happening, I'm thinking of just completely removing the GPU from the loop and sticking the Chipset block in instead. I kept it out originally because I only have a 1 fan radiator and figured the temperatures wouldn't be worth it.

I'm going to try two things. 1) Remove the GPU block and insert the Chipset block in its place. If I still get bubbles coming from the pump then I'm going to do 2) Remove the res from the pump and put in a T section. And if that still doesn't work, I give up.

With the GPU and CPU in at the moment, I'm getting temperatures of 46c GPU and 37c CPU, which isn't too bad. 10c off what I used to have. The noise from the pump is just too much though, with the air being channeled through it.

Just as a side question. The 3 pin on the pump. Is that meant to be plugged in to anything, or is it an optional source of power? Not sure what to do with that.

Well Splattered · 16 Nov 2007 at 12:48

UnusualSuspect said:
Just as a side question. The 3 pin on the pump. Is that meant to be plugged in to anything, or is it an optional source of power? Not sure what to do with that.

I suspect that the 3 pin cable is for RPM monitoring. Seen in Bios and any Windows monitoring software. Yellow and black wires only?

Air usually finds the highest part in the loop and will collect there. I would suggest rotating the case so all the air is trapped in the res and then top it up.

YZFAndy · 16 Nov 2007 at 18:38

If the bubbles aren’t getting any better you must be getting air into the system some how. Is the bump at the lowest part of the loop? And is there enough fluid in the res?

Tooks · 16 Nov 2007 at 19:39

Loop order makes zero difference to the cooling capability of your wc system, or so small you can't really measure it. It's another of those watercooling myths!!

It's a closed system, with a finite ability to shift heat via the rad. Where you put heat into the loop, and in what order, makes no odds as you're still dumping the same watts into it somewhere before it hits the rad. I've tried all sorts of configs, and they never really change temps.

What IS important is minimal restriction, and as short a length of tubes as possible, so I always route my loop with that in mind.

I've got the same pump, and it takes a couple of days for all the bubbles/foam to disappear, depending on what additives you've got in your water.

gurusan · 16 Nov 2007 at 20:34

^
The man speaks the truth.

YZFAndy · 16 Nov 2007 at 20:41

Tooks said:
Loop order makes zero difference to the cooling capability of your wc system, or so small you can't really measure it. It's another of those watercooling myths!!

It's a closed system, with a finite ability to shift heat via the rad. Where you put heat into the loop, and in what order, makes no odds as you're still dumping the same watts into it somewhere before it hits the rad. I've tried all sorts of configs, and they never really change temps.

What IS important is minimal restriction, and as short a length of tubes as possible, so I always route my loop with that in mind.

I've got the same pump, and it takes a couple of days for all the bubbles/foam to disappear, depending on what additives you've got in your water.

No your wrong. The total end result is the same but where the most heat is in the loop is different. What doesn't make a difference is the speed of the flow. For example a slow flow will speed twice as much time in the rad and cool more where as a fast flow will speed less time in the rad but be in the rad more often.

gurusan · 16 Nov 2007 at 21:01

YZFAndy said:
No your wrong. The total end result is the same but where the most heat is in the loop is different.

Your statement is only correct for maybe a few minutes after the system is first turned on. Very shortly the closed system will reach equilibrium.

YZFAndy said:
For example a slow flow will speed twice as much time in the rad and cool more where as a fast flow will speed less time in the rad but be in the rad more often.

oh...so very wrong.

The faster you can move heat away from the CPU (and remove heat from the radiator), the closer to ambient the closed system will become. What is it that we are using to carry the heat away? Water molecules....so if you increase the mass flow rate you will realize a temperature drop.

Of course there's a point of diminishing returns in a PC watercooling setup...but the principle stands.

YZFAndy · 16 Nov 2007 at 21:36

gurusan said:
Your statement is only correct for maybe a few minutes after the system is first turned on. Very shortly the closed system will reach equilibrium.

oh...so very wrong.

The faster you can move heat away from the CPU (and remove heat from the radiator), the closer to ambient the closed system will become. What is it that we are using to carry the heat away? Water molecules....so if you increase the mass flow rate you will realize a temperature drop.

Of course there's a point of diminishing returns in a PC watercooling setup...but the principle stands.

Equilibrium!!!

So what you’re saying is that each component having its heat removed isn’t adding in steps but is a total throughout the loop. No that’s wrong.

Yes for example -100watts (radiator) + 100watts (CPU) +100watts (Graphics) equals 100watts total no mater what way the loop is. But heat will be added at each heat source.
A fast flow will minimise this affect but ultimately and a cpu will be better off being cooled at the coolest point, just after the radiator because it will accept less total heat that a graphics card.

Q=cmΔT

gurusan · 16 Nov 2007 at 21:41

I guarantee you that at the flow rates and heat sources we are dealing with there will be at max a 1c temp difference between any 2 points in the loop.

weird_dave · 17 Nov 2007 at 12:23

UnusualSuspect said:
Hey,
The reservoir makes weird gurgling sounds and the bubbles start travelling up the first tube again.

As the water churns about from having it thrown at speed into the bay, the pump is sucking down the occasional bits of air and blowing them back into the tubes.

Well, you've answered it yourself really.
Deeper water will solve this.

As for the arguments on water speed, if it's slower the water will be warmer coming off the CPU and therefore the GFX will be warmer. If you were only cooling 1 item (or parallel cooling) then you'd struggle to spot much of a difference, unless you pumped it really slow.
play with 4200 joules per litre per degree C, power output and water speed and you can work out the theoretical temperature rise at each stage.
having just looked at the flow rate for the pump, you'd be lucky to get 0.1 L/s, so assuming the CPU is belting out 105W, the temperature rise would be 0.25 degrees C.
The biggest problem is cooling the water, I would suggest putting the rad in a pond and use it as a heat exchanger

you might need a better pump tho to overcome all the frictional losses in all thet extra pipework.

mnemonix · 17 Nov 2007 at 12:24

gurusan said:
oh...so very wrong.

The faster you can move heat away from the CPU (and remove heat from the radiator), the closer to ambient the closed system will become. What is it that we are using to carry the heat away? Water molecules....so if you increase the mass flow rate you will realize a temperature drop.

Of course there's a point of diminishing returns in a PC watercooling setup...but the principle stands.

No, sorry, but this is one of the most frequent misunderstandings. Given everything else is constant, increasing the flow rate makes zero difference... water might be passing faster through the block carrying heat with it, but it's also passing through the radiator faster, giving it less time to release that heat. There is a cut off below which your system will not be running at it's potential (consider a flow rate of zero as an extreme case) but above this running your pump as slow is possible is preferable to minimize heat dump from it.

mnemonix · 17 Nov 2007 at 12:33

To the OP, a leak is a possibility, also depending on your reservoir, a small amount of air trapped at the top of a drive bay reservoir combined with a high pump speed and turbulance in the top of the tank can cause air bubbles in the loop too. Is your pump running at full speed ? The D5 can be adjusted, not sure about yours though assuming it's a 3.2 ?

Tooks · 17 Nov 2007 at 19:27

Just thought I'd post this useful link for those still struggling with how pc watercooling works!!

http://www.overclockers.com/articles1088/

gurusan · 17 Nov 2007 at 19:33

hmm..interesting.

I admit I was wrong about the flow rate thing. I realize now that I was thinking of it as an open system.