Corsair Nautilus-500



Over the last few years the water-cooling industry has been experiencing rapid growth in the OEM sector. Large companies have been seeking proprietary systems to hang their logo on and various component makers have expended their production to supply these orders en masse. Corsair entered the H20 cooling fray over three years ago with the introduction of the Hydrocool 200 external unit. [M]adshrimps reviewed Hydrocool 200 shortly after its release. For their second foray into H20 Corsair "came in from the Ambient" introducing an internal water-cooling kit Corsair COOL. I had the opportunity to review Cool finding the water-kit offered great flexibility as the radiator/fan assembly could be mounted internally or externally and in multiple locations. Corsair's return to external H20-cooling is a promising package, ergonomically streamlined at 1/4th the size of Hydrocool and boasting over twice the performance potential, the Nautilus-500 should appeal to the H20-Novice for its simplicity and to H20-Enthusiasts for its cooling prowess.

Although water-cooling out-performs air-cooling in just about every application, the single greatest obstacle is the additional production cost to the manufacturer. For a particular company to design and build an all inclusive H20-system, various components must be out-sourced, assembled in-house (or at another location) and then "packaged" in a design which is unique to the company selling it. Not always an easy task since there are a limited number of pump, radiator, reservoir and water block makers. Furthermore where cost is a prohibitive factor, by definition it must be passed on to the consumer. To date the desire for H20 cooling (industry wide) has not outweighed its price tag; however, we are rapidly approaching a divergence where cost is beginning to drop and commensurate with H20's superior performance. Fortunately the number of fabrication facilities specializing in water-cooling components has grown dramatically lowering production costs while providing more options for manufacturers and consumers alike.



Miniaturization or scaling is the driving force in the world of microprocessors; CPUs in particular continue to decrease in size while clock-speeds and processing power increase concurrently. Ideologically the goal behind each die shrink is to pack more transistors into less space, transistors which switch faster and consume less power. Logically lower power consumption should equate to less heat, unfortunately the nature of processor architecture is antithetical in this respect. With more transistors occupying less semiconductor real-estate problems such as capacitance and gate leakage raise thermal output considerably. Evidence of this can be seen simply by comparing the size of CPU heatsinks over the years. Air-cooled heatsinks have continually grown upwards and outwards into massive "sky-scraper" sized structures (plumbing included via heat-pipes) as in Thermalright's HR-01 at 518gr, or the Zalman Fatality 1 at 918gr respectively! Water-cooling is not only ideal due to H20's specific heat capacity, the amount of copper required to build a water block is miniscule compared to the required surface area of an air-cooler. Ultimately a well designed water block need not be any larger in it's base plate (contact) surface area then a processor's IHS (Integrated Heat Spreader) and in this respect the water block acts as a "point source cooler." Looking at the photo below, which would you rather have "strapped" to and hanging off your motherboard?



Specialization has always been a concern for manufacturers investing in H20-cooling since most components were borrowed from other industries, then had to be modified. The only components originally designed and constructed for PC/Notebook water-cooling were water blocks and reservoirs. Pumps originally derived from the Aquarium and Fountain industries, radiators were originally salvaged automobile heater-cores. Both areas have seen progress. Insofar as DC-pumps are concerned specialization and miniaturization continue to produce better products from LAING, maker of the DDC (Delphi DC) DC-pump and C-Systems maker of the CSP-MAG DC-pump. Seen below the CSP is about as svelte as they come yet offers respectable pressure and flow-rate.


While pumps, water blocks and reservoirs can be miniaturized the radiator is another matter. In this respect the laws of physics are unforgiving since even water-cooling ultimately uses air to remove heat from the system via convection (passive) or forced convection (active (fan)). This is where the external water-cooling system has an innate advantage over internal options. The problems for internal H20-kits are radiator/fan placement which ideally can only be installed into one of two locations; replacing the front intake fan, or the rear exhaust fan. In either scenario case aerodynamics are disrupted and the radiator’s location inside the case places it in proximity to heat from other components. Components which are now having a deleterious effect on case temps since the aerodynamic flow aiding in cooling has been reduced.

