All you need to know about your products!
 |
 |
|
3DNews Vendor Reference English Resource - All you need to know about your products! |
|
||||
|
|
|
||||
|
|
|
||||
|
|
|
||||
|
||||||
Water-blocks Danger Den and AquaStone for processors and video cardsDate: 04/10/2005
By: Aleksandr Fomin aka eastSiR
Today we are reviewing four quite interesting products which we got for tests at the very last turn but showed their best among other formerly tested products. Participants of our today's review are two water-blocks for the processor and two for the video card, each by Danger Den (USA) and AquaStone (Russia). We'd better start with the foreign representative of the top-line of processor water-blocks by Danger Den:
TDX A64 Lucite TopDanger Den produces water-blocks perhaps from the very onset of water-based cooling of computers. It's been a long time this North American company hasn't renewed its line of water-blocks - TDX emerged on the market about one and half year ago. That was evolutionary development of the three-union design of RBX water-block which was then turned into a two-union water-block. At the same time, the efficiency and the hydraulic resistance dropped a bit, so the difference between the two products in real systems is almost impossible to catch. It should be noted that use of the two-union TDX is far more convenient since it allows to give up using Y-shaped divides in the structure.  The water-block was packaged in a strong cardboard box which had all the needed to install on the processor. The package has all the necessary data on the water-block and the manufacturer's logo. The fastening system additionally has nuts which hold bolts on the reverse side of the motherboard. They don't have to be used since the water-block can be installed without using them and without removing the motherboard out of the CPU unit. This water-block specimen has 1/2-inch unions of 10.5 mm internal diameter, but the line also offers water-blocks with unions of 3/8-inch (9.5 mm) and 5/8-inch (around 13 mm) in diameter.  As you can see, because of the two-union design implying water feed to the center the water-block looks somehow disproportional. All the manufacturer's products have seals with a label warning the guarantee will be canceled if the seals are broken. This requires explanation since the manufacturer thus guarantees that the product is free from leakages. Even if you break the seals, you will anyway have guarantees against flaws of manufacture.  Sealing is done with rubber washers. There are as many as three - one between the copper base and a cover made of plexiglass, with the remaining two around the unions. On the top of the cover the is a laser-cut engraving with the name of the company, names of the water-block and connector for which the fastener is designed. All the facets and even threading in the covers are absolutely transparent, which imparts some charm to the product. When buying it, you can choose a cover of suitable fastener, every additional cover costs almost half the water-block. That is a popular practice for western brands - our producers normally ship water-blocks with either a universal fastener or with a complete set of fasteners of all sizes.  The base of the water-block was glued with a protective film which prevented the copper from oxidation and growing dim. Upon its removal, the surface acquired a mirror-like gleam. The processing quality of the base was simply superb. We only had to clean it with some spirit to remove the traces of the protective film glue along the whole surface.  The manufacturer provides various relations between the water-block efficiency and the hydraulic resistance it produces. That is achieved through use of removable plates which are installed between the central union and the base with fins. To that end, unscrew the bolts meant for tightening the cover and the base, which cancels the manufacturer's guarantee for lack of leakages. The cover can be easily fastened back without fear to break it due to excessive effort upon tightening. The thing is that the sealing rubber ring is of round section (not square, as is usual) and the bolts have to be tightened right up to the point until the whole surface of the ring touches the cover. That is easy to control visually since the rubber turns dark at the point of touching the cover. We can assume that it is impossible to break the water-block during replacement (and further use) with the plate.  The plates are made available separately from the water-block at about $4. Plate No.1 is base (shown on the photo with black sponge), it is there in the water-block even if you don't buy a whole set of plates. Each plate is numbered from 1 to 5 depending on the resistance to the water flow. Plate ¹1 has the widest slit, the slit of ¹2 is a bit narrower and so on up to ¹5 which has some sort of "nozzles" directing fine jets of water at high speed straight onto the special pits on the base which we'll see below. The black sponge on the plate is needed to avoid water losses so as not to let the water spread aside but be directed at the hottest portions of the base. Each plate is equipped with its own sponge. In the kit we found a plate without any slits - perhaps it was meant to make the user's fantasies come true if the other five are not enough. The difference from using plates is within 1 degree, so in the tests we'll be using plate No.1.  The surface of intensive cooler is indeed impressive. Since the thickness of the water-block base is merely about 1 mm, the fins are of very small height. The whole surface takes up no more than a dime.  The fins are not only wavy in profile - there are also pits on the base. It's just due to them the water-block with plate No.5 with nozzles offers the maximum efficiency. If we look closer at the side guides, we can see special grooves in which the plate is fitted.  The procedure of installation of the water-block on the processor is no less interesting, which for the case of A64 does not have to be removed from the motherboard. That's a really rare and valuable feature, although many don't think of it that way. It didn't take much time to install the water-block - for a start, unscrew one bolt in the nominal fastening frame and instead of it screw in a bolt from the water-block fastening so that the frame behind the motherboard not fall (which is possible because the "bolt" has no head). Then replace the remaining bolt and remove the upper part of the frame. At the manufacturer's web site, there is a lot of materials (including video commercials) on installing various water-blocks, so that should not be an issue to the user.  Installation is over, the hoses are on. In testing the TDX, we used hoses of internal diameter 11 mm. On the photo, you can see that the size of the water-block is a bit less than the heat-spreader on Athlon 64. The fastener has rather rigid springs, so there was nothing to complain about the pressing. We'll discuss the test results later on, but for now we are reviewing the AquaStone water-block which is offered by one of the first modding e-shops in Russia - PCdesign.
