People will spend over $1000 on custom water cooling using hard tubing and a fuckton of fittings

>People will spend over $1000 on custom water cooling using hard tubing and a fuckton of fittings.
>For the same price you could just fucking get a phase change cooler

Where do these aspies get their money?

Other urls found in this thread:

tetech.com/product/lc-061/
overclock.net/t/1606680/peltier-tec-hybrid-water-cooling-vs-2-0
ekwb.com/shop/ek-coolstream-pe-360-triple
twitter.com/NSFWRedditImage

...

>condensation
>extra power draw and heat
>refrigerator noise
>only cools the CPU

If you spend $1k on water cooling, you're getting a full loop with CPU and GPUs in it. It doesn't use a lot of extra power, generates little extra heat (just

But now you gotta get another one for your GPU

>300W only
looks like you can't use Skylake-X either. lel.

I've always wanted to play with one of these but their prices never dropped and I don't imagine there has been much for efficiency gain in the last 10 years either. What I'm more interested in anymore is inline peltier coolers which are inherently more reliable and rely only on fan cooling. I'm not an engineer so I'm not familiar with the maths involved in calculating thermal dissipation nor load but for $250 it seems to me like you could effectively remove a CPU from the loop with the amount of cooling an inline peltier is capable of handling.

Anybody capable of deciphering and explaining the capabilities one of these might provide in the home desktop application?

tetech.com/product/lc-061/

There are two TEC air coolers on the market right now and I believe both are inferior to AIO/nh d15. Are you sure you understand the technology right?
I know I don't I'm just saying I looked into it recently.

I think he means like running tubing to a exterior setup with a peltier there, and using that to cool the loop. A peltier with 360mm of rad space for the computer would keep most build nice and cool, and requires only a basic air cooler (read hyper 212) in the chill box if I gather correctly

That's cool. How hard would it be to implement? That sounds like it would be a fun thing to do if no one has done it before.

Designing and building your loop is half the fun of it. Just buying a box and plugging it in is dull.

Honestly what I would do is get 2-4 peltier and run hem in a external rad box, have those all cooled with cheap water blocks on the hot side, cool side chills the loop for computer. Don't need much radiator space fit the peltiers, 2 360 would be plenty since the hot side can get hotter than electronics. Then you can have an internal loop as well, like 2 360s in the case, and route 2 quick disconnect cables out the back, just plug into the loop and turn on the exterior peltier when you want and boom instant cool Temps, if not youre not screwed either. Easy to setup and you could even use a second more powerful pump to move the heat and noise to a completly different area
Wouldn't be that hard, also solves the problem of fuck huge cases being needed to fit extra rads tubing and power supplies, as well as the noise and heat factor from the peltier

Are you idiot 2011 and 2066 are same mounting.

No I meant Skylake-X outputs more than 300W when OC'ed.

dude mineral oil lmao

It will be more ineficient to cool same ammount as the phase changer. Peltier is very ineficient.
Surely you dont need -30c, but to reach that much on a 150w average on cpu you would need +1kw of eletricity on peltiers.

Why you need a cpu running -30c?

Wouldn't it be better to just use all 4 360s in the loop at that point?
I mean at a certain point with enough rad space I'd think temps would barely go above ambient. Like a 1080 and a 7700k on 4x 360mm rads? That would be crazy

You start to get crazy overclocks at temps like that. Even on super cucked pascal cards gpu boost will think you're an LN2 fag and stop fucking you over.

The problem with pelts is that they're inefficient. That makes things difficult even if you don't give half a rat's ass about the electric bill, because it means that in the process of moving heat, they add a lot more of it. The cold-side temperature is dependent on how low you can keep the hot-side temperature, and the hot-side temperature has to be kept under a certain level or the thing will cook itself - pelts die at lower temps than CPUs. Too much heat from the CPU and it can overpower the pelt and you get runaway temperatures.

Back in the socket A days a big Swiftech air cooler with a 226-watt pelt on it was just about able to keep a 130W+ Athlon to something between room temp and ordinary overclock temps. Say 40C or so. The pelt was dumping its own 226 watts of heat into the sink, plus the heat of the CPU that it used that energy to move. It'd be difficult to cool modern housefires with pelts even if you had a big, beefy, high-flow waterblock on the pelt, just because you'd need to be getting rid of something like 600+ watts of heat - and doing it without much of a rise over ambient on the hot side.

