Water cooling for the masses

Water cooling has finally begun to saturate the mainstream PC-market. As a result, technical writers focusing on products once made for H20-Enthusiasts and/or Overclockers must appeal to a much wider and diverse audience. In it's infancy water cooling was an esoteric hobby attracting a niche group of individuals, who were adept at improvisation based on necessity. Given the absence of any water cooling manufacturers per se, many early water blocks were hand-crafted (machined) by H20-Enthusiasts themselves. Pumps and other accessories were found at Aquarium suppliers, heater-cores from Automobiles scrap yards and case fans served as radiator fans. To large extent water cooling is still an esoteric subject since most people either slept or passed notes during their High School Physics classes. As CPUs shrink and transistor counts increase heat is the inevitable result, water cooling may replace air cooled heatsinks in the near future.



Insofar as journalists are concerned, where it once was possible for a writer to delve directly into discussing flow rate, head-feet, Delta-T, impingement zones, metallurgy, and even meteorology (i.e. relative humidity) they must now speak for the layperson. Unfortunately many of the sites whom can afford the proper testing equipment have let this get to their heads. They can be condescending in their treatment of the reader, expecting end-users to be "in the know." Forgoing a lesson in the social psychology of egocentrism, those self appointed "Pro" H20-Cooling sites while providing valuable data for their own Forum members and assorted lackey’s, seem to resent budding enthusiasts.

At present there are only a handful of sites I would recommend which integrate a valid scientific standard of testing with a user-friendly accumulative water block database. Caseumbau continually tests 20 CPU-water blocks updating the list as water block performance dictates, using the following TEST stand. Watercoolplanet continually tests 50 CPU-waterblocks from around the globe using the following TEST stand. And there’s also Overclockers.com growing comparison list. For 2006 I'll be focusing on water cooling kits, providing our readers with some of the best performance for price kits and those stores carrying or assembling those kits.


XSPC X20 High Performance Watercooling Kit

The kit arrived in one of the smallest boxes I've seen considering everything in there would add up to "Performance" or even the term "Kit."





The XSPC X2O WaterCooling Kit includes:

CPU Water block with Universal Mount

Compact 12V Pump/Reservoir

R120 Crossflow Radiator

80mm to 120mm Radiator brackets

Low Noise 1700rpm 120mm Fan

Clear 10/8mm Tubing

Anti-Corrosive Water Additive

Thermal Paste

Designed for:

AMD Sockets 939, 754, 940

Intel Sockets 775, 478, 603, 604



When XSPC delivered their X20 water cooling kit the introduction above seemed most apropos. The X20 is one of the most compact kits I've seen, this is especially true of the pump-station (reservoir) which did give me cause for concern.



A blue LED is included with the pump which fits into the Lucite top-plate, accentuating reactive fluids. Given this system's price point I was surprised XSPC would be as detail oriented.



Although I was a bit concerned about the unit's size, this is not necessarily indicative of performance as the LAING DDC and C-System pumps have shown. I was troubled by the fact the pump was secured to the base of the reservoir with four suction cups.



In three weeks of testing the pump never dislodged and there are advantages to this method of securing the pump. Suction cups are ideal for vibration dampening, hence sound. If the pump has to be serviced removing the station's (reservoir) top-plate is about all it takes. Finally the pump's smallish size should have a negligible impact on water-temps. Although this issue is often exaggerated since pump-stations add very little heat to the water. The copper based water block can be compared in its physical appearance to the Polarflo TT-series water block reviewed here.

The water block; "Where the rubber meets the road."



The X20 CPU water block is based on an 8mm-internal / 10mm-external hose diameter, the block is pre-assembled by the manufacturer making removal of the base plate almost impossible without damage. The only person who this might inconvenience is a reviewer wanting to snap a few photos (ahem). From the photo below the design looks to incorporate deep channels running in parallel from inlet to outlet, this would perpetuate flow across the base.



These channels are not only deep but fairly wide in comparison to more intricately designed blocks which, based on the channel width the X20 is less likely to have it's flow restricted by small particles. A good rinsing after extended usage should remove any foreign objects which may get lodged in there. If your system is bled and filled carefully, flushed before hand, one need not worry about this.



While the base plate finish isn't the best on the X20 it's not the worst either. The quality is high and most importantly the base is flat which is the most important aspect. There were no striations or pits and the surface was smooth.

AMD Socket-939 installation



As specifications indicate on the preceding page the X20 is compatible with Socket-939, 754, 940, Intel LGA-775, Socket-478, Socket-603 and Socket-604. While this may not be surprising, many water blocks are capable of multiple systems mounting, what's unique about the X20 is that all this is accomplished with a single mounting plate. Four spring-mounted bolts are supplied, two of which are unnecessary with AMD platforms. In this respect I must applaud XSPC in their goal to simplify multiple socket-mounting. Below we see the mounting plate and hardware.



