Cooling them Hotties

[Dirk Broer]

Water cooling has a very bright future: the first 11 places in last month's Hardware Secrets monthly cooler test are for water cooling systems -as far as the cooling aspect goes, noise is quite another matter: Intel's high-end i-series stock cooler has a very good score there!-

Best air cooler -according to Hardware Secrets- is the ThermalTake Frio Extreme
It should be possible to fit three 140mm fans to this cooler to obtain even better results, IMHO. I even coin a name for this mod: Thermaltake Trio Extreme.

While I'm busy coining: The Ford Ka Cabriolet should be called Ford Kabriolet, shouldn't it?

[Dirk Broer]

120mm Case fan Shoot-out

What do we want in case fans? Silence and performance! These two criteria unfortunately do not quite match each other. The more silent, the less performance - and vice-versa.
Which fans stick out in the crowd? I've found a review that compares no less than 393 case fans -between 80 and 140mm- that came up with the following results for the best 120mm case fans, the most popular size these days. Are my CoolerMaster Sickleflows that bad and are Vaughan's Noctuas that much better? We'll see!

A small selection of the best fans in the various categories (noise under 12V, air displacement under 12V). I also took the opportunity to set real life performance against manufacturer's claims...

Brand	Type	Diameter in mm	Speed in rpm	Real Noise in dB(A)	Claimed Max Noise in dB(A)	Real Airflow in cfm	Claimed Airflow in cfm	real cfm/real dB
Akasa	Viper 120	120	1900	50.8	28.9	92.1	83.63	1.81
Bitfenix	Spectre	120	1000	28.2	20.0	40.8	43.50	1.45
Coolermaster	Sickleflow	120	2000	44.2	19.0	81.1	69.69	1.83
Coolink	SWiF2 1200	120	800	27.1	8.5	40.6	60.40	1.50
EKL Alpenföhn	Föhn120 Wing Boost	120	1500	39.0	24.8	74.3	106.0	1.91
Nexus	Real Silent Case Fan	120	1000	31.6	18.0	46.5	36.80	1.47
Noctua	NF-P12	120	1300	38.9	19.8	55.9	92.30	1.44
Scythe	Gentle Typhoon	120	5400	68.1	50.5	163.8	150.1	2.41
Scythe	Gentle Typhoon	120	4250	61.9	44.0	129.5	116.5	2.09
Scythe	Gentle Typhoon	120	1850	40.2	28.0	70.8	58.3	1.76
Scythe	Slip Stream PWM	120	1900	52.7	37.0	109.9	110.3	2.09
Tacens	Aura II	120	1200	33.8	12.0	50.2	50.00	1.49

You could argue that, at 5400 rpm, the Scythe Gentle Typhoon is not gentle anymore. Scythe Full Cyclone might be a better marketing name! But it delivers the most cfm (cubic feet of air per minute) per dB(A).
You also need a more or less seriously sound-proofed casing for this Gentle Typhoon, like the Corsair Obsidian below. Note that with a decrease of rpm the value in the last column for the Gentle Typhoon drops like a brick. The Sickleflow performs on a par with the Gentle Typhoon, would a 2000 rpm model of that type had been included. The Noctua does not quite live up to it's imago.

Note that I've left out another important variable in choosing the best fan: the Air Pressure. Given the huge gaps between claimed and actual performance I thought it wise not to supply you with just the manufacturer's claims.

A true Noctua Fan, using -at least- seven Noctua fans (we can't see what he has hidden in the top):

It turns out he has nothing hidden under the top. I'd replace the bottom 92mm fan with a 140mm fan and add an extra two under the top in his case (pun intended),
but that might distort his wind-tunnel....

And why is it that case makers have not aligned the rear-exhaust with the CPU? If that were the case, noctua could add another lump of aluminium between the third fan and the case exhaust.
And by using the three lower drive bays to fit yet more 120 or 140mm fans you then could make one continuous wind-tunnel...

