A different focus in liquid cooling

S

Squalish

Gawd
Joined
Aug 23, 2002
Messages
647
I havn't really been following this topic for years, and am making my annual "What would it cost to bring my system up to date" inquiry. This year, I'm considering liquid cooling. I've replaced maybe a dozen fans in the 6 years I've owned this system (which is getting increasingly haphazard), and I wouldn't be able to play anything if my ti4200 orb-style cooler actually required its fan to function. I've got a marginally newer computer lying dormant because the 6800GTS makes an intolerably loud whine.

Last I checked (and it's been a while), the ΔT of an average crappy kit-based watercooling was still 1/10 that of an excellent, expensive air cooler. I would be fine with the aircooler, if I wasn't concerned with reliability & noise... so I have a lot of headroom.

My needs are:
As cheap as possible (I'd prefer <$200, but ...)
No welding
Near-silence (sleepability)
Extremely reliable/safe
Redundant Fans (& pumps if necessary)
Orderable online (no more junkyard diving)

I can deal with:
Crappy performance
Odd fan/radiator sizes (ripping no-brand fans out of other products, etc, as long as I can get cheap replacements)
Chopping up my case (do people still Dremel?)
Hazardous chemicals (The biodegradable, nontoxic, 'green' biocide really makes me want to strangle a marketting rep)

I'm looking for something more quiet, easier to maintain, & better performing than the stock aircoolers. Most likely I'll have a moderately overclocked q6600 & 2x 4850's in the box. I have a Lian-Li-fulltower case and a 500W Neopower PSU(it's enough) ready to go. I'd prefer to have all the heat routed outside the case to an external radiator... possibly mounted in a push-pull duct with sound-absorbent carpet or something similar.

My main question is durability, reliability, & safety related. I'd like a liquid cooling solution that, after the initial leaktesting, will last me five years of continuous running without maintenance, and won't damage anything if it does spring a leak. I'd like it to be solid enough that it can be moved around, and I never, ever, ever want to have to worry about corrosion/algae/soap blocking a waterblock/pump and destroying hardware.

Again, cooling ability is not an issue here, because I don't feel any need to spend money pushing my PC an extra 5% if it means I'll need to push back upgrades for another year or two... it's just a means to an end: reliable, silent operation, which I'm ready to pay a little extra for.

How would one do this?
 
Less than 200$, and the need to cool a CPU and 2 GPU's?

Eh... I'm afraid that's not going to happen. :eek:
 
Actually, it is possible, if you buy used parts. Look up www.xtremesystem.org. Lots of good "for sale" threads. Just note it takes a long time for you to get registered. If you like, I'll PM fugger/movieman to activate your account.
 
Arcygenical said:
Less than 200$, and the need to cool a CPU and 2 GPU's?

Eh... I'm afraid that's not going to happen. :eek:
Click to expand...

...really...:rolleyes:


...and the op, five years without maintenance...? Stick with air cooling. The Thermalright Ultra 120 Extreme can handle your CPU even with a very quiet fan. Leave the OEM coolers on the cards and change your case fans with more of the same quiet fans. This would meet all your requirements, without water.
 
young boy said:
Actually, it is possible, if you buy used parts. Look up www.xtremesystem.org. Lots of good "for sale" threads. Just note it takes a long time for you to get registered. If you like, I'll PM fugger/movieman to activate your account.
Click to expand...

You could start buying used parts and the potential for a mishap greatly increases. This guy already doesn't want to maintain the system, so there's no reason to even think of buying second hand parts here.
 
I understand it's less than practical, and doesn't fit with usual goals, I'm asking what would make it possible.

I assume that the marketers of speciality non-conductive/non-corrosive coolants *want* you to change them annually, but are there actual physical requrements to do so in a sealed system?

Has anyone around here put together gift systems for people they weren't prepared to teach maintenance to?
 
One other thing to think about is that unless you live in a clean room environment, you WILL have to blow out your radiator(s) regularly. That is something that I'm sure a lot of ppl don't think about. So if that falls under your non-maintenance for 5 yrs deal, that would be a killer right there.
 
roary said:
One other thing to think about is that unless you live in a clean room environment, you WILL have to blow out your radiator(s) regularly. That is something that I'm sure a lot of ppl don't think about. So if that falls under your non-maintenance for 5 yrs deal, that would be a killer right there.
Click to expand...