>TEC air coolers
They can literally freeze CPUs, I imagine those coolers that you're referring to are designed to prevent condensation. I'm interested in integrating one into a loop rather than directly cooling the CPU with a peltier.

That's correct. The temperature differential they list is above my understanding, but if I do get it I believe it will cool 52w before it starts to return to ambient, which is perfect since at load there wouldn't be any worries about sub-ambient temperatures leading to condensation in their respective blocks.

Someone effectively made the same thing, albeit significantly less elegant:

overclock.net/t/1606680/peltier-tec-hybrid-water-cooling-vs-2-0

I've yet to find something relating to the application of a cooler like the one listed though. It makes sense that it wouldn't be a popular solution, too since comparatively it's quite expensive to a standalone peltier, and most of the people interested would be D.I.Y. nuts and hackers.

For cold bugs.

holy wormtit that link is so sick. I wish I had the resources to do crazy shit like that all day. You should do it and shit down everyone's throat in the /guts/ thread with their cute little 240mm AIOs and noctuas.

I've done a little research on on the subject, and I determined that TECs have an inferior coefficient of performance (CoP) compared to compressor-driven phase change. In other words, they use more power (and therefore generate more heat) to move the same amount of heat as a compressor. In that respect, TECs are inferior.

I'd have to get into the numbers to tell exactly what a specific TEC will do, but that one you linked won't cut it for pumping the heat off a modern processor. Also pay attention to the noise level - that one runs at 44dB.

The old rule of thumb was that the pelt had to be rated for twice the power as the chip it was to cool.

I don't use housefire CPUs. There aren't any applications I run that require them, just running a 65w 6700 non-k right now, and undervolted at that. You're absolutely right about them being inefficient though which is why wouldn't work well I don't imagine.

I can't find any numbers on the failure temperatures from the site though, I'll have to look those up, I wasn't aware they were that low.

I think it was something like 70C hot-side above which you risk damaging the pelt. Which doesn't seem all that bad until you remember how much heat you have to dump.

Diminishing returns with rads, actually wouldn't be much better, maybe 2-5c
Oh I know they're not efficient, like at all. It uses more than double the energy to cool that, as in to cool a 95w chip you're looking at over 200w of heat with the peltier. It's just a matter of what you want cool and how you want to cool it
Also eliminates the noise advantage with h2o, and also takes waaaay more power. Still if done right it keeps things stupid cool,but at a huge price

>Still if done right it keeps things stupid cool
Things inside your PC, anyway. Bearing in mind all the extra heat from the pelt has to go somewhere, you might not be too happy with it in the summertime.

Exactly. That's why the only way I would do it is with a expansion loop, somewhere else in the house. Can still move pc, most heat dumped elsewhere, nice and cool even in the ungodly heat of the summer

I am afraid that water/oil with substances that kills algae and works anticorrosive will slowly evaporate and the owner is going to inhale it.

Phase change coolers are a pain in the neck.

I knew a guy running an Athlon (can't remember which, it was before Core2 so probably a 64 dual core) at around -20C on his peltier phase change cooler, overclocked shitloads as well. If the CPU reached 0C the system would crash. Turning on the system it waited until the CPU reached -14C before actually starting the boot process.

But you're still right, there's a lot of problems you have to deal with and the power consumption is ridiculous.

Assuming you don't double the wattage though, simply canceling it out, it would remain at approximately ambient, correct?

Do you think the CPU might be damaged or would it run fine at room temperature if he removed the overclock.

Read this guy's post again: , specifically this part:

>a 226-watt pelt on it was just about able to keep a 130W+ Athlon to something between room temp and ordinary overclock temps. Say 40C or so

40C is already way above ambient (and if it's not, you have bigger concerns than cooling your PC). Reducing the power to the pelt is only going to reduce its ability to move heat, and therefore the CPU will run even hotter.

>I mean at a certain point with enough rad space I'd think temps would barely go above ambient. Like a 1080 and a 7700k on 4x 360mm rads? That would be crazy
Not really, a 1080 and 7700k won't spit out enough heat to overwhelm a single 360mm radiator, much less 2.

Take EK's PE 360 as an example: ekwb.com/shop/ek-coolstream-pe-360-triple
This radiator is rated to dissipate about 500W worth of heat when given fairly modest fan speeds, the most limiting factor with just one of these would be the thermal jizz in that 7700k.
Adding more radiator capacity isn't going to cool things down any further since your 7700k and 1080 aren't outputting 500W of heat in the first place. At best you're just getting the same temperatures with lower fan speeds since you can't cool water below ambient by just passing air through the radiators.

