Inline Chillers?

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
You shouldn't need any insulation at all.
If you have a radiator in series with it, the coolant will never be able to go below ambient temperature.
 

BrainEater

[H]ard|Gawd
Joined
Jul 21, 2004
Messages
1,087
It should go below ambient I'd hope....insulation will only help.

---

Yep I sure have taken the temps etc....the wires for the pelt get hotter than the heatsink does ....

Here's the last run : (no load )

Pelt voltage : 11.5 volts
Current : 12.5 amps

Ambient temp 23 C
Coldblock temp -2 C
Temp on the base of the TRUE : 31 C

Now the pelts only running about 145 watts or so , but 8C above ambient seems like pretty impressive dissapation to me.

/me ups the voltage.

:D
 
Last edited:

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
Hi Brain,
Look at it this way.
The second your coolant drops below ambient, the radiator will heat the water back up to ambient temperature.
 

BrainEater

[H]ard|Gawd
Joined
Jul 21, 2004
Messages
1,087
Ah yes , I see what you are saying.

I had intended to run this *after* the rad.

The radiator will dump the heat , the chiller goes from there.

Whether or not the chiller can reduce the loop temp (right before CPU)below ambient, remains to be seen.

Once I get the cooler running at least 200 watts we'll see......

The system I'm going test this on is a phenom 2 920.Basic WC setup,cpu only , single rad.

There's a lotta work yet to be done.The cooler setup was rushed a bit ,for sake of 'cool pics' . I need to fine tune some stuff, one of the issues being I don't have a sensitive enough torque wrench to check the tension on the assembly.

I also have to build some temp probes so I can get temps along a waterloop in various places.

work continues.

:D
 

BrainEater

[H]ard|Gawd
Joined
Jul 21, 2004
Messages
1,087
Gonna pull the trigger on a 437 watt pelt.....fit's the unit better , and should run more efficiently......

:D
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
Yep, more cooling at a lower voltage... That was my main reason for wanting to try this with a 437Qmax! ;)
 

flogge

[H]ard|Gawd
Joined
May 20, 2008
Messages
1,888
One of two outcomes:

1: water coming out of the CPU will still be below ambient, therefore the radiator does nothing or adds heat.
2: Loop flow rate will be low enough that water temps in various parts of the loop can be noticeably different.

I am curious and I hope you have the ability to slow down your flow rate of coolant, it may improve your CPU waterblock inlet temp.
 

flak-spammer

Limp Gawd
Joined
Aug 24, 2007
Messages
262
Ugh that's a bunch of BS. A loop is a loop if you slow down the coolant it takes longer to cool down and it also takes the same time to heat back up, therefore a net gain of 0. Ask anyone in watercooling and they will tell you that faster coolant flow is better.
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
Ugh that's a bunch of BS. A loop is a loop if you slow down the coolant it takes longer to cool down and it also takes the same time to heat back up, therefore a net gain of 0. Ask anyone in watercooling and they will tell you that faster coolant flow is better.

What will absorb more energy, a gallon of water passing over a surface in a minute, or half a gallon in the same time period?

More volume of water passing over the surface = more capacity to hold heat energy.

Higher flowrate(given the pump isn't pouring heat into your system to achieve said flowrate) will always be better than lower flowrate.
 

flak-spammer

Limp Gawd
Joined
Aug 24, 2007
Messages
262
I know I've done a few tests myself it doesn't work the way people want when you slow down the fluid. Same goes with air cooling which is why I said fluid.
 

flogge

[H]ard|Gawd
Joined
May 20, 2008
Messages
1,888
If you manage to sub-cool the fluid enough that the CPU block outlet is sub-ambient then the radiator will actually warm up the fluid back to ambient before going back to the chiller. In this case the slower fluid will actually serve to reduce the efficiency of the radiator which is fighting the system anyway. I was in need of sleep when I wrote that last night.
 

Dekar12

Gawd
Joined
Oct 2, 2003
Messages
745
Reading the OP's post it brings back memories of things that I had thought about doing at one point or another. After working with my first WC'ing I came to the conclusion that while it would be a neat project, it is more practical to add a radiator and/or adjust fan speeds.

