Seeking Advice About "Custom-Water" Cooling Kits

[BonzaiDuck]

I'm sitting here with a now-dated Sandy Bridge system which actually "evolved" since summer, 2011. Tweaked and tuned, I didn't need water-cooling then to keep temperatures manageable at variously 4.6 and 4.7 Ghz, and I don't need to do anything with it now. A colleague of ours, who may go by the handle of 996GT2 now, had a similar rig with the 2700K CPU under custom-water cooling. He's been able to achieve 5GHz, but for 24/7, he said he ran it at 4.8.

Now I'm counting my Quicken accounts and putting together a plan to build a Haswell-E system next year. It will likely use the i7-5820K or i7-5930K processor. These "E" cores are still apparently fabricated with indium solder, no less because they have a TDP of 140W!!

There seem to be early indications that you can overclock the 5820K to 4.4Ghz with a CLC or AiO cooler -- without raising the VCORE and boosting temperatures into the stratosphere.

Someone else spoke of 4.6 to 4.7 Ghz with custom-water cooling. Of course, until I see some discussion in these forums, other sources such as early customer-reviews at the Egg are sketchy.

I have a choice of using either the NH-D15 air-cooler, a top-end CLC or AiO cooler, or custom-water cooling. Comparison reviews such as I've found at Frosty Tech standardized with a "200W simulation" show such units as the Nepton 280L with just a 5+C degree edge over the NH-D15. I believe, with my homegrown techniques, that I can match that difference and get better results for the D15 than reviews show. Other coolers, like the H100i don't seem to show any better.

I'm looking at custom-water kits. I figure, rather than ordering individual parts with no scars of experience and experimentation, it would be better to buy a kit and then add modifications or improvements later.

996GT2 seemed to indicate his 2700K rig would not exceed 50C with his water-cooling system under stress testing. My current rig reaches (average-of-cores) 72C with a 78F room-ambient.

Perhaps someone can recommend a custom-water kit -- Swiftech, AlphaCool, Phobya, Koolance -- or anything else.

I'm not so inclined to build a dual-loop or use it for the GPU(s): I think the new Maxwell cards will make air-cooling just fine for that.

If there are recommendations forthcoming, perhaps I could get an assessment of "weakest parts" included in the kit, or what I might expect after over-clocking a 140-Watt-TDP processor with it.

This will all drive my choice of a computer case, and I'll get to that bridge when I cross it.

Thanks.

PS. Price is not the primary consideration. In fact, it's the least important factor. The kits I've seen don't seem to exceed a distribution of prices with $400+ beyond the upper tail.

 

[aigomorla]

case you intend to install.
components you wish to watercool.

how big of an appearance impact you wish.. aka.. Bling.

need to know these... before we can help you piece stuff out.

The last one can inflate the value of the set by more then 300%
No joke... bling bling fittings can get pricey.

 

[BonzaiDuck]

case you intend to install.
components you wish to watercool.

how big of an appearance impact you wish.. aka.. Bling.

need to know these... before we can help you piece stuff out.

The last one can inflate the value of the set by more then 300%
No joke... bling bling fittings can get pricey.

Well, Aigo . . . "Bling" is an art-form. I spent a lot of time on my 2007 case-mod, mirror-film on the window, the paint -- all of it. I was totally prepared to continue with the project of cutting Lexan splinters for that window to use with LEDs connected to the RAID activity lights to make "lightning." I had downloaded pictures of actual lightning from the g**d*** Weather Channel. That's when I finally saw the folly of it.

So "Bling" is an art-form, but for me, it's a waste of time. The interior of my sig-rig is "neat;" I may even have matched LED intake fans, but that's as far as I go.

Up to this point, I was thinking I could mod one of my HAF 922 cases; as you know, because of that possibility, I was also thinking to use 200x200mm radiator(s). But there, I would've started at the case, then the radiator, and worked backward.

I've looked at the HAF-X -- which seems like the old 932 full-tower. That's an option, but it's also bigger than I'd like. And the point of it here, is this. I have no hesitation to make modest modifications to any case to accommodate a cooling kit -- and for this exercise, that means the radiators. One could buy a kit and then find alternative radiators, but it might help to avoid doing that.

