Date: May 4, 2006
Author(s): Matthew Harris
In computing we have many cooling options. Air, water and phase change being the most common in the enthusiast arena. Today we will be looking at a top grade kit from Swiftech and comparing it to a similarly configured kit from the competition and see if there is a clear winner or loser.
As today’s computers are getting faster we are starting to see a wider variety of cooling options hitting the store shelves. There are three basic types that enthusiasts normally encounter and those are air which is the cheapest and easiest to configure. Water which adds a degree of difficulty but is far more efficient at removing heat and phase change which is the most expensive and the most difficult to configure. Today we’re focusing on water.
Water cooling is gaining in popularity and we see new kits hitting the scene on a nearly daily basis. The kits are varying in cost and difficulty from the fairly cheap small self-contained kits by the likes of Evercool, Coolermaster, Thermaltake and Kingwin to name a few and going to the more elaborate fully DIY kits by such names as Aqua Computer, Innovatek, Swiftech, Danger Den and Asetek. I know my lists aren’t nearly a complete listing of the major players but it does sample several known players in the low dollar and high-end water cooling niches.
Many DIY makers also offer pre-configured kits that are ready to install in your PC. These usually consist of plumbing, pump, radiator, fans, clamps, mounting hardware and normally a CPU water block. Today I’m looking at the Swiftech 220-H2O Apex Ultra kit and the logical add on of 2 MCW60 GPU blocks for an SLI/Crossfire system.
From Swiftech’s site:
Features and benefits
Capturing the heat
The extreme performance APOGEE Universal water-block is responsible for absorbing the heat generated by today’s hottest CPU’s with incredible efficiency.
Adjustable fan speed: the kit is delivered with 12V to 7V and 12V to 5V fan adapters, enabling users to reduce the fan speed -and noise, to whisper quiet operations (~25 dBA).
Moving the Heat around
The Extreme performance MCP655™ 12 Volts DC industrial pump pushes the coolant throughout the circuit at high velocity.
Larger 7/16" ID (5/8" OD) tubing is used to maximize the coolant flow rate inside the system.
Adjustable Pump speed: the pump speed can be adjusted to virtually silent operation.
Dissipating the heat into the atmosphere
Minimum toll on performance in "silent" mode
The new MCR-220 copper and brass radiator doubles the surface area available for heat convection into the air compared to a single 120mm radiator, and uses two 72 CFM 120mm fans, at a reasonable* sound noise level of 37 dBA.
The MCB120 "Radbox" Revision 2 optimizes cooling by using fresh air from outside the computer, without adding any thermal load to other components.
The MCR-220 radiator fin density has been optimized to reduce air noise and operate at high efficiency with low-flow fans
At the lowest fan speed settings, the H20-APEX kit still delivers superb performance while allowing substantial overclocking. For example, at 100 Watts, the CPU temperature is only 3°c higher than at the highest fan settings
Convenience and reliability features
Kit noise/performance ratio can be fine-tuned at the user leisure.
12 Volt DC MCP655 pump plugs directly into computer power supply.
*For reference, stock cooling solutions for current high-end CPU’s are specified by manufacturers between 34 and 38 dBA.
As you can see the kit features a lot of extras that can be very helpful when adding a water cooling kit to an existing PC. The tubing is a nice compromise between overall flow and lower bend radii. The Radbox comes in handy for users that don’t have the room inside a case to internally mount a 2 X 120mm radiator (Or for those users that don’t like the idea of cutting up a new high dollar case) and it offers better performance in most cases by getting the radiator out in the ambient room air rather than keeping it inside the case where in most cases (No pun intended) the rad will be acting as an exhaust.
I mention both of these things because they’re not typically in kits. Most kits offer either 5/16" or 3/8" or 1/2" I. D. tubing and none offer the radbox as it’s exclusive to Swiftech. Another addition that’s pretty much exclusive to the Apex kit is SmartCoils. SmartCoils are coiled plastic that resembles springs. In use you wind it around tubing that has a very tight bend radius to prevent that tubing from collapsing. I mention these because I’ve seen people talking about the price of this kit being higher than others but when you offset the fact that you’re getting extras for the price difference it still works out as added value.
After opening the box we see that Swiftech thoughtfully includes a rather sizable instruction manual to help the neophyte user with setting up his or her new water cooling system. The manual weighs in at a hefty 22 pages and includes everything from mounting the Radbox to configuring and bleeding your loop. Not only does the manual cover the basic set up the manual also covers additional water blocks for single and multi-CPU and single and multi-GPU setups. The manual covers configuration of every separate component of the water cooling system along with the configuration of the entire water cooling system as a whole. We also see the Swiftech thoughtfully includes a small funnel and a bottle of HydrX coolant additive.
In the included hardware we find a 5 V four pin to three pin adapter 12 V to 7 V resistive three pin adapters along with a PCI shield that is cut out to allow the fan wires to pass through for rear radiator mounting. Swiftech was also thoughtful enough to include grommets not only for passing the water tubing through but also a grommet for the PCI cut out so that the fan wires don’t chafe and eventually end up shorting out. Unfortunately despite what it appears you cannot pass the tubing through the PCI cut outs. Instead you have to cut a 7/8" hole in the rear of your case with a hole saw (not included) and pass the tubing through those holes.
Looking at the working parts of the Swiftech Apex Ultra kit I’m reminded that Swiftech not only cares about how well the product works but also about how well it looks. Both the Apogee and the MCW60 share of the diamond pin-fin configuration. In fact looking at the base of the MCW60 it appears that the MCW60 and the Apogee share the same base. It looks like the MCW60 just has the outer edges of the base machined down to allow the MCW60 to work as a GPU block. Personally, I feel that this is a good approach given that this allows the GPU block to exchange more heat.
As you can tell by the pictures the finish on the bottom of the blocks, while not shiny, are both very smooth. Just out of curiosity I took a straight edge to both the GPU and the CPU water blocks and both were extremely flat. If you look at the bottom of the CPU block you’ll notice that there are machine lapping marks. While the marks show up in a photograph you don’t feel them with your thumbnail whenever you run it across the surface. At least I didn’t.
After you’ve decided how and where you’re going to locate your components you have to mock up the system and fill and bleed the loop. Normally the best rule of thumb is to keep the reservoir as close to the top of the loop as possible if not at the very top. The reason for this is that during filling and bleeding of your loop it prevents gravity from playing havock with the fluid in your loop. If your reservoir is at the bottom of the loop what will happen is whenever you’re trying to fill the loop you’ll end up only being able to get a minor amount of fluid into the loop because gravity will be holding the fluid in the reservoir rather than allowing it to flow out into the loop.
If your reservoir is at or near the top of the loop when you fill the loop you can tilt the case to get the fluid to flow into the blocks, pump and radiator. Swiftech does not include that bit of wisdom in their instruction sheet. Since I’m bringing up the instruction booklet I should point out that Swiftech’s instruction manual is in large part a compendium of the individual parts instruction sheets. I have to admit that at times I found it rather confusing that the instruction manual Referencing 3/8" fittings as well as 1/2" fittings when there were no 3/8" fittings included. Now the MCW60 does include 3/8" fittings but since the tubing is 7/16" the 3/8" are just a trifle too small.
The two pictures of the radiators are to illustrate the differences between the design of each. The reason that I’m illustrating the differences between these two radiators is that I’m going to be comparing the Swiftech Apex Ultra kit against a Danger Den kit consisting of the Black Ice Extreme II, TDX, Maze 4 X 2 and D. D. D5 (Same as the Swiftech MCP655). The Danger Den kit uses 1/2" tubing rather than 7/16". Getting back to the differences in the radiators, the Black Ice Extreme II is a double row radiator design.
It also has a higher fin per inch density than the MCR220 which is what is present in the Swiftech kit. The MCR220 is a single row radiator and has been tuned with a lower density of fins per inch to take better advantage of medium to low flow fans. What this means is you will get better heat dispersion with lower flowing lower noise fans. Going by Swiftech’s performance data, if you do choose to go with a higher flowing fan you will see better thermal dispersion at the sacrifice of added noise. What this all means is that at a given CFM the Swiftech radiator will outperform the Black Ice radiator due to its lower internal losses.
Moving right along, you can see that the radiator is mounted upside down in the front of the case. You can also see that I used the Smartcoils on the output of the pump going to the radiator. While I’m thinking about it, getting the 7/16" tubing to stretch over the output and input for the pump was rather challenging. I found that I had to submerse the tubing in near-boiling water. I also had to put dish detergent on the fittings. For the rest of the tapered fittings, although they were 1/2", I found that smearing a little dish detergent on them and simply wetting the tubing did the trick just fine. The trick is not to use so much dish detergent so that you end up with your reservoir filled with foam like I did.
Here are a couple of pictures of the MCW60 water block mounted on a 7800 GTX. You’ll note that I did not use the included RAM sinks because I was testing this on a pre-existing system that already featured RAM sinks. The MCW60 attaches to the card with screws going through springs with tension limiters on the face of the card, passing through the chrome mount on the GPU block, passing through the card and into the rear stiffener. The rear stiffener features a rubber captive bumper to prevent it from coming into contact with the surface mount components on the backside of the card.