However, if BrainEater is attempting to go below ambient with this then most likely he will need to just remove the radiator from the loop. Because the radiator is going to be adding heat back in as it goes below ambient which makes it harder to go below, etc.
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
Reading the OP's post it brings back memories of things that I had thought about doing at one point or another. After working with my first WC'ing I came to the conclusion that while it would be a neat project, it is more practical to add a radiator and/or adjust fan speeds.

However, if BrainEater is attempting to go below ambient with this then most likely he will need to just remove the radiator from the loop. Because the radiator is going to be adding heat back in as it goes below ambient which makes it harder to go below, etc.

Right Dekar. We've been saying this the entire thread. The object is not to go below ambient. It is in fact to just get closer to ambient, and avoid needing to insulate anything.

Distilled down to pure ideals, I think the goal is to find out if we can use a peltier to more efficiently dissipate heat than an additional radiator would, in the same amount or smaller amount of physical space.
 

BrainEater

[H]ard|Gawd
Joined
Jul 21, 2004
Messages
1,087
Distilled down to pure ideals, I think the goal is to find out if we can use a peltier to more efficiently dissipate heat than an additional radiator would, in the same amount or smaller amount of physical space.


For 5 times the cost.

:D

--------

Sorry for the lack of updates on this.

I just had a basement flood , and well the last 5 weeks has been rebuilding that.

This project is on again now though.

---------

I want to make some definitions clear.

The term 'below ambient' is vague.

The output side of the CPU block won't be below ambient .I don't think it will work that well , but we'll see.However , the inlet side (peltier block output) very well might be.

The radiator will not be adding any heat.Once again I doubt It'll work that well....the radiator will be the primary heat dump , the chiller subtracts from the radiator output temp.

:D
 

flogge

[H]ard|Gawd
Joined
May 20, 2008
Messages
1,888
If you do manage to get the peltier block outlet below ambient, you do have the possibility of condensation dripping off the hose running from the peltier block to the CPU block. It won't matter if the CPU itself never goes below ambient, you still have something that could be cooler than the dew point inside your case.
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
If you do manage to get the peltier block outlet below ambient, you do have the possibility of condensation dripping off the hose running from the peltier block to the CPU block. It won't matter if the CPU itself never goes below ambient, you still have something that could be cooler than the dew point inside your case.

The water will be spending so little time in the peltier block, there's no way it would be able to drop the temps below ambient.

You just don't get it. One of the primary objectives, if you read the thread, is to design this specifically in a way to never get condensation.
 

flogge

[H]ard|Gawd
Joined
May 20, 2008
Messages
1,888
According to your logic: The water spends so little time in the CPU block, how could it ever be warmed above ambient? If he has a radiator cooling the water to near ambient temp, then the peltier block certainly does have the potential to cool the water further below ambient. I realize that the objective of the idea is to avoid condensation. how does that make it wrong for me to point out the possibility? You don't have to cool the CPU itself or the waterblock below ambient to have a problem. If the water and the hose are at ambient -5C, you could drop below the dewpoint and condensation occurs.

ambient is ~25C. Lots of WC rigs boast load temps of ~50C or lower. I saw frost on some of those pics, so that block was under 0C at no-load. I'm betting that if you push 25-30C water through a -10C waterblock it has a chance of coming out below 25C.
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
According to your logic: The water spends so little time in the CPU block, how could it ever be warmed above ambient? If he has a radiator cooling the water to near ambient temp, then the peltier block certainly does have the potential to cool the water further below ambient. I realize that the objective of the idea is to avoid condensation. how does that make it wrong for me to point out the possibility? You don't have to cool the CPU itself or the waterblock below ambient to have a problem. If the water and the hose are at ambient -5C, you could drop below the dewpoint and condensation occurs.

ambient is ~25C. Lots of WC rigs boast load temps of ~50C or lower. I saw frost on some of those pics, so that block was under 0C at no-load. I'm betting that if you push 25-30C water through a -10C waterblock it has a chance of coming out below 25C.
Please link all of these watercoolers 'boasting' load temps below 50C. You'll either be showing us a liar, or someone who spent $1000+ on their cooling, not someone who spent less than $400 and is looking to eek a little more out of their loop in a small space.

Read the thread. It is clear you have not read, comprehended, or understood the goals and conditions surrounding the initial inquiry.

The whole point is that you are _WRONG_ about the output between the pelt and cpu block being able to get anywhere below ambient. If you had read the thread, you would know why.

1 Gallon of water = 3870 grams.

The energy required to lower 1 gram of water by 1C is ~4.2 joules.

At an approximate 1GPM, that's (4.2*3870)/60 for 270.1W of power required to drop the coolant by 1C. Which is basically the entire power of the 437Qmax 24v peltier running at 12v.
 
Last edited:

flogge

[H]ard|Gawd
Joined
May 20, 2008
Messages
1,888
Here is a list of air coolers keeping a 1.45V i7 under 50C on the heatspreader. I certainly hope a WC rig can outperform a $50 air cooler.
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
Here is a list of air coolers keeping a 1.45V i7 under 50C on the heatspreader. I certainly hope a WC rig can outperform a $50 air cooler.

Nice list of nothing. A heatspreader measurement doesn't mean anything against an on-die T-junction measurement. Way to go for measurements of the hunk of metal on top of the transistors, instead of the transistor temperature itself. Stop making excuses.
 

flogge

[H]ard|Gawd
Joined
May 20, 2008
Messages
1,888
I've re-read the entire thread trying to find this blinding evidence that sub-ambient temps are impossible. I didn't find it. I see you discounting the possibility with words but not facts on multiple occasions.

I don't know what the die temps are on an average WC loop, but the heat load as far as the water is concerned is the block, which takes heat from the heatspreader, so the block metal will be under 50C unless you did it wrong.

According to triple rad review http://www.hardforum.com/showthread.php?t=1467444 for a 2 degree delta at 1800 rpms the average wattage dissipated by these rads is as high as 136W.

I'll use these givens
Ambient=23C
flow = 1 GPM
CPU load = 135W
block temp = 40C(based on air cooler review of 135W stock CPU)

If your water going in to the CPU is at ambient, which is optimum performance, and the CPU is making 135W(half of 270) and flow is 1 GPM, then the water would only gain .5C on its way through the CPU. Ignoring the radiator because of the tiny DeltaT, the pelt block must dissipate 135W to bring the water back to ambient. The 437W pelt at 12v makes ~220W, so the TRUE must dissipate 355W.

If the TRUE could dissipate 144W(11.5Vx12.5A) with its base at 31C, how hot would it have to be to dissipate 355?
 
Last edited:

flogge

[H]ard|Gawd
Joined
May 20, 2008
Messages
1,888
Whatever that temperature is on the hot side of the peltier, subtract the Delta-T of the peltier. This will be your cold-side block temp. average this temp and the CPU-block temp and you will find your final water temp neglecting the radiator.
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
I've re-read the entire thread trying to find this blinding evidence that sub-ambient temps are impossible. I didn't find it. I see you discounting the possibility with words but not facts on multiple occasions.

I don't know what the die temps are on an average WC loop, but the heat load as far as the water is concerned is the block, which takes heat from the heatspreader, so the block metal will be under 50C unless you did it wrong.

According to triple rad review http://www.hardforum.com/showthread.php?t=1467444 for a 2 degree delta at 1800 rpms the average wattage dissipated by these rads is as high as 136W.

I'll use these givens
Ambient=23C
flow = 1 GPM
CPU load = 135W
block temp = 40C(based on air cooler review of 135W stock CPU)

If your water going in to the CPU is at ambient, which is optimum performance, and the CPU is making 135W(half of 270) and flow is 1 GPM, then the water would only gain .5C on its way through the CPU. Ignoring the radiator because of the tiny DeltaT, the pelt block must dissipate 135W to bring the water back to ambient. The 437W pelt at 12v makes ~220W, so the TRUE must dissipate 355W.

If the TRUE could dissipate 144W(11.5Vx12.5A) with its base at 31C, how hot would it have to be to dissipate 355?