The External H20 Solution



The Nautilus-500 is well packaged in a box size belying its 500W cooling capacity. Opening the box revealed a neatly packed system with an in-depth manual.



Unpacking the box, Corsair has provided just about everything you need with the exception of distilled water. I prefer Poland Spring Steam Distilled H20 however the choice is yours. Corsair COOL liquid additive for H20-cooling systems is provided and while I've been an advocate of pre-mixed "cooling fluids" for their "non-conductive" properties, at water-cooling temps between 18C to 45C distilled water with propylene glycol type formulas provides the lowest temps. I was impressed Corsair had thoroughly researched their choice of additive, the formula is essentially Propylene Glycol which acts as a surfactant / lubricant helping to prevent dissimilar metal corrosion an reduce laminar surface flow. Unlike Water Wetter which I stopped using over two years ago due to residue build up, what Corsair has left out of their formula are ingredients recently discovered in Water Wetter which have cast a new light on that product.



Perhaps foremost on the mind of the PC-Enthusiast would be; What CPU Water block is Corsair using? The unit is most likely manufactured by or related to the Swiftech MCW6000 series. Although dropped from Swiftech's product line, purportedly these blocks were and still are manufactured specifically for use in Corsair COOL water-cooling kits and now the Nautilus 500. Nonetheless the base of this copper block sports one of the best finishes in the market as can be seen in the photo below in which I purposely left the protective plastic in place.



With the protective cover removed the block's finish is extraordinary, a benefit of Corsair's high standards which of course have made their memory one of, if not the best DRAM on the market.



Standing on its own the primary heat exchanger including pump, radiator and reservoir is an attractive design. Certainly a space saver it should easily fit on top of any case, or on a desktop, although I would recommend leaving the unit is mounted higher then the CPU itself. The fan is located on the top drawing air from beneath the unit which is raised a few cm to insure the radiator isn't obstructed. The design is so simple as to be a potential award winner.



Viewing the unit from the rear we see the High/Low fan speed switch and Quick Disconnect nozzles. These nozzles made installation, and especially bleeding so easy it's no wonder the entire industry doesn?t employ them more often? Given the Retail price for the Nautilus one tends to doubt cost is a limiting factor since only two quick releases would be needed if placed strategically. The DC power connector is also located on the rear of the unit attaching to a power cable of ample length. This cable then feeds to a PCI-mount which has inlets to accommodate both tubes and the power-cable. At the end of the cable feeding into the case are male/female molex plugs.



With a top-mounted fan air will be drawn from the bottom, as was evident above four cushioned legs raise the unit up allowing air to flow freely up into the radiator. Turning the unit over we get an idea of the radiator size as well as its aluminum fin pattern.



A look inside Corsair's Nautilus 500 ->

System Mechanics



Disassembling the Nautilus 500 reveals Corsair may have used a radiator identical to that found in their original Hydrocool, as seen in these Tweakup.DK Hydrocool photos (scroll down). This is by no means a criticism since this radiator performs on par with many of today?s 120mm single radiator separates. Inside Nautilus the radiator is located on the base above a grill allowing ambient air to be drawn in via a fan mounted directly above. Corsair chose a fan from Vette Corp specifically the 120x25mm model A1225L12D which runs at 1800RPM producing 74.4CFM. While not as powerful as I might like the unit is certainly quiet and does do the job. Insoafr as the radiator the design follows today's higher end models which eschew atypical tubes in place of flat ellipses which force more water molecules along its interior surface. Two radiators pictured below exemplify the shape of the water "tubes" used in radiators, on the left the Alphacool NexXxos Pro II on the right the Thermochill PA 160.



Corsair's choice of radiator and those pictured above share many similarities as well as differences. An entire article can be written on radiator design when you consider the number of variables alone such as pressure and flow rates across pumps feeding the radiator, metal type, wall thickness. What seems to work well are the elliptically shaped "tubes" which can be seen in the close-up below. When I thought of a radiator I thought of rounded tubes snaking through a series of thin cooling fins, where the tubes would have to "exchange" thermal energy via the fins. From the macro below the ratio of fins versus "tubes" seems to be juxtaposed in a much more efficient design. Where the majority of surface area was once cooling fins it's now water filled "ellipses" or foils these are very narrow in width yet relatively tall. This begs the question how will laminar flow be affected and how will this ultimately affect heat dissipation?