I suppose I could see possibilities in the Phanteks Enthoo, but this and others I've looked at don't exactly raise my willie, either.

I'd rather buy a kit that offers optimum cooling, and then select a case for it. Does that make sense? I thought it made sense . . .

 

[BonzaiDuck]

Oh. Here's a thought to the case business.

Any case I have in the future is going to have a rather simple mod from the git-go.

Double caster wheels, lifting the case up off the floor by . . . . jus-a-minute-while-I-get-my-ruler . . . well-- 2.5". $5 at the local metal-store for some 1/4"-thick aluminum bars, some precise measurements, and prepping of the bars to fit the 6/32 holes for the rubber standoffs on the bottom of any case.

I'm thinking there's room there for fan or radiator installation -- outside and on the case-bottom. That's an option.

 

[krnmastersgt]

How much work do you want to put into tinkering with the kit? I will say I enjoyed my adventure into water, that being said there are a few areas for frustration/annoyance I could have dealt without but it was fun for me regardless, though I do question some of my component choices still.

Knowing you and your past projects/threads I'm assuming you'll want to tinker with it and fine tune it, which is always fun! Don't think it'll be necessary to do external rad mounts for just a CPU loop, while it'd be fun after a nice size rad or two you'll be getting so close to ambient that adding more rad space isn't going to affect the temperature curve that severely.

I think starting with the kit presents problems in your situation on the basis of where do you want to have the pump mounted and are you going to go nuts with multiple points in the loop? There are some budget pumps out there that work fine for the kits they ship with, but when you want to go crazy with adding more components into the loop or super high restriction parts you'll require/want a stronger pump.

Also very importantly as aigo pointed out, bling components make your prices rise quite quickly. I'm roughly 80% done with my WC set up so far, and I've sunk a good 800 or so in the components in my loop, along with more on parts that didn't make it in/changed my mind/going into next project so yeah, they get expensive quickly.

 

[BrightCandle]

A really hi specced water cooling kit that is well over the top is theoretically 10C better than a normal custom kit, and realistically its only 5C cooler. That is typically enough to get maybe another 100-200Mhz out of a CPU so long as you are willing to push the dangerous voltage through the chip. A well specced custom loop in my experience isn't more than about 5-10C cooler than an AIO and air cooler for just cooling a CPU.

A custom kit these days is mostly about noise rather than top level performance. Most Intel CPUs today are limited by the voltage and not thermally when overclocked, so any high end cooler can get near enough the safe maximum out of the CPU, high end air and AIO's are more than sufficient. Where the custom kit comes in is it can cool the GPUs, CPU and anything else and do so by spreading out the heat over a larger area of radiators and allow lower speeds of fans. 4 120mm fans at 600 rpm are nearly silent and out perform a single 120mm fan at 1500rpm or more of an AIO. More cost of course but much quieter.

I think just for the purpose of overclocking a CPU the custom kit isn't worth it unless you are willing to go into well above the 200W range of cooling, which is going to use potentially CPU killing voltages (up to or even exceeding 1.4V) for the purpose of squeezing out the last 100-200Mhz the CPU can give.

 

[BonzaiDuck]

A really hi specced water cooling kit that is well over the top is theoretically 10C better than a normal custom kit, and realistically its only 5C cooler. That is typically enough to get maybe another 100-200Mhz out of a CPU so long as you are willing to push the dangerous voltage through the chip. A well specced custom loop in my experience isn't more than about 5-10C cooler than an AIO and air cooler for just cooling a CPU.

A custom kit these days is mostly about noise rather than top level performance. Most Intel CPUs today are limited by the voltage and not thermally when overclocked, so any high end cooler can get near enough the safe maximum out of the CPU, high end air and AIO's are more than sufficient. Where the custom kit comes in is it can cool the GPUs, CPU and anything else and do so by spreading out the heat over a larger area of radiators and allow lower speeds of fans. 4 120mm fans at 600 rpm are nearly silent and out perform a single 120mm fan at 1500rpm or more of an AIO. More cost of course but much quieter.

