• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Hot chip in need of powerful cooling

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

ClownPrince

New Member
Joined
Dec 6, 2017
So we've all seen the fancy pants "underwater" computers and most of us have thought about a build at some point in time, for me that time is now.

3M, as well as a number of youtubers, have submitted footage of computers being partially or completely submerged in this seemingly miracle liquid. Several more science based channels show the basis of cooling the vapors toward the top of the tank and even suggest using a bellows or balloon top to prevent loss from the gasses created by the rising of heat being pushed past the cooling zone, but none have quite addressed my concerns.

C4F9OCH3 by nature would create water as well as various gasses when boiled and the subsequent vapors cooled unless I'm mistaken, and I may be though with 2 hydrogen atoms and an oxygen atom present in the formula one can easily see where the concern comes from. So how would one deal with this unwanted downfall without just letting it fry your board? How often would one need to change/add coolant to an otherwise sealed system because of water exposure as well as other factors? How would one even go about creating an enclosure to house this thing while still being able to plug in ones mouse, keyboard, hdmi, etc? But most importantly how does one make this system anything more viable than a expensive data timebomb due to the bubbling inevitably leading to capacitors and the like coming detached from the board?

The reason I ask these questions is because my next build centers around the Intel Core i7-7820X Skylake-X 8-Core 3.6 GHz LGA 2066 140W BX80673I77820X and the X299 Aorus Gaming 9 motherboard. The processor it's self is known to run rather hot; lets face it with an rgb integrated board it would be a shame if the rest of the system didn't have some flair to it, and what's more flair than than committing the ultimate pc taboo of drenching it in liquid?

Obviously the 7100 eats plasticizers for breakfast which would cause a bit of problems holding the led strips to the board but simply waiting for the lights to start floating and re-securing them to the board using a good clear epoxy wouldn't be much of a problem for anyone with a steady hand.

Back to the matter at hand tho Has anyone been able to work out the problems of making this a viable cooling option? Does it remove enough heat to actually be considered a match for such a hot chip? Most regrettably I also ask that if submersion cooling still isn't a viable long term solution to dealing with the heat (and it's only going to get hotter as I plan on upgrading various parts including the chip when I save up the cash for it) what should I look at as an alternative?
 
Rigs submerged in mineral oil have been around for quite a while. Fans turn slower in it, but you can still move the liquid over a cooler like thick air (think Miami in July). Being non conductive you can dunk the PSU and all. I wouldn't put a HDD in it, due to the vent hole and lack of sealing options, but SSDs shouldn't pose any more problem than RAM. Turn the mobo 90 degrees and put the I/O panel up top for mouse, etc..
 
- Boiling is a physical, not a chemical reaction.
- Novec is 500$ a gallon
- Creating a closed loop system while still having input functionality will NOT be cheap, and will be a custom job
- the reason you're seeing these "science based channels" showing cooling vapors; is because phase change uses the most energy ergo provides more cooling. Subnucleate boiling is ideal in heat exchangers due to the FAR greater efficiency than standard conduction into a liquid
Back to the matter at hand tho Has anyone been able to work out the problems of making this a viable cooling option? Does it remove enough heat to actually be considered a match for such a hot chip? Most regrettably I also ask that if submersion cooling still isn't a viable long term solution to dealing with the heat (and it's only going to get hotter as I plan on upgrading various parts including the chip when I save up the cash for it) what should I look at as an alternative?

The reason you're not seeing it done is because its ludicrously/prohibatively expensive, and for a home computer system is absolutely not worth it. To do this as a "Science experiment" would have some merit; but for "some flair" would be preposterous. That said, its your money, but the infrastructure alone is going to cost you a few thousand + the amount of coolant; probably another couple thousand.
 
It can be done "relatively" cheap. But if you want the bling of a custom case, well that'll cost you a bit more.

Here's a "less-expensive" set-up that has function over form.
 
OP was referring to a special cooling liquid, which "boils" to cool down things. Mineral oil is not really comparable to that. Thinking more, is mineral oil any better than a good custom loop?