I wouldn't think that's any different than any other computer with fans. As far as the xtremesystems.org forums go, you need 100 posts to access the buy/sell forums.
 
ATLPIMP said:
You could start buying used parts and the potential for a mishap greatly increases. This guy already doesn't want to maintain the system, so there's no reason to even think of buying second hand parts here.
Click to expand...

You just have to go with reputable buyers. I bought a number of second hand parts that have been used for almost 1 year. One of which was a MCP655 and it still works great after 4 months.

And I don't see the difference between maintaining a first hand and second hand parts.:rolleyes: You still have to wash the parts upon buying, and refill the loop once in a while.
 
well, i know off the top of my head, that a dtek v2 cpu block costs around 60 dollars, i think the gpu blocks are also around there, so 180 dollars on blocks, i know newegg sells an mcp 655 or something like that for around 50 dollars, so 230 dollars so far, a micro res from swiftech is 25 dollars or so, so 255 dollars, a triple fan swiftech/hw labs rad is around 60 dollars(more rads may be needed), maybe 10 dollars on tubing, and i would suggest either ptnuke or iandh killcoils along with some distilled water, but that should be changed probably every year to 6 months and maybe 10 dollars for tubing too
 
You're looking at spending ~$350 for a watercooling setup worth buying in the first place.
 
Squalish said:
My main question is durability, reliability, & safety related. I'd like a liquid cooling solution that, after the initial leaktesting, will last me five years of continuous running without maintenance, and won't damage anything if it does spring a leak. I'd like it to be solid enough that it can be moved around, and I never, ever, ever want to have to worry about corrosion/algae/soap blocking a waterblock/pump and destroying hardware.

How would one do this?
Click to expand...

That is near impossible with an expensive kit (yearly fluid changes, wear and tear on the pumps), I just don't see how you will build a kit to last 5 years continual without maintenance. While the WC parts have become more reliable in the past couple of years, the delivery system (distilled water/glycol) still require service. Think of it in similar terms as your cooling system in a car. It still requires a flush every so many years, and nothing is long term leak proof.

I've seen a few people on other forums that have had old Athlon XP systems running the same fluid for 4+ years, and when they take it apart, all the blocks/tubes and parts are throw away items.

If you can pull it off, good luck, but your $200 budget for a CPU + 2 GPUs and redundant pumps with an external setup is going to cost more especially if you consider quality parts that you don't mix metals and stay all copper/brass.
 
I understand that chemically, the molecule H2O was gifted with an extremely high specific heat... but based on other posts indicating that the biggest temperature differential in between different parts of a coolant loop is 1C or so, I suspect one could get by without it.

How about using mineral oil or some coolant not based on water? Transmission fluid is used as a long-term coolant in cars that have grinding metal surfaces all day long.
 
Squalish said:
I understand that chemically, the molecule H2O was gifted with an extremely high specific heat... but based on other posts indicating that the biggest temperature differential in between different parts of a coolant loop is 1C or so, I suspect one could get by without it.

How about using mineral oil or some coolant not based on water? Transmission fluid is used as a long-term coolant in cars that have grinding metal surfaces all day long.
Click to expand...

Right, but if the radiator's cant cope with that 1c, it will rise in temperature until equilibrium - which can be quite high if the radiator is undervalued. Water is better than any mineral oil or anti-freeze/alcohol based coolant.

This translates to higher CPU temps, and poorer all around performance.

Even if you get things at 50% (and sometimes higher) off you're looking at:
CPU block 25$
GPU1 + Ramsinks = 35$
GPU2 + Ramsinks = 35$
Pump = 35$
Radiator = 30$
Tubing + Fans = 10$

170$... But these parts would be *very* hard to find at these prices...
 
Squalish said:
How about using mineral oil or some coolant not based on water? Transmission fluid is used as a long-term coolant in cars that have grinding metal surfaces all day long.
Click to expand...

To start, water is the best justifiable coolant for a pc, like we're using.

I, with the educated mind that I have, would like to hear anything (proof/independant sources) you have to suggest that 'transmission fluid' is used as any type of coolant. This should not be confused with a transmission oil cooler that is designed to cool the fluid, and keep the fluid from breaking down faster.