[Terry1953]

I just ordered 5 of these for a box I'm building si I'll let you know how it works out. http://www.newegg.com/Product/Produc...82E16835186080 The fans are actually costing me more than the box but I really like the boa since I already have 6 of them. If I get what I asked for for Xmas I'll finish it up then. I7, 32 Gb. 128Gb SSD, 2 7970's and room tor 2 more on the MoBo. It'll need some serious air flow.

[Dirk Broer]

140mm Case fan Round-up

And how will Terry's Arctic F-14's fare? Unfortunately this model was not amongst the tested fans. All figures again under 12V. Noise was measured by placing the fan in a sound-proofed box and taken at an angle of 90 degrees at 10cm distance.

Brand	Type	Diameter in mm	Speed in rpm	Real Noise in dB(A)	Claimed Noise in dB(A)	Real Airflow in cfm	Claimed Airflow in cfm	real cfm/real dB
Antec	TrueQuiet	140	800	26.6	20.0	51.10	32.40	1.92
Antec	TwoCool	140	1200	38.4	26.0	78.10	58.90	2.03
Be quiet!	SilentWings USC	140	1000	35.4	16.5	77.50	60.40	2.19
Cougar	VortexHDB	140	1400	41.9	19.2	104.0	70.50	2.48
Noctua	NF-P14 FLX	140	1200	44.2	19.6	97.30	110.3	2.20
Noiseblocker	NB-BlackSilentFan XK1	140	800	29.4	19.0	54.90	35.00	1.87
NZXT	Enthusiast	140	2000	55.6	37.0	150.6	98.30	2.71
Scythe	Slip Stream	140	1700	50.0	36.4	114.6	92.40	2.29
Thermalright	TY-140	140	1300	39.4	21.0	92.80	74.00	2.35
Titan	Kukri PWM	140	1800	47.2	28.8	110.9	89.43	2.35
Xilence	2ComponentFan	140	1200	48.1	26.7	111.0	71.80	2.31

[Nflight]

After years of learning and appreciating my system for how it works, and works well. I think the best solution in any build from now on is to remove the PWR Supply from the mainframes location. This unit creates heat by itself that can overwhelm the whole system, and thus it takes up space as well. So why not devise a remote PWR supply for any hardware application. Heat produced and located out side the realm of the rest of the heat generating units like the CPU & GPU etc.

My Thinking is to take the PWR supply one step farther and create a true supply of PWR by developing a Heat Pump inside the PWR Supply that could allow an installer to arrange the pipes transporting the working fluid into a systematic arrangement of individual chillers or condensers to locate each excessive heat machine unit (aka CPU & GPU) and take each units heat and evaporate the working fluid then capture the fluid in a circulating system to return the fluid back through a radiator and then back into the system. There is one location I did not mention and that is the incoming and exhausting air flow for each box. The incoming Air to the boxen would have a radiator/condenser that would chill the incoming air coming into the boxen where all the working crunching is taking place. A single fan would be necessary to be either in the very beginning or the very back of the system to bring the air in. No more 10 or 12 fans per box. Just one fan and channeled air flow.

Having that PWR Supply sitting farther away from the existing operating electronics may cause inherent spurious emissions of flux or electromagnetic emissions that could cause communication interference. So how about some shielded power cabling to solve that dilemma before it even gets mentioned.

Its a National Holiday and I had some time to postulate my thoughts...

P.S. One more thing I forgot to mention is that I plan to use this type of scenario with incorporating my own Geothermal home unit with a secondary zone in my yard providing for chilling capacity for my boxen once I reach the threshold of income I hope to one day come by!

[Brucifer]

If a system puts out X amount of heat, it really doesn't matter how it exhausts it when viewed from a purely heat produced perspective as no matter how you get rid of it you are still getting rid of X amount of heat. The only thing this impacts on is the noise factor......