So, my math and science, calculating the amount of power required to change water's temperature at 1GPM isn't blinding evidence?

You change your story again. Way to drop down to an unoverclocked CPU, because that's what we do here, watercool unoverclocked CPU's below ambient.

You also admit to knowing nothing about watercooling, and you clearly know NOTHING about the typical loop. Average loop deltas are between 5 and 10C over ambient. Most are near 10C, and with my heavy overclock I'm at a 12C delta on my triple rad. Why are you posting all this FUD?

Also, if you knew ANYTHING about ANYTHING you would know that TDP doesn't mean ANYTHING and processors frequently exceed TDP in both power consumption and heat output. Especially overclocked, because the second you raise the voltage you've left the realm of measured TDP, by a lot.

None of your questions or anything you have said changes the fact that in order to change the coolant of a system at 1gpm by 1C, you need 270w of power(watts are joules/second).

I won't be responding to any more of your posts any more because you are clearly here just to argue even though you are flat out wrong. You already admitted to knowing nothing. Hell, you don't even watercool. Which makes what you say worth... oh right. Nothing at all.
 

flak-spammer

Limp Gawd
Joined
Aug 24, 2007
Messages
262
Geez..guys take it easy. However Zoson is right, you won't make a huge difference on the temperature. In any single pass a TEC loop (with no rad) might drop the water temp by .01C, and that number is really depended on TEC power, water velocity, and TEC block surface area. That being said in this particular setup it would be physically impossible for a loop to undergo a 10 degree delta between water in and water out of the TEC block. The only way this would be possible is if the water was moving at .01GPM or so. In which case why watercool at all.
 

vjcsmoke

Supreme [H]ardness
Joined
Dec 5, 2006
Messages
4,511
Geez..guys take it easy. However Zoson is right, you won't make a huge difference on the temperature. In any single pass a TEC loop (with no rad) might drop the water temp by .01C, and that number is really depended on TEC power, water velocity, and TEC block surface area. That being said in this particular setup it would be physically impossible for a loop to undergo a 10 degree delta between water in and water out of the TEC block. The only way this would be possible is if the water was moving at .01GPM or so. In which case why watercool at all.

So how many degrees does the TEC actually drop the temp of the ENTIRE water cooling loop? It sounds like a lot of energy and money spent to drop temps by a few C, when just adding another triple or quad radiator would probably achieve the same result.

Am I the only one who has observed this?
Any updates on this experiment?
 

Zoson

Supreme [H]ardness
Joined
Aug 4, 2001
Messages
6,113
So how many degrees does the TEC actually drop the temp of the ENTIRE water cooling loop? It sounds like a lot of energy and money spent to drop temps by a few C, when just adding another triple or quad radiator would probably achieve the same result.

Am I the only one who has observed this?
Any updates on this experiment?

You clearly missed the entire point of this conversation. The thread was started in the context of 'I only have a tiny bit of space left inside my case, but I want to fit more cooling in - damn the cost.'

Not everyone wants to have a huge watercooling loop outside of their case. Some of us seek elegance and professional looks.
22377_490877725296_612230296_11169994_3786805_n.jpg

22377_491144720296_612230296_11173017_5863460_n.jpg


No triple or quad is going to fit INTO my case.
 

Emission

Supreme [H]ardness
Joined
Dec 6, 2005
Messages
4,420
Any updates on this experiment? I've been thinking about doing something like this, but in a grander scale that could be more effective depending on how you look at it.

I've been thinking about using several small TECs (or maybe several larger TECs depending on how the wattage numbers work out) attached to one water block and one air-cooled heatsink each, the air-cooled sink would go on the hot side and the water block would go on the cold side, and the total heat load would be distributed among maybe 2 or 4 of them. This way, you can use cheaper, smaller air-cooled heatsinks, and you would draw less power due to the increased efficiency of running each TEC at a relatively low voltage.

I apologize if it is off-topic at all, trying to throw some ideas around and help you guys somehow. I know my solution might be too large for Zoson's setup, but what does everyone think of it?
 
Top