AquaStone water-block for a processor with two and three unionsFor tests, we received a three-union model with fittings for a 12-mm hose. Using it, we analyzed the performance of the two-union model, and in that case the center hole for the fitting was closed with a cap.  The water-block did have its own package - it was simply wrapped into a plastic bag together with the fastener and sealed in a way sufficient for the Russian post service. Unfortunately, there isn't much to say regarding the package bundle of the water-block: all it has is shown on the photo. As you see, they forgot to put nuts for pressing the water-block. So we had to invent our own ways.  The cover of the water-block is peppered with fastening holes thus providing universality of the product for use with any platforms. On the cover, there are traces from the milling cutter, which spoils the first favorable impression a bit. The base is nickel-plate, the way it is for Cooled Silence CPU. This coating will not let the copper grow dim with time, which is not a problem, albeit not pleasant. The water-block uses quick-disconnect Camozzi fittings. At the base of the threading, there is a sealing rubber washer. The fittings should be screwed in with caution to prevent leakage. The water-block itself is sealed with a sealant which is not seen anywhere outside, but is there within.  The processing quality of the base is satisfactory. It is smooth but there are traces of processing which are even felt to touch. Since the surface is nickel-plated, you can't polish the base to the ideal condition without spoiling the appearance, otherwise the nickel layer will go off.  The inner surface of the water-block is made up of many "pyramids" which don't touch the cover. That is, between the water-block base and the cover there is a gap, which combined with the use of pyramids promises good results at hydraulic resistance. In the three-union model, water is fed to the center, in the two-union model - to any of the unions since there is no hole in the center. Inside the water-block, due to partitions the water runs in zigzags and washes the whole surface uniformly, which will certainly be good to the efficiency (remember the Cooled Silence CPU modification which resulted in the rise of efficiency).  It is hard to choose a more precise word for the procedure of water-block installation on top of the processor than "torture". Among the causes if that is deficiency of supply, but that is not the only cause. As you can see on the photo, the fastening bolts are low, approximately on par with the height of the fittings. That does good to the appearance, but also brings up problems since the springs stick up above the bolts. Imagine that you have to hold the water-block on the processor with one hand, with a nut in the other hand, press the spring and try screwing in the nut while the fastening bolt is freely turning together with the nut. Imagined? Now add another fastening bolt to the picture and that as you press the spring on one side, the other side is lifting up, which makes the procedure more complicated. And the springs are good, rigid... All in all, I would rather not recommend a water-block for processors with the chip open. The "patient" would rather die from the cure before you manage to push the temperature down. Once the water-block has been installed, you momentarily feel the true happiness as if David defeated Goliath (or vice versa). We have to admit that because of low bolts the water-block on the processor looks better than in the hands, and the pressing is good indeed.  To simulate performance of the two-union model, the center hole was closed with a plug. The Camozzi fittings which we used need some explanation. The water-blocks can be equipped with fittings for a hose of 12 mm and 10 mm in outer diameter, while the hole in them is approximately the same. In the tests, we used hoses of 8 mm inner diameter. If you can use a 10mm/12mm hose, the results may somehow change to the better. Make sure the hose is rigid enough, otherwise the fittings will be simply unable to keep hold of it.
Measuring the hydraulic resistance of water-blocksFor the procedure of measuring the hydraulic resistance, read one of the previous test reviews. We just remind that as the pumps we used powerful Hydor L30 devices, and regular Hydor L20. On the pumps, we removed the suction union. As the injector, we used the bundled union of 9 mm inner diameter for AquaStone, and 13mm for Danger Den TDX.  Hydraulic resistance diagram with Hydor L20 pump  Hydraulic resistance diagram with Hydor L30 pump
The blue-green bars stand for the flow losses because of the hydraulic resistance, other colors stand for the produced flow through the water-block. The quantity if unions on AquaStone water-block is marked as x2 and x3. TDX showed all the advantages of "large-caliber" unions. That is a good margin, which may serve best when using another plate. AquaStone indeed surprised in the case of two-union model. The flow was expected to be less than produced. Perhaps the gap between the cover and the pyramids had a favorable effect, and the use of pyramids instead of pins, with other things being equal, in most ways tends to produce smaller hydraulic resistance. The three-union model took a noticeable lead over the two-union and minimized the major shortcoming.