>This radiator is rated to dissipate about 500W worth of heat
This is a somewhat misleading simplification. The capacity for heat dissipation (or more accurately, heat transfer from water->rad->air; expressed as the rad's delta T) varies depending on heat load and airflow (and flowrate).

Take pic related, a performance graph for a thick, 480mm rad. With a heat load of ~300W, the temperature rise is anywhere from 2 to 9C, depending on airflow. In other words, the water in the loop will be 2-9C warmer than the ambient air going through the rad.

Keep in mind that this is only only tells you the water temp in relation to the air temp, and doesn't say much about the temperature of the CPU and/or GPUs, as the water blocks and the chip-IHS interface each have their own delta T. Expect the CPU/GPU to be a good 10-20C higher than the water temp.

Sandwiching your peltier into the existing water cooling loop isn't sane. It should be used to sub-cool already cooled coolant just before enters the block, pic related. Peltier hot side can sizzle away under air cooling (outside the case) and doesn't matter, inlet coolant still went from ~ambient to hopefully near -8C etc. with that bit of extra pull down.

And I'm getting banned for posting web shop pics because no advertising. Anyway there are inline coolers out there search 12V 240W inline peltier for example

What's the argument for cooling the water instead of cooling the CPU directly?

Some Athlon 64s had a bug that caused problems if the on-die thermal sensor read below 0C. I think they fixed that in later models/steppings.

Well you can use em just as water chillers. Still not exactly efficient though, and it doesn't remove the need to dump all the pelts' heat somehow.

Also whatever pic it doesn't like try just opening it up in paint or whatever and cutting/pasting it to a new image, thats worked for me sometimes.

Lets you get some benefit out of a pelt that wouldn't be powerful enough to use directly. You don't have to worry about mounting a pelt (and probably a cold plate) underneath your waterblock. An unpowered or dead pelt is a good thermal insulator, using it as a water chiller lets you turn it off when you don't want it.

The peltier is inefficient. All its waste heat must be got rid of somehow - far easier to move that heat source away from the CPU to where it can get relatively hot and radiate efficiently and no effect on CPU. After all, the most useful feature of a peltier is how it can stack temperature differences with no limit, making small cryogenic systems possible for example.

I can see people shimming a peltier element directly under their AIO cooler head for kicks, because kind of fits. Dumb idea desu

I feel like it's a wholly different equation, though.

He would've been using direct contact cooling which is definitely sub-optimal. From what little I've gathered the saturated water temperature of a custom loop is approximately 8c, a far departure from taking the brunt of 90w+ TDP. To top it off if you positioned it right out of the outlet of the radiator I'd posit that with well tuned pump speed you'd be able to keep a perfect ambient with relatively low power consumption off the peltier. As a matter of fact it's within the realm of possibility to run below that without a radiator, but targeting ambient would probably be the most optimal.

Though both of you give good points to cool the fluid rather than the CPU, you'd have to insulate pretty much all your tubing and piping to prevent condensation elsewhere. The benefits of Peltier on CPU is mostly practicality.

yes but for one thousand fucking dollars the net gain over a basic closed loop cooler is not worth it in the slightest

t. college student

I'd rather spend nothing and use the stock cooler.

People doing this really should be wanting to go below ambient, the whole point is "I need extra cooling can't get from a normal setup". Theres no practicality in 15% efficiency.

>the saturated water temperature of a custom loop is approximately 8c
What is "saturated water temperature" supposed to mean? Water will keep getting hotter until it boils. In the context of a water cooling loop, the water will get hotter until the heat loss (to air) matches the heat input (i.e., it reaches equilibrium). The greater the temperature difference, the greater the heat transfer.

Where's the carb on this thing

I always considered custom water cooling to be too much of a hassle and too expensive to really ever care about it. I did however, once use an all-in-one water cooler made by Thermaltake for my i5 3570k build. The pump went out and I basically had to throw it away. I switched to Noctua since they have solid hs/f combos and having a pump is just one more moving part that makes your build likely to break. Like a mechanical hard drive. I honestly think fans are the only moving part I want in my system, even optical drives are better when they're external so I can move them between machines but maybe that's just me.

Not for a college student, but when you're out in the real world and earning enough disposable cash that you want things to be a little more aesthetic, then moving up to custom loop cooling is nice.
Its also better than using two closed loop coolers; which I was with my Fury X and Kraken X61 prior to going open loop.