Installing the radiator section was simple; four risers create an approximate 3.81cm buffer between the case rear and radiator ample for unrestricted air flow. The radiator can be mounted internally as well per X20's instructions. I mounted the radiator at the rear of the case with fan and fittings facing out to eliminate sharp bends or possible kinks in the hoses; and to ensure ambient room air was being drawn in. The radiator is well constructed thin walled steel with ample copper internal tubing. The unit is finished in black with two pre-chromed plates which screw into the sides of the radiator, these wrap around the front and rear of the radiator providing mounting holes for fan and riser screws. Over time I've come to discover the radiator is an integral piece of hardware in any H20 system and from our test results XSPC is well aware of the radiator's role. They sell some high-quality units including a series known as the Crossflow.





The system mounted easily onto the DFI Lanparty nF4 UT and into the Thermaltake Kandalf aluminum tower. What makes the unit a space saver is its incorporating the pump into the reservoir which has both its advantages and disadvantages. One advantage is the elimination of a stand alone reservoir and of course the additional length of tubing. As hose lengths increase pump pressure and flow rate is affected. A disadvantage to a submersible pump would be the amount of heat that pump dissipates into the water itself, raising temps. The ramifications of a submersible insofar as raising water-temps are widely exaggerated; in fact even the largest submersible pumps have very little impact on overall water-temp.

AMD Tests

AMD Test System
CPU	Althlon-64 3500 Retail Socket-939
Mainboard	DFI LAN party UT nF4 Ultra-D
Memory	Corsair TwinX2048-4000PT (2x1024MB DC CL3-4-4-8) Mushkin XP4400 (2X512MB DC CL2.5-4-4-8)
Graphics	AOpen Aeolus 7800GTX-DVD256
Power Supply	PCPower&CoolingTurboCool 850 SSI
Cooling	XSPC X20 Alphacool CPU-waterblockNexXxos XP nickel plated. Alphacool pump AP1510 Centrifugal w/Voltage regulator. Alphacool CAPE Cora 642 passive radiators. 8/10mm Internal/external tubing
Operating System	Windows XP

For our AMD test system the A64 3500 was run at default speed and Vcore 11x200FSB=2200MHz / Vcore=1.34V and then overclocked at 10x250FSB=2500MHz / Vcore=1.475V.

CPU temps were then measured at IDLE and LOAD in which the CPU was pushed to 100% LOAD using the CPU/Memory/VGA stress-test utility S&M. Measurements were taken with the Thermaltake Kandalf tower case bottomed up, the fan compliments = 120mm front intake, 90mm rear exhuast.

Tests were performed over a four hour period with a consistent ambient room temperature of 19°C/66°F. Thumbnails below enlarge CPU-Z, WCPUCLK, Smartguardian and S&M screenshots indicating system speed, voltages; temps and system status (IDLE or LOAD). CPU-Z was mis-reproting voltages which 9as stated above) ran deafult and then 1.475V overclocked.

2211MHz IDLE | 2211MHz LOAD | 2500MHz IDLE | 2500MHz LOAD

Intel Socket LGA775 Installation



As seen above the Socket-T installation on our Asus P5AD2-E Premium (and P5ND2-SLI Deluxe) didn't go smoothly. I decided the best solution was to remove the "Fanless Design" mosfet-sink Asus uses on the P5 boards. This is of-course and ad-hoc solution and most likely not conducive to a water-cooling kit sold perhaps for an entry-level user. None-the-less at this stage it was my only alternative and once removed I was in for quite a surprise.



It's obvious somewhere along the way an Asus employee confused toothpaste application with thermal paste application. Where this amount of thermal paste is used an opposite result is obtained. The paste acts as a insulator where it's needlessly applied as bare copper would normally radiate heat much more efficiently then if it were "insulated" with a foreign substance. I set about at cleaning things up.




The Ring



After consulting XSPC about this issue I realized once again I'd just about lived up to my reputation as an idiot savant. An item supplied by XSPC intended to raise the mounting plate above any cap's or other devices is supplied with each kit. Unfortunately its ambiguous label had me overlook it. The item is, haunting in its simplicity.



Applying the Ring takes about 1-second, it simply drops onto the water block. Made of aluminum the ring has a lip matching the mounting plate securing it from sliding off the CPU water block surface.



Unfortunately the height of the Ring was just about 2mm shy raising the mounting plate above the Asus Fan less Mosfet heatsink.



XSPC informed me they were willing to ship out another ring to stack upon the first; however, I thought the point was clearly made. While the ring is an ideal solution several should be included with each kit which would truly make the X20's mounting plate "universal."