On hard disks, they don't have to be cooled in the liquid anyway. If you wanted to for whatever reason, look at the helium filled ones as they have to run sealed.
 
OP didn't really mention what liquid he would like to use, only that it be submersed and preferably a sealed system. The oil cooled system would fit that bill the closest as very few vapors are created and requires the least amount of maintenance in terms of replacing fluids. You can definitely create some major bling with these but I wanted to show you could do it at a low cost as well.

EDIT: After re-reading i now realize that "C4F9OCH3" is the chemical make-up for that 3M Novec. I missed that. END EDIT

IMO a custom water loop is your most efficient cooling.
 
OP was referring to a special cooling liquid, which "boils" to cool down things. Mineral oil is not really comparable to that. Thinking more, is mineral oil any better than a good custom loop?

On hard disks, they don't have to be cooled in the liquid anyway. If you wanted to for whatever reason, look at the helium filled ones as they have to run sealed.

Mineral oil is inferior to a proper water cooling loop. Its just 'unique' and it works out "ok", so people like it.
 
Delidding and watercooling. The submersion idea is neat, but, not practical.

There are several threads here talking about it, few have gone through with it.
 
The initial cost of startup for a project like this is no doubt staggering but I do happen to have an old fish tank sitting around as well as some metal scraps I can fashion a lid out of so there goes some of that start up cost. Let me just reiterate that the cost of construction is not my concern here, but rather how to construct in a way that minimizes the amount of additional fluids later on due to loss and how to prevent the introduction of water into the cooling chamber. If these two aspects can align to create something I can actually afford to maintain money wise (as I'm no stranger to the physical work) then I can play around with it and come up with more efficient ways in the future. While $500 seems like a steep price for a gallon of hfe 7100 if said gallon lasts (ie will continue to cover my hot spots and transfer heat away from them) for say a month then maintaining the proper levels of liquid would be within budget on a $200/week check while still allotting about $200 a month in pocket cash after deducting my living expenses just for an example.

Tho I realize that adding fluid once a month may be a fools dream at this point due to the high evaporation rates etc, etc.

I've also recently stumbled onto the calyos NSG S0 which looks promising. Tho it shelfs at like $823 and wont be available until march of 2018 at earliest. Not to mention it only supports a cpu and a single gpu to be cooled.

Traditional watercooling just seems like it would be a hassle with having to worry about things growing in your loop as well as worrying about spillage/leaks. Any advice to combat against this?
 
Last edited:
Wait, so, to be clear, you believe that water cooling is a bigger hassle than a submersion project? Interesting... though I certainly would not agree.

Also, the real problem with these CPUs running hot is the TIM used between the IHS and die. The product doesn't cool any better so chances are results will be worse than water, regardless.

I mean, its neat, people have wanted to and walked, and a few brave souls have done it. But my take is make sure you are doing it for the right reasons. To cool this CPU better is not one of them. :)
 
"An old fish tank"? That ain't where the dollars are. Pay shipping and I'll send you a few. They may smell like snake, but...
ED has it nailed. Moving the heat away from the source(s) is the way to go if cooling is your actual concern. Fill a custom loop with Novec. Now that will produce some interesting results.
 
Wait, so, to be clear, you believe that water cooling is a bigger hassle than a submersion project?
Yes, simply because with water cooling you have multiple parts with small inner diameters to clean on a regular basis to prevent erosion or biological growth in a water loop. Certainly the project of making the submersion chamber would be a larger undertaking however the back end, adding more fluid, would be as simple as open hatch and dump coolant.
Fill a custom loop with Novec. Now that will produce some interesting results.
I hadn't thought of that, if anyone has tried that and know the ins and outs I would love to hear more about it and the challenges associated with it.
 
I don't know of any efforts in that direction. That's why I'd find it interesting. :D And the quantities would be in much more financially manageable chunks.
 