If you have any water cooling experience, more then likely you've come across the term Laminar Flow while reading a review or in the forums. You can always tell when "flow rate" is being discussed because the thread is usually 10+ pages in length and includes multiple heated arguments, as found here at OCForums. To put it simply Laminar Flow describes how a viscous fluid "behaves" as it flows through a tube or over a surface. For our purposes as a stream of H20 flows through a length of tube with a fixed radius the stream divides into layers or Lamina which flow at varying speeds beginning at a rate of 0 at the surface progressively increasing in speed until the center layer, where maximum flow-rate is achieved. Water molecules in contact with the surface area of the tube "adhere" to it based on a set of criteria one being surface friction. Laminar flow is explained below.



Laminar Flow plays a critical role in thermodynamic processes where H20 is used as a transfer medium in PC-cooling. Tube diameter, length, fluid viscosity, H20's specific heat capacity and of course thermal conductivity of the radiator material will all play an integral role in the cooling process. For example a small increase in tube radius will have a powerful impact on flow as explained by Pouiseuilles Law, seen below.



Back to less technical matter's, Corsair has done an excellent job organizing an entire water cooling system into a small container. From the front view the radiator consumes most space while the reservoir and an inverted LAING Thermotech pump consumes the remainder. Electronics have been kept simple, DC-power enters where the wires are split to power the pump and twin speed fan which has its switch mounted on the rear panel. While a temp display or water level indicator would have been nice their absence reduces the price and simplifies installation.



Rotating the unit 180-degrees provides a better understanding of water-flow direction. Heated water returning from the CPU water-block enters through the Quick Disconnect nozzle on the right where it feeds into the reservoir. Although difficult to see from this angle at the bottom of the reservoir an outlet feeds the inlet of the LAING Delphi DDC mounted top-side up. The Delphi or DDC as it's more commonly known then feeds the radiator inlet.



Taking a closer look at the LAING DDC pump we see Corsair organized the system so this pump had to mounted upside down. While this doesn?t seem to affect flow, I did immediately wonder how it might affect the pump long-term? I'm sure Corsair doesn't intend to RMA thousands of units over something as silly as improperly mounting the pump, so I wrote LAING.



Below we see the Delphi DDC model number and serial number. Laing Thermotech is one of the few companies to print so much pertinent information on their pumps, and given their wide use across the industry this is a smart move. This took a little "peeling" back of the sticky pads used to mount the pump although it is snug no worries.



To exemplify how Corsair's choice of flow direction through the cooling systems I've inserted color-coded arrows onto the photo below. The red arrows indicate water returning from the water block which is heated from the CPU, or more accurately carrying the kinetic energy from the processor conducted to it through the water block. Blue arrows indicate water which has transferred that energy (heat) dissipating it through the radiator. This returns to the processor in a continual cycle.



Beyond the specialization of the radiator itself, I was happy to see Corsair organize the cooling cycle such that the water leaving Nautilus and entering the CPU water block has just been passed through the radiator and will be at it's lowest temperature in the loop.

Installing the Nautilus 500 water block.

The water block itself sports an exceptional finish. When out-sourcing a specialized component such as a CPU water block, there are varying degrees of base-plate finish a manufacturer can choose from. All manufactures seek flatness this cannot be compromised (relatively speaking); however, finish is a result which can be specified based on cost. When ordering many thousands of parts (water blocks) a manufacturer may want to keep costs low so as not to pass them onto the consumer. Ordering thousands of water block base-plates with a mirror fish would be a costly endeavor and in this respect Corsair did not compromise or pass the cost onto the consumer, quite a feat.



In our next photo below the block was mounted and tested over a short period using an Arctic Silver rapid setting thermal paste. When applying thermal paste I use the same process for every heatsink install, centering a small dollop of paste on the processor IHS (Integrated Heat Spreader) then mount the water block allowing that pressure to ingrain and disperse the paste. To reiterate, this is preferable to "spreading" paste with a credit card etc. which incorporates air into the paste as it drags the material over striations and micro-pores in the base. Removing the block to examine the "TIM-spread pattern" reveals how effectively included mounting hardware distributes pressure across the water block. We're not simply looking at the diameter of the pattern, it's also important to look closely at build up, for example has the paste been "forced" into the micro-pores and or striations. From the photo below is it evident what began as a droplet sized amount of thermal paste has been dispersed about as well as I've seen.