Test ResultsThe procedure of tests and conditions for all water-blocks are identical. For details, read one of the previous reviews of water-blocks. We ran the tests with the following configuration:
In the BIOS of the motherboard we set the maximum possible voltage, in this case it is +0.35 V. The processor was overclocked to 2700 MHz.  Results diagram The blue color stands for the data on Hydor L20 pump, red - for Hydor L30. The test results proved somehow a surprise. It turned out that all the three participants of today's review took first places, and while it was expectable of our North American guest from the very start, then AquaStone with two unions proved a pleasant surprise (especially with Hydor L20). Such success was due to the fact that water uniformly washes the whole surface of the water-block, which does give a minor performance boost sufficient to become one of the three leaders. Nevertheless, this model offers higher hydraulic resistance and the results will definitely be worse in a real system. That can be traced from the difference in results produced for various pumps - it is greater than the average versus the rivals. The three-union model is free from this shortcoming, which makes it attractive but causes other problems typical of such a design. First of all, it imposes some restrictions upon the components used (application of dividers, serial connections, expansion vessels with two inputs and one output etc.). In should be kept in mind that Danger Den TDX was tested with plate ¹1 and offers small margin for improving its efficiency (and decrease of the hydraulic resistance). Danger Den TDX is an excellent water-block of superb efficiency and exterior - that's about all what can be said about it, so we'd better move on to reviewing water-blocks for video cards.
Danger Den Maze4 GPU Acetal TopThe name of the model suggest almost all of its key features, namely, the inner structure specific of the previous line of processor water-blocks, and the upper part made of plastic, durable, strong and amenable for complex processing. The line also has other "upper parts", e.g. chromium-plated, so the buyer can choose not only by taste but color as well. On the cover, there is an engraving of the company logo made as double D. 
The water-block, like the one aimed at processors, arrived in a separate package containing all the basic information about the product. There is also a notice of compatibility (to all modern video cards by ATI and NVIDIA), as well as a note saying that the fastener is sold separately. It happened that there was no fastener in the parcel, so those who want to get an idea of its look can find on the company's official web site. Again, we had to invent our own ways. It should be noted that it is juts the fasteners which provide compatibility of the water-block to a particular video cards, and the cost is sometimes as high as 10$. 
The base of the water-block was glued with a protective film, and the unglued surface turned slightly dim. On the base, there is a small ledge above the VPU core. It is smooth and well processed to the mirror gleam. On the water-block there are also seals providing the manufacturer's guarantee against leakages, which may be canceled if you dismantle the product.  The surface is styled for "labyrinth" which the name suggests. The copper surface looks a bit untidy because of some local dim points, unlike TDX, but that should not affect the efficiency. Sealing is done with a rubber ring without any sealants.  The cover appears to be two tanks with a very thin partition right on top of the water-block center. Quite an interesting solution which is aimed at reducing the hydraulic resistance of the water-block (thus, the whole structure) at the expense of efficiency a bit. Albeit it gives saving of the precious flow in the line, this approach will let concentrate efforts where it is of most need - on the processor. Video cards have a temperature mode of somehow different range than that of processors, their overall performance depends also on the memory bandwidth, so water-blocks on video cards do not need record-breaking efficiency to the prejudice of hydraulic resistance. On the photo, you can see a number of holes, and only four of them are used to tighten the base and the cover. Fasteners on NVIDIA video cards can be attached on both the four holes around the chip and the two furthermost holes over the diagonal. We decided to try the latter way.  The fastener is tightened in quite an unusual way as compared to the domestically manufactured products, all the manipulations have to be done from the reverse side of the board. On the photo you can see standard fastening holes for NVIDIA cards look through the four holes around the chip. The "native" fastener implies using a cruciate stiffener with a soft porous washer to protect the onboard components.  The water-block is easiest to install if turned upside down under the board, which allows minimizing the risk of core chipping. Its side is even greater than in TDX, but when assembled it takes up just one more PCI connector.