The initial cost of startup for a project like this is no doubt staggering but I do happen to have an old fish tank sitting around as well as some metal scraps I can fashion a lid out of so there goes some of that start up cost. Let me just reiterate that the cost of construction is not my concern here, but rather how to construct in a way that minimizes the amount of additional fluids later on due to loss and how to prevent the introduction of water into the cooling chamber. If these two aspects can align to create something I can actually afford to maintain money wise (as I'm no stranger to the physical work) then I can play around with it and come up with more efficient ways in the future. While $500 seems like a steep price for a gallon of hfe 7100 if said gallon lasts (ie will continue to cover my hot spots and transfer heat away from them) for say a month then maintaining the proper levels of liquid would be within budget on a $200/week check while still allotting about $200 a month in pocket cash after deducting my living expenses just for an example.

Tho I realize that adding fluid once a month may be a fools dream at this point due to the high evaporation rates etc, etc.

I've also recently stumbled onto the calyos NSG S0 which looks promising. Tho it shelfs at like $823 and wont be available until march of 2018 at earliest. Not to mention it only supports a cpu and a single gpu to be cooled.

Traditional watercooling just seems like it would be a hassle with having to worry about things growing in your loop as well as worrying about spillage/leaks. Any advice to combat against this?

To submerge and get proper coolant levels, you will likely need 4-5 gallons of the stuff. As well, it will not "last forever" either.


Spending thousands of dollars to cool a cpu seems impractical at best. If you're not looking for "hassle" then just get an AIO, or a premade kit. I really cant think of any practical phase change application that wont cost a fortune.

- - - Updated - - -

Fill a custom loop with Novec. Now that will produce some interesting results.

Couple things ->
pumping a coolant with a low vaporization temperature is asking for trouble. As well, if you make the system air tight (feasible), the expansion volume of the gases will create a drastic pressure increase (unless you account for some expansion chamber or sufficient reservoir gapping to allow for the displaced liquid). As well, you're probably going to dead head the pump around the cpu or gpu block, due to the boiling increasing pressure around those areas.


These ideas are neat and fun to think about, but are really impractical and would need a ridiculously engineering piece of kit to support it.

edit: I feel like Im being a downer here, but its really a question of benefits vs cost. Yes, phase change cooling is cool (no pun intended), but at the end of the day, you can get really great temperatures from existing and far cheaper solutions without the headache. IS this type of system feasible? Its been done before. I can't even imagine the costs associated with it, but its probably approaching 10 grand, if not more for what you're trying to do.
 
then you need a powerfull pump;)

I'd be more worried about cavitation tbh.

At any rate, did some short math on component sizes, and the very bare minimum amount of liquid you would need is 4.7550969424 gallons based on standard ATX component dimensions. Thats 2500$. Using 8mm plexiglass, the custom sheets would cost you 213.08$
You would need a simple copper coil cooler, so like 30 bucks. Then you need to get a silicone gasket for the lid, and get it bolted down ACCURATELY; which your best bet would involve someone with a CNC machine. So lets say another 300$ for that. For the inputs, you can drill out holes and make them permanent with silicone sealant... The way I'd personally do it is get extensions for the usb and I/O devices and that way you can still change things over time. So an extra cable for each usb; lets say 7-10$ each, so pragmatically we're talking 80 bucks or so give or take.

So I guess the basic cost for a rag tag setup is going to be over 3 grand most likely. Food for thought.
 
Sure this can be done...... Here are 3M's best practices.

3M™ Two-Phase Immersion Cooling –
High Level Best Practices for System Fabrication*


This is not nearly as simple as running mineral oil submersion (and that's not simple). In the long run, it will be more work than a traditional custom W/C loop and vastly more expensive.

As far as safety goes, the stuff is relatively non-toxic LINK to msds

I looked into doing this myself a few years back. I came to the conclusion that it was not worth the expense for the cool factor and the only reason to do it is for fun and the cool factor.

If you do it..... make sure you do a detailed build log with lots of pics, I want to see it.
 
If money is not an issue here just go get one of the sealed case ln2 systems. This will keep the chip at whatever temperature you want while also keeping the ambient temperatures around the other components at will below ambient room temperatures.
 
Back