I think just for the purpose of overclocking a CPU the custom kit isn't worth it unless you are willing to go into well above the 200W range of cooling, which is going to use potentially CPU killing voltages (up to or even exceeding 1.4V) for the purpose of squeezing out the last 100-200Mhz the CPU can give.

THANX, BrightCandle!! Very much. As I said, I had some off-the-cuff indications about these E chips. I hadn't even ruled out the NH-D15.

I thought MAAAY_BE -- from things I'd read in these forums over past months and years, that 20C reduction was achievable with custom-water. But the bench results at Frosty-Tech and other places put the margin of high-end AiO coolers over the D15 at just over 5C.

If custom-water only nets 10C+ over AiO, I think I can close the gap between AiO and D15-air-cooling. I sure as heck did it for the D14, and with not much in the way of "extra noise." That is -- I didn't "close the gap," but I'm pretty darn sure I got 10C lower than what the benchies showed for D14 under over-clocked stress.

Well, like I said in a recent PM I sent -- there's the 6-P principle: Piss-Poor-Planning-Prevents-Proper-Performance.

 

[BonzaiDuck]

Incidentally -- and adding a successive post to assure it won't be missed.

Different people have different ideas about "hanging their a** over the edge of the envelope" with voltage settings. For me, it just doesn't make much sense.

With each die-shrink, not only do the operable and normal voltage ranges get smaller, but the "strange, quantum-physics" stuff in addition to electro-migration would SEEM more likely to occur. I say "seem" because I don't have any serious exposure to "electrical/electronic engineering," I got through relativity-physics and Thermodynamics before 1968, and I know PhD physicists and physicists whose careers turned toward information sciences.

I think the succession of die-shrinks in their relative numerical values could be the only guideline we have for choosing a "safe-range" voltage upper limit. And if you can't over-clock a processor without going beyond that limit, it might be best to stick with what you have. This then drives your choices over cooling options. You either want to attack the cooling problem because you're pushing the edge of the voltage envelope, or you choose a cooling option that's "acceptable" given your voltage . . . frets, concerns, self-imposed limits.

 

[Essence_of_War]

I think just for the purpose of overclocking a CPU the custom kit isn't worth it unless you are willing to go into well above the 200W range of cooling, which is going to use potentially CPU killing voltages (up to or even exceeding 1.4V) for the purpose of squeezing out the last 100-200Mhz the CPU can give.

I agree with this. I don't have any WC experience personally, but the accounts that I've read of it suggest that WC actually gives the most dramatic results when applied to GPUs. Especially given that you're already willing to do some homebrew to tweak your high-end air coolers, Duck, I think you'll be much less impressed by the results of a custom WC loop applied to your CPU, than you would if you applied it to your GPU.

Here are some sample plots from Dustin at Anandtech's experience:

Here's CPU:
http://images.anandtech.com/graphs/graph7363/58370.png
Meh and meh.

Here's GPU:
http://images.anandtech.com/graphs/graph7363/58371.png
:awe:

And his cooling is quite a bit under spec'd! He's throwing a pair of OC'd GTX 780s, and an OC'd i7-4770k for something like 750+W at a total of 5x120mm fairly thin rads that might be good for ~100W each at safe coolant temp increases if you're not into pushing the fans to gale-force winds.

Your gpu might actually shiver if you spec'd your loop properly!

 

[BrightCandle]

My GPUs idle in the 24C range (a few C over ambient) and at load they run at 40C. Even overclocked it makes basically no difference. Watercooling dramatically improves the GPU temperatures but what it doesn't do is change the overclock much. I suspect it makes a little bit of difference, 10's of Mhz maybe but the temperature change that water brings just isn't sufficient to really get much out of it. Improvements with watercooling come from being able to dissipate more heat and hence drive more voltage.

The issue with the new Haswell-E is that its a really high TDP chip. So the AIO coolers are kind of necessary when overclocking and you might find the benefits of custom start to play out. Lets say you have a Haswell 8c16t, you overclock it from 3.0Ghz to 4.5Ghz. That is likely going to near double its power consumption. That might well put you into 300W of cooling necessary.