AquaStone water-block for a video cardAquaStone water-block for a video card arrived wrapped in the same plastic bag as that for AquaStone CPU. The fittings were not installed - they were put in a separate package. The water-block in size is smaller that the fittings, so it is a bit uncomfortable to wrap them.  The whole package bundle was there in place. There are few parts in the fastening system, merely two for each: dielectric washers, springs, bolts, and wing nuts. All is intuitive, so the user can hardly lose way even though the package bundle provides no setup guide. The water-block follows the general style of the line and copies all the main traits of the model aimed at processors: Camozzi fittings, pyramidal design of the surface, nickel-plating, sealing with silicon sealant, a universal cover "for all" (including some chipsets of motherboards).  The surface of the base is polished poorly. There are not only grooves after processing which are felt even to touch, but their depth leaves much to be desired. The water-block is unlikely to lose the precious degrees, but the buyer's mood over the lucky buy may seriously get down.  Because of the large size of fittings, the holes are placed apart, closer to the edges. The photo shows traces of sealing, as well as the partition which forces water to wash the whole surface uniformly.  The installation didn't take much time at all, and in the end the water-block was positioned a bit turned relative to the video chip. The springs provided adequate pressing, but the installation would have been easier if there had been washers for springs or holes in the cover instead of slits. The former is much easier to implement - in this case, the springs do not fall within the slits and stop against the cover and the wind nut with sharp ends, thus impeding its tightening.  When assembled with the AquaStone water-block, the GPU will take up two PCI connectors, but it is more convenient to lay hoses inside the housing due to rotating Camozzi fittings.  To run tests, we assembled a separate water-based cooling system of relatively lower performance. Presumably, it displays the real temperature mode of a video cards in a high-efficient water-based cooling system at three points. The elements of the water-based cooling system were the Hydor L20 II pump with 9 mm unions (ID), a radiator from ProModz for a 120 mm fan (standard Camozzi fittings were replaced with compression fittings because of greater compatibility, a pass hole of 8 mm diameter). The fan was powered for 5V. A hose of 11 mm diameter was used. In the case with AquaStone GPU, hoses from the pump and radiator to the water-block were replaced with 8mm (ID). During the tests, the ambient temperature was about 24 C. The reference data on performance used in the table were produced with the nominal cooling system ASUS EN6600/TD/128M/A based on NVIDIA GeForce 6600 chip with the temperature monitoring activated. The video card was overclocked from 300/550 MHz to 540/678 MHz. Use of water-based cooling brought additional 10 MHz for the chip. Perhaps we attained the limit capabilities of a particular specimen, so improving the cooling conditions doesn't help. It should be noted that nominal air-based cooling system of a video card is very noisy, whereas water-based cooling system did not produce any noise. The tests implied running a number of 3DMark2005 cycles with default settings until the temperature is stabilized. The temperature was controlled by RivaTuner v.2.0 RC15.6, but unfortunately it rounded the readings to integers.
As you can see, there isn't much of a difference between the temperature mode of the video cards with various water-blocks. On the other hand, not only we produced a noiseless system with nominal air-based cooling (at quite a lot, I must admit), we also reduced the chip temperature by 17 C in the idle mode and by 38 C under load. As regards the choice of a specific water-block for the video chip, we would advise to choose on the base of other parameters rather than pure efficiency (e.g. the number of blocked PCI connectors, cost, compatibility, cooling of not only the video chip, but memory etc.).
If we weigh all the pros and cons after tests, today's participants will be definitely in the pros part. Of the cons is deficiency of fastening in AquaStone CPU and Maze4 GPU, poor processing quality of the base in the case with AquaStone GPU, but that anyway applies to specific specimens. Here is the overall summary: Danger Den TDX A64 Lucite Top
Danger Den Maze4 GPU Acetal Top
The "fly in the ointment" is in that the produce of Danger Den can be still made available abroad only, which imposes certain temporal restrictions and adds more to the price. Recommended price for TDX A64 amounts to about 52$ in the basic package bundle (up to your choice). Additional cover, e.g. one to install on LGA (or on A64), will cost extra 20$. A kit of plates costs another 4$. Additional fastener for Maze4 GPU will cost about 10$,and the water-block with a faster to choice will be about 40$. All the should be ordered from abroad either at once or wait it to arrive and pay for the delivery. Many might not like it, but the last word is always after the buyer.
No doubt, the produce of Danger Den deserves most compliments and best recommendations.
 AquaStone CPU
AquaStone GPU
In addition to the above, we should note that AquaStone water-blocks stand out with their common style and some harmony. That is, two or more water blocks in the CPU unit look better than one. The price of the products is moderate and meets the declared. A water-block for the video card is about 1000 rubles (~ $35), one for the processor varies within 1150 to 1300 rubles depending on the number of unions. These water-blocks can be easily made available via e-shop following a well-established scheme. The author is highly appreciated to Dan Stephens from Danger Den, administration of PCdesign, e-shop, as well as to Aleksandr Pronoza for the specimen water-blocks presented for tests.
|
|
|
| |||||