Installation isn't exactly an appropriate term for the Nautilus 500, for the simple reason you?re not spending any length of time inside your case. With many kits you must remove your motherboard to gain access to the reverse side of the board to secure the mounting hardware. While this was once he preferred method many manufacturers use the mounting bracket which is just as secure since they use back plates themselves. Preparing to install the water block for Socket-939 requires items seen below outlined in AMD orange. I must stress even with the LGA-775 hardware, you do not have to remove your motherboard to install the Nautilus 500's CPU-water block.



Following Corsair's directions one would mount the water block and then fill the system. More experienced users tend to err on the side of caution, preferring to leak-test any system which introduces water into their aluminum box filled with expensive electronics. I chose to go this route and delay mounting the water block to the DFI board.



Insofar as Leak-testing this simply entails filling, bleeding and running the system for a extended period of time on a towel or other absorbent material outside of the case. I chose to leak-test the Nautilus for about 45-minutes. Corsair certainly isn't taking risks with their customer's hardware which is why hoses attached to the waterblock are clamped securely at the factory. They have been repeating this method since the introduction of Corsair COOL. If a leak were to occur which would damage costly components that leak would most likely occur at the water block connectors.

Mounting the Water block



Mounting the block involves a few simple steps; placing the foam cut-out over the block, placing an adhesive backed metal plate on that foam, and slipping the brace over the tab.



The next step consists of slipping the shorter brace with screw thread over the opposite tab, placing pressure on the elongated brace and aligning the screw holes. Adding more pressure to bring the braces together and threading the down the mounting screw.



The most difficult aspect installing the water block involved laying the retention brace over the foam, slipping the second brace over the tab and under the elongated brace. This wasn't very easy since the foam had never been compressed before. My only concerns would be in time would the foam loose its elasticity.



Onto testing ->

Testing the Nautilus 500



Testing Nautilus 500's cooling ability entailed overclocking and stressing the CPU using the tried and true CPU/subsystem-stress test software S&M. Processor temperatures were measured placing an external thermistor at the edge of the processor's IHS between the CPU and water block. Primary temps will be measured using the processor's internal thermal diodes. Self appointed PRO-cooling "experts" will frown upon this, this usually involves a temper tantrums and labeling some a "Buffoon." I understand the pathology as pseudo-intellectuals often have too much knowledge in one area lacking any real life experience they become myopic. This may be a result of post traumatic stress disorder from the socket-thermistor days and/or troubled childhood. Sociopath diagnostics aside since CPU makers have been integrating thermal diodes into their processors they have an innate advantage simply by virtue of proximity. While its true software interpreting these devices may be inaccurate, they are at least consistent, therefore temp fluctuations will provide useful data.

AMD Test System
CPU	Retail Opteron 148 (CAB2E) Socket-939
MAINBOARD	DFI LAN party UT nF4 Ultra-D
MEMORY	Corsair TwinX2048-4400PRO (2x1024MB DC CL3-4-4-8)
GRAPHIC CARD	AOpen Aeolus 7800GTX-DVD256
POWER SUPPLY	XG Duro 900 PCPower&Cooling TurboCool 850 SSI
COOLING	Corsair Nautilus 500 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

Overcocking our Opteron 148 involved the following settings:

9 x 275FSB = 2475MHz @ 1.34Vcore | 10 x 285FSB = 2850MHz @ 1.53Vcore

Temps will be recorded at LOAD with the fan speed at Low then High. The Thumbnails below record data entered into our chart below.

2475MHz Low | 2475MHz High | 2850MHz Low | 2850MHz High


As the thumbnails above indicate (and the chart below will clarify) there was very little difference in temps between High and Low fan settings. This is most likely attributable to Opteron?s thermal output while overclocked. Temp differential is even less prevalent at 285FSB (2850MHz) even with the additional Vcore and this is due to the cooling system beginning to reach a state of thermal equilibrium.