Even with the improvement of the mounting system over the competition, there are a few drawbacks that I encountered. When attempting to attach the blocks to the cards with the blocks already plumbed into the system I found that attempting to hold the backing plate captive with my hand while holding the card in my hand and holding the block centered over the GPU core with my other hand I ran out of hands. When I spoke of this to Gabe at Swiftech his response was "Normally you would attach the block to the card before plumbing it." While this is true I personally do not leave the equipment in the case attached to the water cooling blocks while I do my leak test. This means that you either remove the block from the card, run your leak test, reattach the block to the card or remove the block from the loop then reattach the block to the card and re-plumb your loop.
I think it would be easier to locate both the block and the rear backing plate if you threaded two 2" 4-40 pieces of all thread into opposite corners of the backing plate and used those to locate the backing plate on the card so that it could not move around. When you’re holding the card with the 4-40 rods sticking through the card and the card facing up you could simply slide the block down the 4-40 rod and put the springs, screws and spacers in the opposite corners from the rods. After that you would just remove the rods from the backing plate (since they would only be finger tight) and put the screws, springs and spacers where the rods had been. Had this been the mounting method it would have saved me, roughly, an hour and a half of inventive language.
Now don’t take this is as bashing Swiftech and their mounting system. If you don’t normally remove your motherboard and video cards whenever you are doing a leak test you won’t notice a problem. It all has to do with your personal preferences, me personally, I don’t like the idea that if my water cooling system leaks it’s going to leak all over my expensive parts.
Now that I’ve gotten through the basic explanation of the kit lets get down to the compared systems:
With these components I’ll be testing:
Which will be going up against the Swiftech Apex Ultra H2O-220 and 2 X MCW60 GPU blocks.
For testing I will be using Everest Ultimate Edition to monitor the CPU and GPU temps and the temps will be logged after the system is allowed to idle an hour. For loading I looped 3D Mark 2005 while doing a 32M PI calculation with Super PI Mod 1.4 which takes roughly 30 minutes to complete. This insured that I had a 100% load on the CPU and a heavy load on both GPUs.
I ran the fans included with the Swiftech kit at 12V as that was the same acoustic level as the Thermaltake fans set to roughly 75% on a rheobus. That was reached by matching the drone from the PSU fan so that the water cooling system in each case was not the loudest part of the overall system but instead did not stand out acoustically. Your PC will only be as quiet as the loudest part and in this case it was the PC Power & Cooling 850SSI PSU. The room temp was kept at a constant 19.1*C throughout the testing of both systems (Yes, I like living in a meat locker)
The results for the Danger Den setup:
The results for the Apex Ultra with MCW60 blocks:
As you can see, the Swiftech fared very well for itself. The CPU temps were exactly the same for both idle and load. The GPU temps were the same for idle but under load the Swiftech setup gained some ground.
The cost on both kits is basically the same as delivered and the Danger Den setup doesn’t have the added benefit of a reservoir which makes filling and bleeding the system decidedly more tedious. Also due to the radiator used in the Danger Den setup you’re penalized more for lowering the airflow through the radiator by higher temps. While I didn’t document this I did see the Danger den setup giving decidedly higher temps when the fans were set to very quiet settings of about 1000 RPM whereas the Swiftech kit only suffered by 2*C on the CPU and 3 or 4*C on the GPUs when running at 7V and at 7V you can’t (Or at least I can’t) hear the fans at all.
All in all I’m very impressed with the Swiftech Apex Ultra H2O-220. The performance is on par with one of the heavy hitters that is commonly pimped by the hardcore overclocking crowd and the quality of the product is exceptional. The added extras do come at a bit of a premium but they’re very useful and are commonly something that would be bought out of necessity anyways. The MCW60 GPU blocks are pretty nice by their own right and aside from the problems I encountered while mounting them pre-plumbed I have to say that the mounting system is the best I’ve seen.
There’s no guesswork at all. You tighten the springs down until the screws hit the limiters and you’re done. There’s nothing sticking up on the back of the card to get in the way of the slots above the card so you don’t end up blocking 2 slots (One above the card and one below) by deciding to go with water cooling on the GPUs. Trust me here, on certain motherboards such as the one I tested on and the MSI X16 board the Maze4 GPU blocks can be a nightmare.
Oh, by the way, here’s a pic of the SmartCoils in action:
As you can see they constrict the tubing causing it to keep it’s circular dimension when it’s under compression. Without them the tubing in this picture would have kinked causing a loss in flow.
And now without further delay, the score. I’m awarding the Swiftech Apex Ultra with MCW60’s a 9/10 and the Editor’s choice award for being such an outstanding kit. Swiftech doesn’t leave you out in the cold, if you run into any problems you have phone tech support along with a very good manual to fall back on. I wish that their competition could take a page from Swiftech’s playbook and include such complete documentation.
You’ve read the review now go discuss it here.
Copyright © 2005-2019 Techgage Networks Inc. - All Rights Reserved.