An air cooler can't cope with that, they tend to top out around 150W effectively, so its going to be unnecessarily hotter than it should be. An AIO on the other hand with 2x 120mm fans will likely cool about 200W to 10C water delta, but this is a 300W source so its going to be about 17C and probably a bit more because most AIO's are really low quality CPU and radiator coolers and it will probably be running at 1200-1500rpm (quite noisy), but at its top range it might be dissipating to maybe 250W @10C.

The custom kit in comparison we can put a 4x120mm radiator in there, something like a Thermochill PA series which can dissipate 130W per 120mm at 800rpm fans. So we can be looking at a water temperature of about 6C, which is 10C less than the AIO and its going it with fans less than half the speed because we could (a) get a better quality radiator, (b) we could get more radiator space and (c) we control the fan speed.

The AIOs are a little better than air cooling, it tends to play out more with the bigger radiator AIOs and at higher heat dissipation where spreading the thermal load over more space becomes beneficial. Watercooling is not magic, the cooling effectiveness of it is entirely down to the fact that we move the heat away very easily with water (that takes heat energy very easily but takes a lot of it to gain temperature) and then get to spread that energy out over an efficient system for exhausting the heat via a radiator, the more space we spread the heat over the lower the water temperature will be and the cooler the CPU will ultimately be.

Personally I comply with Intel's recommended voltage ranges when they give them. Many people exceed these (most review sites do) and I think that is kind of foolish, if Intel says its the maximum then they are the people that would know. The last 100-200Mhz is normally very hard fought and its just not worth the instability IMO, I use water to reduce noise while maintaining a safe 24/7 overclock.

 

[BonzaiDuck]

I agree with this. I don't have any WC experience personally, but the accounts that I've read of it suggest that WC actually gives the most dramatic results when applied to GPUs. Especially given that you're already willing to do some homebrew to tweak your high-end air coolers, Duck, I think you'll be much less impressed by the results of a custom WC loop applied to your CPU, than you would if you applied it to your GPU.

Here are some sample plots from Dustin at Anandtech's experience:

Here's CPU:
http://images.anandtech.com/graphs/graph7363/58370.png
Meh and meh.

Here's GPU:
http://images.anandtech.com/graphs/graph7363/58371.png
:awe:

And his cooling is quite a bit under spec'd! He's throwing a pair of OC'd GTX 780s, and an OC'd i7-4770k for something like 750+W at a total of 5x120mm fairly thin rads that might be good for ~100W each at safe coolant temp increases if you're not into pushing the fans to gale-force winds.

Your gpu might actually shiver if you spec'd your loop properly!

Why do I think the bar graphs in the first link are mislabeled, or reversed? It doesn't make sense, that you would have higher temperatures with water-cooling, unless I misread the bar graph to begin with.

There is at least one thing you and BrightCandle imply -- and I would've assumed. The larger the radiator with sustained airflow, the greater the thermal wattage capacity for these high-end CPUs.

On the other hand, the die-shrink and reduced size of the processor under the IHS, the more limited the heat removal over a range of TDPs.

This then goes back to my idea of using a dual-radiator approach to "totally-custom" individually-ordered parts. AigoMorla insists that using 200mm radiators with 200mm fans will fail, because the static pressure on the fans is low. On the other hand, I'm wondering why they would then make those radiators in the first place.

My dilemma here, whether I were to order parts and start building now, or wait as planned until next year -- I want to know exactly what I'm doing before I press any check-out buttons, and I want to feel confident that it will be adequate.

At minimum, we can order one of the 6-core K processors, boost the clock a little bit to a point before VCORE needs to be higher, then call it a day.

But it was evident with IDontCare's delidding project -- which doesn't apply to a soldered E processor -- that even profound success in reducing temperatures matters little to mitigating voltage increases. You're going to run up the voltage to about the same level, even if you can keep the processor at 40C.

Phase-change, anyone? That's a $900 price-tag. Also a power-guzzler. And always the risk that something can fail-- even within some warranty period.

If I had the money to spend, maybe I could be optimistic. We just replaced a 30-year-old refrigerator -- still running -- with a 21st-century marvel. After playing with the touch-screen, I was hoping I could download an "app" for it so I can watch the "Food" channel in the kitchen "on the fridge."

I can keep more "stuff" in the fridge, now. I'm not sure I can "do more stuff" with an "E" processor.

 

[aigomorla]

well... if your still going to commit to water...
Personally i like Koolance blocks...
Koolance pumps are also great, as they carry the best warranty, being both flexible, and also being located on stateside.

People like swiftech pumps because of the pwm feature they have, however, swiftech has the most stringent of warranty in being you can never open them up.
The owner gabe has even told us, that he recommends user to use pine sol to clean the pump instead of open them up as it would void your warranty.

Koolance allows full disassembly of pump, as long as there is no physical damage you have caused, they will replace them, and again, located in the USA. (the state of Washington to be exact.)

I love koolance waterblocks, due to the very high quality in nickle plating they use.
I have yet to own one which started pealing, unlike eK. Koolance has been plating things longer then all the vendors combined including swiftech, so they know how to plate.

For radiators, i like XSPC typically the RX series.
I like big thick rads, due to the noise / performance ratios they pack.

For fittings, i will always be a Bitspower fan.
I have worked with Vincent a lot, (owner of Bitspower) and he is a very friendly guy to get to know.

 

[BonzaiDuck]

I'll block copy Aigo's response and paste it in a file for future reference.

If I planned to build the system next year, I think I've got a lot of work to do in the meantime just to get ready . . .

 

[nitrousninja]

I started out with a Swiftech H220X and added a block to my 290 and an extra radiator. Not a bad start to water cooling. The H220X gives you a lot for $150.

 

[BonzaiDuck]

I started out with a Swiftech H220X and added a block to my 290 and an extra radiator. Not a bad start to water cooling. The H220X gives you a lot for $150.

Again -- thanks. I'll hope for even more from others.

Maybe this is a good thread for the C&C forum -- focusing on the E processors and how one addresses the elevated thermal power in overclocking them.

With air-cooling -- if adequate -- a lot of folks don't need to think much about it except for measuring the case and choosing the fans. [Although there ARE some things you can do.] AiO -- again -- not much trouble. But in those cases, effectiveness is limited in this context of the E processors.

With the potential thermal wattage of these E cores, though, one has the first level of trouble just choosing parts or a kit for custom water. The second level means figuring out how to make it more effective for the processor TDP elevated to the levels others mentioned here.

 

[Essence_of_War]

Why do I think the bar graphs in the first link are mislabeled, or reversed? It doesn't make sense, that you would have higher temperatures with water-cooling, unless I misread the bar graph to begin with.

For the CPU graphs, my guess is that it's because as I mentioned, the loop is under spec'd. The GPU and CPU are sharing X-watts of cooling, and the CPU is getting less of a share than it does with air when the HSF is sort-of decoupled (or at least more loosely coupled compared with a water loop) from the GPU air cooler.

I'd need to read more carefully, but I also seem to recall his throwing a lot more voltage to squeeze out an extra +100 MHz on the water loop, just because it could take it.

 

[BonzaiDuck]

For the CPU graphs, my guess is that it's because as I mentioned, the loop is under spec'd. The GPU and CPU are sharing X-watts of cooling, and the CPU is getting less of a share than it does with air when the HSF is sort-of decoupled (or at least more loosely coupled compared with a water loop) from the GPU air cooler.

I'd need to read more carefully, but I also seem to recall his throwing a lot more voltage to squeeze out an extra +100 MHz on the water loop, just because it could take it.

I suppose that's a reasonable explanation.

I also took a fresh look today at the phase-change options. There are forums on other web-sites, maybe "Overclockers." You can buy compressors and build your own -- some people do that. I saw a conversation over the question of "power consumption." I'm wondering if -- choosing a pre-built phase-change cooler -- you could simply keep the processor temperature in a range around 35C. I did that with my "home-built wine-cooler" -- a deep-freeze with a Johnson Controls digital controller that keeps the wine between 53 and 57F. The power-draw for the freezer is minimal. But there's little in the way of useful comparison there: you'd only suspect that an expensive device like those sold at Frozen-CPU would provide that level of control.