There was also very little difference in noise level between the Low and High settings. I measured this using a Intelli Instruments AR824 SPL meter from 50cm. With the fan on Low noise was measured at 6dBA above ambient (for my room) and 10dBA above ambient wit the fan set to High. Corsair has chosen a fan from Vette Corp. whom make numerous OEM products in addition to a large number of DC-fans. For the Nautilus 500 radiator fan, Corsair chose the VETTE model A1225L12D (PDF) this is a 12VDC, 1800RPM, 74.4CFM.

As I stated above the temp differential while overclocking was minimal between High and Low fan settings and i believe this is due to the system reaching its maximum performance range. If there was one area which I felt limited Nautilus it was its fan. This inspired me to replace the VETTE fan (74.4CFM) with a Sunon KD1212PTB1 which moves 90CFM. Given the Laing Delphi is basically a low flow pump then increasing the air flow certainly couldn't hurt anything. One reason I delved into laminar flow in the opening section was to exemplify here in the "real world" there are always variables which arise when mating H20-components which make universal rules such as "higher flow rate = lower temps" susceptible. Unfortunately the fan Corsair has mounted requires a Torx wrench to remove, therefore I was forced to remove the entire top half or cover and simply lay the Sunon fan on the radiator allowing it to run flat out as I bypassed the switch. This ad hoc method introduced a number of variables once the cover was removed.



Thumbnails below exemplify results running the Sunon fan. In so far as noise levels running the Sunon, from 50cm from the case/unit measured 19dBA over ambient; however, with the fan improperly mounted this reading isn't representative as to what the fan might sound like if properly mounted, that is secured and covered. This was an ad hoc experiment since I didn't have a Torx head driver to remove the Vette fan.

9x275FSB=2475MHz Sunon High | 10x285FSB=2850MHz Sunon High


Summary of Results

The chart below summarizes all results including the ad hoc experiment where the Sunon was used. Temps are recorded in Celsius, comparing performance between the Nautilus 500 and the Stock AMD air-cooler provided with our Retail version Opteron 148.



PRO
Great overclocking performance.
Very easy to setup.
No need to remove motherboard.
Ergonomics, svelte unit, streamlined, LED, good looks.
Silent operation on High or Low fan speeds.
PRICE!

CON
Foam retention mounting system.
A more powerful fan with rheostat would be ideal.

Price at time of writing. The Nautilus 500 can be found through Froogle at around $/? 150.

Conclusion

Corsair has done an outstanding job bringing the Nautilus 500 external H20-kit to market for the SRP of $150. There are a growing number of inexpensive compact internal/external water cooling kits beginning to hit the market; however, they all seem to fall short in one area, their pump. Most are spec'd with underpowered, unreliable submersible designs which leave the system wanting. The Laing Thermotech Delphi DDC offers a respectable balance of pressure and flow-rate, not to mention the DDC is found in more water cooling kits then any other pump today. If I were asked what the secret ingredient is in the Nautilus 500's successful recipe, I would have to say it's the radiator. Corsair's decision to use a previous model from the Hydrocool series not only contributes to its efficient cooling, but most likely the low price as well. I have just one concern and one "wish" for the Nautilus 500. My concern is the foam insert acting as the retention mechanism for the mounting system. Perhaps a more traditional mounting system might be in order as I fear the "sponge" will eventually loose its rebound. Although I'm sure it would take some time.

My "wish" would be a more powerful fan mated to a rheostat. Of course keeping the cost down for the consumer was key for Corsair and to this end small concessions must be made, none of which negated performance. The charts above speak loud and clear as to this external water-cooling kit's ability. When you consider you can be water cooling your CPU in just about the same time you can install a high performance air-cooler, many of which require removal of the motherboard while Nautilus does not, I wouldn't hesitate to recommend this system to anyone interested in virtual maintenance free H20-cooling.

In the very near future I plan on re-visiting Nautilus 500 replacing the fan with a much more powerful model and switching out the water block with a model more conducive to low-flow or the Laing Thermotech in particular. Stay tuned for some real fun as this unit is highly mod-able!

Questions/comments: forum thread