Intel Tests

Intel Test System
CPU	Pentium 630 Retail (SL7Z9 3.0GHz 2MB L2 1.25V ~ 1.388Vcore) Socket-775
Mainboards	1.) Asus P5AD2-E Premium (BIOS 1005)
Memory	Crucial Tracer Balliztix 5300 (2x512MB DC CL4-4-4-12)
Graphics	1.) AOpen Aeolus 7800GTX-DVD256
Power Supply	PCPower&Cooling TurboCool 850 SSI
Cooling	XSPC X20 Alphacool 12V Cora 662 XP (passive H20-system)
Operating System	Windows XP

Were performed running the Pentium 630 at Default speed and Vcore 15x201FSB=3034MHz / 1.31Vcore then overclocked at 15x270FSB=4073MHz / 1.450Vcore.

Temps were recored at IDLE and LOAD using the CPU/Memory/VGA stress test utility S&M to stress both threads on the 630 creating LOAD. Identical condiitons were recreated in Thermaltake's Kandalf case with 120mm front intake and 90mm rear exhaust.

Ambient room temps = 19°C/66°F. CPU temps were recorded using the Prescott 630 internal thermal diode. Thumbnails below exemplify speeds, voltage and temps.

3034MHz IDLE | 3034MHz LOAD | 4073MHz IDLE | 4073MHz LOAD

Noise measurements



Recently all madshrimps staff collaborated in obtaining similar SPL meters. This is to provide our readers with the most accurate and consistent sound measurements we can. My Intelli Instruments AR-824 Smart Sensor SPL meter is currently being re-calibrated; however, I wanted to include a short section on acoustics. Sound and noise are one of the most misunderstood phenomena primarily because what may be noise to one person may be music to another. What is needed a is a universal standard of measurement, the basis for that standard is the unit known as the dB, defined below from Understanding dB;



When we take into account frequency and human hearing we add what is known as A-Weighting to the db equation the result being the dBA standard of measurement. When frequency and its effects on human hearing are factored in we have a universal definition for "noise" (unwanted sound) a definition which takes into account noise levels damaging to human hearing. I've chosen to omit my results until my meter has been calibrated. I have been testing and re-testing with the new meter and results seem off. I can say this, at 1-meter from the source the case fans in Thermaltake's Kandalf veil any noise made by X20's fan/radiator which was mounted externally for all tests.


Temperature Test Results




I've been running Midwest Cooling Technologies MCT-5 for over 8-months. To eliminate as many variables as possible I ran X20 with its included mixture and distilled water and then with MCT-5. Results found in the chart below represent X20 running with MCT-5.

When comparing MCT-5 to the distilled water/XSPC solution, temps at IDLE were 4°C lower across the board with the distilled water mixture.

When running overclocked, temps at LOAD were 3°C lower on the AMD test system and 6°C lower on the Intel system running MCT-5.

On both the AMD and Intel test-beds, the X20 kit was compared to each manufacturers stock HSF for that particular CPU model. The X20 was then compared to the Alphacool CORA 662 Passive Water Cooling kit. This is also a 10/8mm ID system and one of the better performing passive systems on the market.






Conclusive Thoughts

This review went two weeks past the deadline for one reason; I simply couldn't believe X20 nearly out-performed the Alphacool CORA 642 mated with the NexXxos XP and their best pump. I re-tested several times ensuring ambient temps remained steadfast. I also found that using MCT improved performance under LOAD, making it an ideal alternative for Overclockers.

XSPC has produced one of the best water-cooling kits money can buy. The best news is that the X20 kit is also one of the least costly. To be honest, when I first saw this kit packed in a box only slightly larger than a CPU Retail box, I thought its performance would match its size - small. Not only did this system perform as well as systems costing much more, it was fairly easy to install and quiet. One of the most appealing aspects is that X20 can accommodate AMD Sockets 939, 754, 940 and Intel Socket-T (775), 478, 603, 604 allowing end-users to upgrade without having to re-invest in their H20 system. Size matters in computer-science, such that good things do come in small packages where your PC is concerned. If you've been dipping your toes in the turbulent white-water of H20-cooling choices, I recommend taking the leap into X20's placid water.

PRO
Excellent Performance
Affordable
Masked front panel
Compact, and compact to ship, ergo lower shipping costs to you.
Without question one of the most universal mounting bracket on the market, made even better with the provided "O-Ring."

CON
Possibility of pump detaching from reservoir base.
Extra adjustment ring needs to be included, two should suffice.
Stiff tubing - difficult to fit over fittings.

At the time of writing the X20 high Performance water cooling kit is for sale in Europe/UK for £79.99 at Watercoolingshop.com. XSPC is currently looking for more resellers in UK, Europe and North America drop them a line if you a reseller and interesting in offering your clients a high quality water cooling kit.

Question/Comments: forum thread