This is all uncharted territory. The Haswell-E chips are "uncharted territory." Highest TDP I ever remember was closer to 135W.

[Freakin' earthquake might up-end my plans to buy cases of 2013 vintage -- touted last year as "going to be the best." Only the Shiraz and the whites are appearing right now . . . ]

 

[krnmastersgt]

I already have plans for when I'm tempted to go phase change, converting a mini-fridge into a cpu cooler/fridge hybrid so that I can get my game on and have nice cold beverages. I think rigging that up wouldn't even be that difficult aside from the anti-condensation steps you'd need to take for the motherboard and other components.

As for the temps you get with custom loop vs aio on CPUs, I think the primary reason you see such a relatively low difference is firstly surface area, the die of a GPU is considerably larger than a CPU so you're able to cool the whole down more effectively. But there's also what most de-lidding experiments seem to reveal, that the lack of solder isn't the issue but the gap that is a result of using TIM instead, making for a very poor contact in most cases.

Not much you can do about the first but the second when addressed I believe should fairly dramatically affect CPU cooling curves, also why you might see better deltas with the Haswell-E chips since they don't have that problem.

 

[guskline]

BonzaiDuck: A custom water cooling kit is probably the easiest way to get started but there is no "easiest" way other than taking the leap.

Personally, I started slow with BrightCandle's sage advice along with aigomorla and others.

I scoured ebay offerings and bought a used RX360, Swiftech 655b and XSPC Acrylic bay/res. I continue to use all 3.

As you get deeper into the "adventure" called custom water cooling you start wanting more rad space. Heck, I now own a MO-RA3-420 Pro!

I've spent some serious$$$ at FrozenCPU etc.

Here's some lessons I've learned.

For custom water cooling buy the biggest rad friendly case you can afford or be prepared to use external rads like I do.

Be prepared to spend $$$ to finish the job.

If you are only cooling the cpu I would stick to a AIO but if you are cooling the gpu(s) custom is the way to go.

If you are going to leap into the Haswell-E, especially the 5960x, Please buy a rad friendly (i.e. Big) case. Tons of heat!

Good luck.

 

[BonzaiDuck]

BonzaiDuck: A custom water cooling kit is probably the easiest way to get started but there is no "easiest" way other than taking the leap.

Personally, I started slow with BrightCandle's sage advice along with aigomorla and others.

I scoured ebay offerings and bought a used RX360, Swiftech 655b and XSPC Acrylic bay/res. I continue to use all 3.

As you get deeper into the "adventure" called custom water cooling you start wanting more rad space. Heck, I now own a MO-RA3-420 Pro!

I've spent some serious$$$ at FrozenCPU etc.

Here's some lessons I've learned.

For custom water cooling buy the biggest rad friendly case you can afford or be prepared to use external rads like I do.

Be prepared to spend $$$ to finish the job.

If you are only cooling the cpu I would stick to a AIO but if you are cooling the gpu(s) custom is the way to go.

If you are going to leap into the Haswell-E, especially the 5960x, Please buy a rad friendly (i.e. Big) case. Tons of heat!

Good luck.

If I chose the 5960X, I wouldn't attempt to overclock it. Somewhere in reviews I recently read, the reviewer observed that the two extra cores (over the lesser models) made the thermal problem and overclocking -- greater. The X processor would probably be a great candidate for AiO.

Anyway, I think I fit the pattern of many folks here. I'd rather get the 5930 or 5820 and spend the money on cooling, case and peripherals.

If I had to pick a case to follow your prescripts immediately, it would be a HAF 932 or HAF X. Ugly -- I don't mind. I've had good experiences with the 922's I'm using for three systems here. But there are other cases which may offer similar features. And it's too easy to go with what you already know.

SOMETHING ELSE: Last year saw the cooler-makers in high competition. For instance, I think it was CM which introduced the Nepton 280. We also had a big promotion for Noctua's D15. I suspect -- by the time I get to the starting gate on this, there may be some newer entries.

AND AGAIN . . . We crossed paths on this thread during the spring months:

http://forums.anandtech.com/showthread.php?t=2380437

Interesting -- a "420" that would accommodate an array of Akasa Vipers. Of course, moving the computer around would be a two-fold operation, but I said the ssame thing in that thread . . .

 

[BonzaiDuck]

I already have plans for when I'm tempted to go phase change, converting a mini-fridge into a cpu cooler/fridge hybrid so that I can get my game on and have nice cold beverages. I think rigging that up wouldn't even be that difficult aside from the anti-condensation steps you'd need to take for the motherboard and other components.

As for the temps you get with custom loop vs aio on CPUs, I think the primary reason you see such a relatively low difference is firstly surface area, the die of a GPU is considerably larger than a CPU so you're able to cool the whole down more effectively. But there's also what most de-lidding experiments seem to reveal, that the lack of solder isn't the issue but the gap that is a result of using TIM instead, making for a very poor contact in most cases.

Not much you can do about the first but the second when addressed I believe should fairly dramatically affect CPU cooling curves, also why you might see better deltas with the Haswell-E chips since they don't have that problem.

So far, there is only sparse "hum-int" on the E chips. Not sure whether it was 5930 or 5820K, for which someone boasted he could get to 4.4 without increasing the voltage. We'd want to see some lab reviews -- maybe they're already "out there."

Again -- good idea if you can hook up the mini-fridge to do double-duty, but it's also the same problem with all external solutions: umbilical cords. I was looking at that MO-RA3-420. The number of fans (140mm?) required reminds me of Howard Hugh's "Spruce-Goose." Maybe you could replace nine 140's with four 200's. Can't rightly say at the moment. This sort of thing requires commitment -- for the down-side as well as the up-side.

 

[BonzaiDuck]

BonzaiDuck: Heck, I now own a MO-RA3-420 Pro!

Folks who've been around for the last ten years would agree that I'm always looking for some cheap, simple -- ingenious -- mod opportunity.

Why wouldn't it be possible to mod a case side-panel and mount that 420 there? You could mount the fans either way and either internally or externally -- depends on how it's all mounted to the case-panel and/or frame. Or mod the case some more so they'd fit. I'm thinking there must be a way to do it without permanently impacting the "warrant-ability" of the radiator -- no changes to the basic metal part with its welds.

Also, are you using an internal reservoir for that rig? Or could you go either way, only to make sure that the "T" is above the radiator, or flips upward to assure it? Enlighten me. . . . . really . ., . I don't know s***.

And also, I can't see that it would make any difference as to what sort of sheet metal you might add -- if any -- really. Aluminum for the lazy man -- it works more easily under cutting tools if you know what you're doin'. Where or possibly how would any corrosion occur, especially if covered with primer and paint? Even if it mattered, you'd be able to take measures.

What's the capacity of that radiator? I'll do another web-search, but you would know.

 

[guskline]

The MO Ra is made by Watercool. It is the equivalent of three 420 Rads!

 

[guskline]

I have nine 140 mm fans on one side. The Pro, which I have, can use nine additional fans for push/ pull

 

[BonzaiDuck]

If "looks" take a back seat to neatness, structural integrity and balance, here's what I think could be done -- [and I'll look at the other radiator you mentioned last.]

I fit all my HAFs with 2.5" double-braked-casters. These have the square plates with the holes in each corner. I buy 1/4" x 2" x 12" aluminum bars -- usually precut to an approximate excess -- at the local metal shop. I tap the 6/32 holes for screws fitting the caster-plates. I drill two holes in each bar slightly looser than the 6/32 taps that CM puts in the case-bottom for the rubber feet.

So far and always, these bars were the exact width of the case, flush with the frame. However, by extending the bars an extra 3+" to accommodate the radiator, there is then the interior space available for the fans. Or somehow the radiator must be made to fit with clearance for video cards and other case interior items, and with fans on the case exterior.

I actually think that a CM Stacker could be used for this or that option, because it has a separate acrylic frame for 140mm fans -- x4. Or maybe one could appropriate the frame and its hinges, fit it to a wider case, modify the acrylic construction to accommodate larger fans, and even mod the interior parts (a drive cage position, etc.) accordingly. If there were a way to fix four 180mm or 200mm fans to the case side panel, you might have options for selecting the size of the radiator.