[Dirk Broer]

As for getting rid of heat: We're mounting our coolers in the wrong direction!
Traditionally we have, after years of either blowing down or sucking up, coolers that blow in the direction of the rear ventilator.
This appears not the ideal solution for modern cases that have ventilation on top.
The new way of blowing is in the direction of the top-ventilator, which needs to suck the hot air out of the case.
The CPU-cooler doubles as secondary GPU cooler by sucking hot air from the back of the videocard upwards to the top ventilator.

[Terry1953]

I find that the PWS is a fairly closed off unit in most boxes. On my bottom mounted boxes it takes in room air with it's own fan and expels it out of the box back into the room. On a top mount it takes the warmest air in the box and expels it out of the box and into the room. And on it's worst day it can't begin to match the heat output of 2 6970's that are in the center of the box. The trouble is the total amount of heat generated and what to do with it. In the winter it is useful once you get it out of the box as a supplement to the furnace by moving it around the house. In the summer it would be more helpful to put it out of the house, or focus it on something useful like supplementing the water heater. At this Jason seems to excel with his watercooling. He has spent years learning to move the heat where he wants it to either dissipate it or use it to his advantage and turn it into money savings. The real problem as I see it is getting use out of the excess heat at a cost of less than the cost of just expelling it and eating the cost of generating it in the first place. As in if it cost me $10 to generate 1000 btu's and I expel the btu's outside, I have a total cost of $10. But, if I take my $10 worth of btu's and spend $4 extra to run a system that moves it to supplement the water heater and it only saves me $1 I have a net cost of $13 and I'm better off tossing it out the window. For some reason I get a picture of a Chevrolet Leaf popping into my head.

[Jason1478963]

Nflight did you share a similar cooling tactic used in data centers? It was done with cooled air delivered to the front of the racks and the hot pulled from the rear of the rows of racks. I believe it was two rows back to back with a door to access the back of the racks. I believe they reduced the fan noise greatly with this setup. I'm not sure how the air was treated, but with a newer building and heat pumps connected to water loops this heat could be put to use in the building elsewhere. I look forward to pictures of your completed systems for your geothermal home.

Dirk Broer This sounds like a great idea if you have a cooler that allows different mounting positions. I have still been using stock AMD coolers for my systems that are on air yet and they aren't that fancy yet. When the case allows it i try and make an intake duct from a plastic juice bottle to draw air from the vented cover. This works ok for systems I have running cpu based projects.

Terry1953 thanks for the recognition of the system for heating water. The biggest chunk of the cost for me was the initial waterblocks and pieces parts. I would say with my setup in the basement it doesn't really use more electricity to pump the water for heating as one 50 watt pump replaces 6 or more cpu fans and/or gpu fans. The week spot in my system is the thermal siphon heat exchanger and small tank. I don't overclock all that much as I like the ability to heat water and don't always use the hot water fast enough in summer time when running several systems. In the winter time I plan on maintaining a certain temp with the koolance controller in the pre-heat tank by circulating the water into the living room zone to help heat it. I'm not sure you could run an exhaust fan for the power the 2 cooling loop pumps draw. :P What are your doing to cool your facility with all heat producing machines your running? If you can do it without using a/c to fight the extra BTUs it would be a considerable energy saver. (1 kWh is the energy equivalent of 3412.3 BTUs)

[Dirk Broer]

This sounds like a great idea if you have a cooler that allows different mounting positions. I have still been using stock AMD coolers for my systems that are on air yet and they aren't that fancy yet. When the case allows it i try and make an intake duct from a plastic juice bottle to draw air from the vented cover. This works ok for systems I have running cpu based projects.

AMD are still using old-fashioned down-blowers for their products, even for the 125-Watt FX line. Though I would gladly put it on one of my 65-Watt APUs (it looks more than twice as high as the AMD coolers that came with my 65-Watt APUs and the fan looks bigger too), this seems a bit inadequate for overclocking a FX as compared to a good air-cooler: