|
I have water cooled more of my computers than not, the first time I did it was in the 90s before they even had proper water cooling parts. I sawed the square part off a 4" PVC pipe cap and used tub and tile calk it to glue it the top of an AMD K6 and drilled holes in it with tubing similarly calked in so water was just flowing directly on to the aluminum heat spreader. It had no radiator, the entire system worked because the reservoir was a 25 gallon plastic storage carton that just had such enormous thermal capacity that it was impossible for a 16w CPU to heat it by any significant amount. Tap water, well water, distilled water, hard water, soft water, ice water, dish water, carbonated beverages, pure alcohol...if its wet, I've probably run it through a cooler loop at some point or another. Through the years I went through tons of water blocks, from cheap custom milled copper held in the super 7 socket by janky clips to a bunch of innovatek parts, including several blocks that had copper bases with aluminum cap pieces and had about the expected endurance you would get from mixing those metals, the aluminum slowly dissolved into the water over a year. How the industry has learned since those early days. Though one thing I can give a nod to, the eheim 1046 pump I got back then deserves much of its legendary status, mine is something to the tune of 15 years old with a good 10 years of 24/7 operation behind it and it still just works if you ever plug it in in something wet. I eventually dumped the bucket for much smaller internal reservoirs and threw in a 120mm heater core for a radiator. I abused the ability to dump the heat somewhere else and stuck the radiator under the floor in the basement with a ridiculous 190 CFM 4000 RPM 120mm Delta fan on it. Which I eventually replaced with a 360mm radiator with three 110 cfm Deltas on it, could feel the air moving from that radiator from 12 feet away, cooling a core 2 duo by then. Ultimately after moving to a new house and time, maintenance and financing being a pain, I abandoned my custom loop and ended up with a couple AIOs in the form of of a Corsair H100i and later a MSI Sea Hawk 1080. However, last year I started to get the itch to return to a proper custom loop cooler. Between wanting something fun to do to escape the daily grind, thinking it was about time to upgrade from my aging 3770k that was having trouble keeping the 1080 fed, and once again having more money than sense. I decided to build a whole new system and to stuff a small fortune worth of EKWB parts into it with the goal of it being a extremely stealthy system both audibly and visibly (Just say no to RGB!) while still making no compromises on performance. The result after voiding several warranties by cramming a 420 radiator into the top of a Corsair Obsidian 750D and removing the stock H55 from the Sea Hawk:
|
# ¿ Apr 2, 2017 16:37 |
|
|
# ¿ May 3, 2024 13:42 |
|
You can tell this is on the opposite side of a memory module on the board, so depending on what function it does that memory module could be either lacking proper power delivery or be outright disconnected from the GPU. If it is part of the power delivery, it wouldn't be the only pin for that so the card could still possibly work although the memory may become unstable at higher frequencies, but if it is a trace to the GPU itself the card may be unable to initialize the module at all and the system would fail POST. I doubt powering it on to find out one way or another would cause any further damage the card so no harm in testing, but even if it does still work I would get it fixed just to be safe for the long term. Also be aware that parts like these often have a proper polarity, so when it is soldered back on it has to be oriented correctly or it won't work. Now with the 1080 Ti, one memory module is missing (this board is the same board as the Titan Xp which would have all 12 modules populated), I don't know how nvidia bins these chips and if the memory is always missing from M12 where as you chipped that part of off M8. But if there is no memory chip in that location on the other side of the board, then damaging that component would likely be harmless. Also even if the part is cracked or broken, you can probably safely harvest one off of the empty memory channel to replace the damaged one. Indiana_Krom fucked around with this message at 13:07 on Jun 10, 2017 |
# ¿ Jun 10, 2017 12:54 |
|
Unnecessary, you can turn off a water cooled PC the same as an air cooled one, there shouldn't be any significant sediment in a good loop. The worst you should get is the occasional air bubble noise on startup if there has been a drop in atmospheric pressure recently while it was off (dissolved air forming bubbles), however the reservoir should capture it all within a few seconds and it will be back to normal.
|
# ¿ Jun 10, 2017 23:14 |
|
I'm also somewhat curious how you managed to do that much damage on card that ships with a back plate, but then I looked at a disassembly video and saw the plate and the cooler are held on by separate sets of screws. https://www.youtube.com/watch?v=3AQ5R23FIHU Also worth noting that M11 is empty in the card in that video, so looks like nvidia is actually binning defects here instead of just always disabling the same channel. Edit. Ah, well that is a $700 mistake you probably won't soon repeat. FYI a trip to the local hardware store will usually turn up a decent set of screw driver bits in both imperial and metric for less than $20, so probably a good idea to have them on hand (especially when dealing with PC water coolers which often results in a significant mix of both standards in the case). Indiana_Krom fucked around with this message at 00:49 on Jun 11, 2017 |
# ¿ Jun 11, 2017 00:44 |
|
You don't even have to measure soft tubing, just start routing the whole coil through the case from one fitting to the next then pinch and cut it as you go. Soft tubing is so easy to work with it is basically cheating, with the only real downside being it won't look as good as a well done hard tubing rig (which is the whole point of hard tubing).
|
# ¿ Jun 22, 2017 01:57 |
|
For #1, If you are blowing that much money on it already, how much more would a aquacomputer aquaero controller (or something like that) and a few temperature/flow sensors for the kit hurt? That way instead of depending on the CPU or GPU temperature sensors on one machine or the other, you could control the pump and fan RPMs based on the actual coolant temperature deltas to ambient and always get the optimal cooling solution regardless of which machine is busy at the time. Use some PWM hubs/splitters to drive the radiator fans from one channel, and your pumps (plenty of PWM compatible D5 pumps out there) on another. You use PWM hubs because they can pull the power directly from a high amperage 12v source so you aren't limited to the 30w maximum of the controller, while still spliting the PWM signal from a that controller to everything connected to it. One system would still be the "master" in that the controller requires USB2 input, although it may be possible to program it and then simply power it off a USB wall charger afterwards. Same for the fans and pumps which could be powered from a 12v DC power brick independent of either computer from there. Obviously everything would have to be PWM because only it allows you to split the power source from the speed control, but that way the coolant loop could be run entirely independent of both machines without any guesswork. Plus if you stuck that much radiator area in the system, you could probably dump 5000w into the loop and the only thing that would heat up significantly is the room its in.
|
# ¿ Jul 22, 2017 17:15 |
|
Deuce posted:I was already thinking about a fan controller of some sort. I hadn't considered an independent 12v power brick to power everything. I can't quite picture how that setup would work. I assume that's just a wall plug on one end with a molex/sata on the other end? And then the molex/sata plug powers the PWM hub, which itself is plugged into the fan controller? Maybe with a molex/sata splitter so that two PWM hubs can plug into a single DC power brick. Mine is from a long since recycled viewsonic flat panel monitor but it still reliably supplies up to 45w if I need it. Very useful hardware to have for anyone who wants to do loop maintenance/filling/draining without using one of those hotwire plugs for ATX PSUs.
|
# ¿ Jul 22, 2017 20:03 |
|
I live in the central US and I ordered parts for a complete water cooler loop direct from EKWB, it got here in less than a week for a $40 shipping charge via UPS. The only problem you might have is a lot of credit cards will deny the charge if you try it direct, so just use paypal (it costs you nothing extra) instead and it will go through.
|
# ¿ Aug 7, 2017 02:02 |
|
Mr Shiny Pants posted:So I got my new ThreadRipper system running with a Kraken X62. One question: Do I need to use that godawful CAM software to have the fans change the RPM or does the thing change its RPM automatically depending on load? Indiana_Krom fucked around with this message at 00:10 on Aug 26, 2017 |
# ¿ Aug 26, 2017 00:07 |
|
PerrineClostermann posted:My H50 never made any noticeable pump noise Neither did my H100i, pump was always silent, fans not so much. Same for the pump in my MSI GTX 1080 Sea Hawk. I've never heard the pump sound from an AIO and I've seen units from both of the major OEMs. On the other hand the D5 in my current loop is clearly audible if I crank it to 100%, but that makes sense because the flow rate at that throttle is ridiculous, I run it at 35% (~1700 RPM) and it is inaudible there.
|
# ¿ Sep 5, 2017 23:07 |
|
Scholtz posted:
|
# ¿ Sep 24, 2017 12:18 |
|
I decided my 7700k was running too hot for all the money I spent on this 420mm custom loop cooler, so I grabbed some thermal grizzly conductonaut and a rockit 88 delid tool. Yielded exactly 20C reduction in load temperatures under the same conditions. The factory thermal compound Intel uses really is complete garbage.
|
# ¿ Oct 15, 2017 01:37 |
|
It was hitting 88C in barely above stock configuration, only running all core turbo of 4.5 GHz with a heavy AVX workload (prime95 v2.85), the power draw was about 120w at stock voltage. Mainly what I did it for wasn't overclocking but for silencing the fans which start ramping up as the CPU approaches 70C, now under more typical workloads in the 50w range it rarely rises much past 50C and so the fans never have to ramp up. Amusingly enough the same prime95 test now tops off at about 115w/68C, so lowering the temperature by 20C cut 5w worth of leakage. Although as a result of this change the GPU runs slightly warmer because the fans aren't spinning up as aggressively which allows the coolant temps to rise a little more than before.
|
# ¿ Oct 19, 2017 17:05 |
|
GRINDCORE MEGGIDO posted:Watercooler noob checking in.
|
# ¿ Dec 3, 2017 17:46 |
|
Go big or go home. Note, the radiator, pump and reservoir do not *have* to be inside the case. Especially for a server system with 8 hard drives that is presumably kept somewhere out of sight so it doesn't matter if it is a little ugly.
|
# ¿ Jan 29, 2018 23:49 |
|
A good AIO isn't really inferior to a good tower cooler, but it isn't really superior either. The big advantage at that class is flexibility, AIOs have more mounting options.
|
# ¿ Mar 28, 2018 00:28 |
|
The typical GPU die that you would use a water cooler on is also a significantly larger surface area with a lower power/heat density than a typical CPU die.
|
# ¿ Apr 23, 2018 23:07 |
|
If your system was throttling on an air cooler, a larger and more powerful water cooler will actually cause it to dissipate MORE heat/energy not less, because system would no longer throttle. The efficiency gains of running at a lower temperature are trivial, for reference I de-lidded my CPU and replaced the stock transfer compound with liquid metal that yielded a ~20C drop at peak load which only reduced the CPU power consumption from 120w down to 115w. Also the amount of heat a PC dissipates into a room is measurable but not terribly significant. The formula to use is 1 Watt equals 3.4121 BTUs/Hour, so for instance my entire PC, both monitors and an audio amplifier uses roughly 400 Watts in a gaming load, which if it was all going to heat losses would be ~1360 BTUs/Hour. It is there, but it isn't going to break the air conditioning or significantly warm an average sized room.
|
# ¿ May 27, 2018 20:51 |
|
The motherboard in any system worth attaching a custom loop cooler with two radiators will have power delivery so overbuilt your CPU would explode with the force of 1000 suns long before you reach the thermal or mechanical limits of the power phases. And before that, the over current protection would force the system to back off or halt entirely. Orient the radiators and fans any way you would like and ignore the VRMs.
|
# ¿ Jun 30, 2018 23:20 |
|
Ah, even then, as long as it isn't like some $89 special motherboard I wouldn't worry (and if it was, the VRMs overheating would probably be a feature not a glitch). The math at least according to gigabyte is about 60 watts per phase (they built a board with 32 phases that could supply over 2000w through its LGA1155 socket).
|
# ¿ Jul 1, 2018 00:15 |
|
Enos Cabell posted:Posted this in the overclocking thread before I saw this one, might be more appropriate here. Quiet while extracting "max OCs" and cooling a big GPU is firmly custom loop territory, because you could easily be approaching 500w heat output in that scenario.
|
# ¿ Nov 6, 2018 20:28 |
|
While doing a system upgrade I decided to finally get rid of the optical bay in my Corsair Obsidian 750D: Before: After: It isn't a custom water loop if you haven't taken a Dremel (or other high speed cutting tool) to your case to make it fit better. Makes that 420mm EKWB behemoth look slightly less out of place since a third of it isn't inside an unused drive bay anymore. I might get a right angle fitting or two now to straighten up the tubing runs, although the length and flex as is allowed me to reuse the whole loop with a new motherboard without having to cut any additional tubing. Also glad I didn't waste money on putting a drain in the loop after all, since it is soft tubing I just released the pump/res mount opened the port at the top and tipped it over into a bucket which drained the whole loop with minimal effort. If it was a hard tubing loop it would be a different story for sure. As for temps, I set BIOS to let 250w through this i9-9900k, threw Prime 95 v28.5 (AVX) small fft at it which made it pull 196w, and it hit about 80C after a couple minutes. Once I get some proper fan/pump speed profiles setup (speedfan doesn't work with this motherboard! ) I might go for a longer run just to see where it equalizes, but I know from experience this loop can vent 300w and change without making a sound and not rising more than about 5C from the initial surge.
|
# ¿ Nov 24, 2018 15:48 |
|
forbidden dialectics posted:Yeah, unfortunately even with a really good water setup, the bottleneck for heat transfer is still the processor itself. The IHS, the somewhat lovely solder, the extra thick silicon on the top of the die. I delidded my last processor and gave it the liquid metal treatment, but 80C in prime95 AVX hell while dissipating 196 watts is acceptable for now .
|
# ¿ Nov 24, 2018 18:26 |
|
Indiana_Krom posted:Once I get some proper fan/pump speed profiles setup (speedfan doesn't work with this motherboard! )
|
# ¿ Nov 26, 2018 03:23 |
|
Partial Octopus posted:I'm looking to build my first custom loop. I used the EKWB configurator and it is suggesting a 360 rad up front and a 120 rad in the back. I'm running an overlocked 8700k and a rtx2080ti. Are two radiators overkill? Should I just go with a 360 up front? I have a overclocked GTX 1080 and a i9-9900k running on a single 420mm radiator and it works fine while barely making any noise. For 99% of my usage my fans and pump never rise out of their 30% idle PWM throttles. All I did was swap one of the plugs on the GPU block with an XSPC thermristor plug, then set my radiator fans and case fans to gradually start ramping up from idle when the coolant temp hits 35C, reaching 100% throttle at 45C. The pump is set to start ramping up its throttle at 40C with 100% throttle at 50C because honestly idle on it is good enough since 100% on the (D5) pump circulates the entire volume of the loop in probably less than 2 seconds which is ridiculously overkill. After ~15 minutes of gaming the coolant hits 36-37C and the fans barely rise out of their ~600 RPM idle setting, most of the time they don't even exceed 800 RPM and the coolant temperature pretty much stops there like it hit a (cold) concrete wall. The GPU basically never climbs over 50C, and outside of prime 95 AVX or other similar torture tests that pull >200w through them the CPU cores rarely pass ~70C. Right now my system has been idle for several hours, the ambient temp is 25C, the coolant temp is 27C, the GPU is a solid line at 30C and the CPU cores are all hanging between 28 and 30C. How many radiators you need depends on what exactly you want to accomplish with this build. If you want super low temperature deltas, then you need every last millimeter of radiator you can possibly cram into the loop combined with powerful high static pressure fans and a pump strong enough to circulate it every few seconds. But if you just want "good" temperatures without a lot of noise, one 360 will do it as long as you set reasonable goals. If I'm going to recommend one thing, it would be to get a coolant temperature sensor so you don't have to guess. My motherboard has two thermristor headers so I just use that, but there are also fan controllers out there that can use one and it makes a world of difference in how the fans and pump throttle. Basically your CPU might hit 70-80C but as long as your coolant is still 30C there is literally no point to ramping up the fans or pump whatsoever. An all copper 360+ mm radiator like the types most custom loops employ can easily dump 1000w and change worth of heat at a moderate noise level if you allow a ~15C temperature delta over ambient on the coolant.
|
# ¿ Dec 14, 2018 23:37 |
|
Corsair makes this thing: https://www.amazon.com/Corsair-Fan-Controller-Commander-CL-9011110-WW/dp/B0725HP1J2 Which can presumably control PWM fans based off the temperature reading from am xspc sensor like the one listed in a previous post.
|
# ¿ Dec 16, 2018 00:24 |
|
TipsyMcStagger posted:So kind of debating trying my first water-cooling build but want to know how much maintenance needs to get done once set up. A sealed system shouldn't need to be refilled but also wondering if you put dye into it if you need to clean or flush the system regularly Things that can and will happen in custom loops: algae will eventually grow in anything that allows light exposure to the coolant (meaning, clear tubing + RGB LEDs shining on it for a couple hours a day). Additives inhibit growth, but eventually lose potency and should be replaced. Silver kill coils are another method, but using one will corrode and eventually destroy any nickel coating which is common in fittings and blocks, so only use that if you have no nickel in the loop. Generally if you want low maintenance, you should avoid clear tubing/light exposure, avoid dyes, always use distilled water with a manufacturer approved additive, and avoid mixing dissimilar metals. Copper/nickel/brass are ok together, but do not mix them with aluminum or silver. If you do all that, you should be able to get away with fairly long maintenance intervals, other than topping off the coolant as needed from evaporation. You can also make maintenance and flushing a lot easier on yourself by including fill and drain ports in the build, which is probably the best approach.
|
# ¿ Feb 9, 2019 20:35 |
|
TigerXtrm posted:That either takes a shitload of rads or your rig sounds like an air tunnel even in idle. I suppose it depends heavily on the loop. This also describes my experience almost exactly, GTX1080+9900k in a single loop 420mm (3x140) radiator. Idle fans are ~600 RPM, things start slowly ramping up at 35c, 100% throttle is 45c, and it usually stops around 37-38c at load with a very moderate rise in fan RPM. Once I exit whatever game is loading it up, the temps usually drop to under 35c in only a minute or two. After a long idle the loop has been down as low as 27c in a 24c room, and it isn't terribly unusual to have a couple CPU cores matching the loop temperature. The delta is basically everything in cooling, water or otherwise. The higher the delta the easier it is to dump heat, but the hotter the thing being cooled will be. If you are aiming for silence, allow a higher delta and the cooler will dump more heat at lower RPMs. On the other hand, ramping everything to 100% doesn't really yield much improvement in deltas, because holding +3 or +4 at load is really hard to do. Trying to transport 300w of heat with only a +5 temperature delta requires ridiculous amounts of airflow and surface area and is going to be expensive and noisy.
|
# ¿ Feb 23, 2019 01:07 |
|
Yeah, it isn't particularly unusual for a water cooler to move and draw more air through a case than a traditional cooler. Likely it would be worse, but depending on how/where the radiators are mounted it could be slightly easier to dust out with compressed air than a more common tower cooler (you could literally just blow compressed air straight down through a top mounted radiator and dislodge most of the accumulated dust without even opening the case if you were willing to let some settle inside). Dust filters on any intakes and keeping the case off the floor are definitely going to help more than any sort of different cooler type or layout.
|
# ¿ Mar 3, 2019 16:01 |
|
Important things to keep in mind about water cooling: How you use it doesn't count, Size Is Everything. Water cooling doesn't make your PC quiet because its better than a heat pipe tower cooler, it makes your PC quiet because the radiator can be three times the size/volume/surface area of even the biggest lump of metal heat pipe tower out there and you can throw a half dozen slow and super quiet 140mm fans on it. This is also why the quietest water cooling setups are usually in gigantic full tower cases. You can water cool in a small case, just curb your expectations about noise and be ready for a long painful exhausting build (the smaller it is, the louder it must be in order to equal the same performance, and the more cuts and scrapes and going back shopping for something 2mm smaller that will actually fit this time). Get a D5 pump, EKWB makes some with PWM control, at full speed they are louder than a hard drive, but at 35% you cannot hear it at all unless you literally touch your ear to one. Coolant volume doesn't really impact how big of a pump you need or how fast it needs to run, what does is how restrictive the components in the loop are and how high the pump has to push the water up (the pump should be at the lowest point of the loop, directly under the reservoir). EKWB also makes 120mm and 140mm brackets that let you mount the pump to the other side of a fan and it looks clean. Drilling holes and bolting or zip tying it down is also a fairly common method. The best fan/pump control for a water cooler is by using thermristors, there are various types between in-line and plug types that screw in to the loop one way or another and can then be connected to a motherboard thermristor header or aquacomputer that tells the coolant temperature, which you then use to determine the PWM throttles of the fans/pump. Note doing this requires a motherboard with a thermristor header, which isn't guaranteed, otherwise you would have to use an aquacomputer as your fan/pump controller. Upgrading the PC after building a loop varies in difficulty, also depending on what type of tubing you use. Soft/flexible tubing and you only need to drain the loop when you change a component that requires a different water block. Hard tubing requires draining the loop every other time you look at it. Drains are absolutely necessary for hard tubing loops, but are optional on soft tubing (you can just tip the reservoir upside down into a bucket on a soft tubing loop). So if you are using soft tubing, and for instance you want to upgrade your motherboard/cpu, but you are sticking with a similar or compatible socket (like Intels LGA 115x sockets), then it is totally possible to swap without draining a loop with a little care. Your video card's full cover block on the other hand is specific to one type/make of GPU and will almost certainly not be compatible with your next GPU or even a different GPU in the same family, so you will have to drain the loop for that. As for maintenance, there are a trillion things out there that like nothing more than to grow in warm water if they are given a little bit of light. So if you want low maintenance, use dark colored opaque tubing, turn off those LEDs and avoid light exposure to the coolant as much as possible. Also use coolant mixes that are approved by the manufacturer of your loop components in the proper ratio of distilled water, and avoid dies/colors like the plague they are. Basically bling/LEDs/showcase windows/colored coolant/etc increases maintenance demands. Clear/uncolored coolant + no light = long maintenance intervals (this is also why AIOs have no method or way to extract or replace the coolant).
|
# ¿ Apr 13, 2019 22:23 |
|
Xerophyte posted:Temperature control: A note on this, 60C is an incredibly high coolant temperature, like water that warm will burn your skin pretty easily, and it is extremely unlikely you would ever have a proper loop hit that high of a temp unless the temperature in your room was also crazy high. I run a loop with a single 420mm radiator and I prioritize noise over absolute temperature performance so it runs a coolant delta of around 12C over ambient at load. (24C ambient +12 delta = 37C coolant at load of ~300w.) I can push it a little higher with torture tests that bring the load up to 400w, but a mild fan speed increase stops the rise at about 14C delta. Anyway, the thing worth keeping in mind is the delta over ambient is the really important part about temperatures. Average room temperature is around 23-25C, it is basically impossible to get your coolant any lower than about a 2C delta over ambient at almost no load. Dump 400w of heat into anything that fits in that case and it will warm up, but here's where a key aspect of cooling in general comes in to play: The hotter something is, the easier it is for it to dissipate energy. Basically the bigger the difference between the ambient air temperature and the coolant temperature, the more heat can be transferred from the water into the air. The last time I flushed my loop, I ran hot water through it and the radiator was so warm you could feel air flowing through it purely from the convection, spin up the fans and its an instant room heater, but the water coming out is still hot. Also on loop order: it is completely irrelevant. Put anything in any order that fits, the only thing that matters is you want your pump+reservoir at the lowest point gravity wise. The reason is simple: my total loop volume is about 900 ml, at full speed the pump does Edit: whoops, read the specs wrong, turns out the max flow on the pump is 1500L/hr not 500 (though the actual flow rate will be less than that because of loop restrictions). Indiana_Krom fucked around with this message at 19:16 on Apr 14, 2019 |
# ¿ Apr 14, 2019 18:53 |
|
Worth noting I have a Corsair Obsidian 750D tower case, which is one of their larger ones, and I still used a Dremel to saw a considerable amount of metal out of the case (the entire 5.25" bay) in order to properly fit my 420mm radiator up top. The problem with napkin math about stuffing multiple radiators into something that small is that radiators are bigger than fans because they stick out on the ends for the chambers and fittings, and there also needs to be room to pass the tubing to them. Then you also have to consider that even a slim radiator is ~40 mm thick once you add the fans. You might end up in a situation where even if the radiators fit, you can't install fans on half of them because the fans interfere either with other fans or with another radiator even. Usually cases that advertise space for so many different radiator sizes are only ever actually engineered to fit ONE radiator of any of those sizes at a time, and often it comes at the cost of being unable to install fans in some of the immediately adjacent slots. Basically look at your case with a fan of the appropriate size in every slot, if any two fans on adjacent slots (like the rear fan and a top fan) are LESS than 40mm apart in any direction, you cannot fit two radiators together in those locations. So for instance if you install a rear radiator, you cannot install a top radiator (but you could still install a front radiator), or if you use a top radiator, then neither the front or the rear would be usable. So basically, I've been doing water cooling since the late 90s, and I wouldn't touch a case that size with a 10 foot pole unless someone was paying me thousands of dollars to make it work and even then I'd stop at one radiator.
|
# ¿ Apr 15, 2019 23:59 |
|
Fedule posted:I thought of another question. I think I know the answer to this but I guess it can't hurt to check. Usually there is enough air in the reservoir to allow for the expansion, but also unless you run a loop with no fans at 500 watts of load the coolant isn't going to warm much past 15C over ambient. Radiators get tremendously effective at higher coolant deltas to a point where short of wrapping them in a blanket you just can't warm them up any further. Also if the loop is using flexible tubing with compression fittings, the designs for those were originally adapted from pneumatic systems which were meant to contain 100+ PSI compressed air, and flexible tubing can expand/contract enough to compensate for changes in volume.
|
# ¿ May 4, 2019 18:03 |
|
Also manifolds generally increase the number of connections and amount of tubing in a loop, plus there is a risk of the more restrictive block getting significantly reduced coolant flow compared to a serial loop.
|
# ¿ May 24, 2019 00:51 |
|
I'm going to second the mini-split, I have a high end fujitsu unit that I had installed several years ago and it is worth every penny. It is virtually silent, does heating and cooling, is more efficient than central air even and doesn't clog up a window.
|
# ¿ Jun 21, 2019 15:26 |
|
Harik posted:Fair enough. How much did it cost you overall? As for long tubing runs, the popular D5 pump used in most custom loops is insanely powerful, you don't need anything else. There is a reason I never crank mine past 35% PWM throttle beyond just the sound (its silent under ~45%), at 100% throttle it would circulate the entire volume of my custom loop in less than 3 seconds. I know the water cooling community settled on D5s because they are relatively quiet, can run directly off 12v and are simple and reliable, but they are also pump like a fire hose and are ridiculously overpowered for the job. I once used a much weaker aquarium pump with smaller more restrictive tubing to run a loop through the floor into a shelf hanging off the basement ceiling where I kept a 360mm radiator with 3x 100 CFM delta server fans blowing through it. The performance was suitably spectacular at not even 10% of the flow rate of a D5 at full speed.
|
# ¿ Jun 21, 2019 16:21 |
|
Not exactly the same, but I took an AIO installed GTX 1080 (MSI Sea Hawk X) and switched it to an ekwb full cover block when I built my custom loop. The GPU temps on the AIO settled around 60C, on the FC block they settle around 50C. I would expect roughly the same (-10C) change on a 2080 going from AIO to a custom full cover. Although some of that might be simply from switching thermal paste from the stock white paste on the AIO to thermal grizzly kryonaut on the full cover. At idle the GPU quickly settles to ~1-2C delta above the coolant. The only change I would expect would be you would probably get one extra boost bin out of it with the same overclock settings applied.
|
# ¿ Jun 23, 2019 16:45 |
|
VostokProgram posted:Would it be a bad idea to use stuff like Novec and other fancy coolants in standard water cooling equipment? Aside from cost (I assume they cost a lot?) It would be a very bad idea; Novec boils at like 40C. That stuff is designed for full immersion cooling, not a traditional loop.
|
# ¿ Jul 16, 2019 12:08 |
|
Actuarial Fables posted:I've got a case, Norco RPC-432, that has two 120mm fans in front and looks as though it could house a 240mm radiator. However, there's only about 3/8ths of an inch between where the left fan end and the case wall begins. I tried putting in an old 120mm AIO (Corsair H75) to see what would happen, and I wasn't able to get it to fit. The AIO extends 1/2 an inch from the edge of the fan. I would check the rack and see if it has enough clearance so you could just drill and then cut open a slot in the side of the case big enough to fit the radiator with the edge sticking out.
|
# ¿ Oct 26, 2019 18:34 |
|
|
# ¿ May 3, 2024 13:42 |
|
Endymion FRS MK1 posted:Another quick question. It occurred to me that I have a few spare Fractal Design Dynamic X2 GP-12 fans. Would I be at any disadvantage running these over EKWB's EK-Vardar EVO 120ER? The Vardars are decent high static pressure fans but I'll be honest and say I was less than impressed with mine, I had 3 of the 140MM ones and one of them started to squeak after only a year of use. I eventually replaced them with the black noctua fans which are a little quieter and perform just as well. The thing to keep in mind about water cooling, or really all cooler types is thermal resistance increases as the temperature delta between what's being cooled and the ambient temperature decreases. So basically the closer to room temperature something is, the harder it is to remove heat from it. And that also means the further above room temperature something is, the easier it is to remove heat from it. So at less than a 10C temperature delta it is extremely difficult to improve the performance of a radiator, but past 10C it becomes extremely difficult to make it get any hotter because these things can dump insane amounts of energy once they cross that threshold. The fan throttle on my radiator has next to zero impact on my full load (~375w) temperatures, but turning on my room air conditioning almost immediately results in a matching coolant temperature drop regardless of the fans or their throttles.
|
# ¿ Dec 22, 2019 14:39 |