As well, in normal operation and theory, there will never be any metal grinding in any automotive engine or transmission. This occurs when parts fail or parts are worn beyond normal service limits.
 
Sorry if this post is a jumbled mess, as I'm doing a little bit of a thinking exercise here. Water has an extremely high capacity to absorb heat without undergoing a phase change to gas, thanks to all that Van der Waals stuff, I think (it's been a while).

But is water's high specific heat the "bottleneck" in a water cooling rig?

AKA: Are there other factors controlling how much heat gets removed from the waterblock rather than the fluid's ability to absorb heat? Such as...

1) Efficiency of the block-to-fluid heat transfer
2) Efficiency of fluid-to-radiator heat transfer
3) Efficiency of radiator-to-air heat transfer
4) Volume of fluid running through the system
5) Velocity of fluid running through the block and rad.

So let's say you have a system at equilibrium, with water running over a 40C block that heats up from 30 to 35C after it leaves the block. It absorbs X calories in doing so.

That same water runs into the rad, and cools back to 30C, dissipating X calories.

The question is, does the equilibrium change if water is replaced by another fluid with a lower specific heat? (Provided other factors, such as block-to-fluid heat exchange, stay the same.) I'm inclined to say "not necessarily," because a fluid with a lower specific heat absorbing X calories will heat up more than the water over the block (eg from 25C to 40C), but will also cool down more than the water over the rad (eg 40C to 25C).

I'm no thermodynamics engineer, but I do have a solid background in the sciences. If anyone feels the need to school me on this, be my guest...
 
As I understand it, the primary limitation in a typical coolant loop is not the pump: the water is nearly all one uniform temperature, because water passes through the waterblock fast enough not to incur much of a temperature hit. So you can think of it as mostly a steady-state system, with only trivial differences in between the coolant temperature at different parts of the system.

Equilibrium example: The coolant is always 29-31C, the ambient air is always 25C, the exhausted air is always 30C, and the processor under the waterblock is always 35C. At both of these thermal interfaces, waterblock and radiator, the heat conduction should suffer because of the 2/3 lower heat-holding capacity(j/ml°) of mineral oil, and that 5 degree delta should increase in both cases. I do not know by how much, or whether it can be improved by increased pumping.

What I suspect, however, is that even in a worst-case-scenario where both of them increase by 3x, the system will still be viable.

Dealing with the increased viscosity is another matter.
 
Squalish said:
the water is nearly all one uniform temperature, because water passes through the waterblock fast enough not to incur much of a temperature hit. So you can think of it as mostly a steady-state system, with only trivial differences in between the coolant temperature at different parts of the system.
Click to expand...

That begs the question: Is the speed of the water the control on how much its temperature increases, or is it the block-to-water heat transfer efficiency? What if another liquid is able to absorb heat faster from the copper water block?

I wish we had a watercooling engineer here..
 
ordovician said:
That begs the question: Is the speed of the water the control on how much its temperature increases, or is it the block-to-water heat transfer efficiency? What if another liquid is able to absorb heat faster from the copper water block?

I wish we had a watercooling engineer here..
Click to expand...

These questions and more explained years ago at http://www.procooling.com
 
Using something else as a coolant in a loop won't help you much. In fact, if anything, it will increase the maintenance that needs to be done, or at least increase the amount you'd have to spend initially. Pumps built for water cooling are not designed to be used with a more viscous fluid such as mineral oil. If you want to use something like mineral oil, you would need to move to a submersion solution (which is viable, but you're looking at a hugely more expensive project).

I think you're asking a LOT from water cooling. Ignoring the fact that your budget is pretty low for what you're trying to do, the fact that you want a setup that will cause no damage should it spring a leak is, well, pretty much impossible. Water + Computer parts = Broken. You MAY get lucky if you catch it quick enough and nothing is on at the time, but if you spring a leak you'll almost certainly be buying new parts. Also, you want something that you can move. While you CAN move water cooled rigs without draining the loop, most people would recommend against it because, every time you move it, you risk a seal popping open, a tube popping off its barb, etc.

At your budget, I'd invest in some really good heatsinks. You can get some really good ones that don't need fans. Also, replace all your case fans with 120mm. They'll be quieter, and you can always put them on a fan controller to turn down their speed when you're not doing something hardware intensive (meaning even less noise!)
 
You must log in or register to reply here.
Back
Top