High-end systems can generate lots of heat. We investigate the wonders of watercooling and see if using an EK watercooled setup & Phobya/Alphacool radiators can help.
Building a high end system is great – you can crank up all the settings in your favourite games, beat your friends in 3D mark and listen to the meaty roar of it purring away…Actually, that’s a lie. Most systems don’t produce a meaty roar, but instead a dull, monotone hum as the fans desperately try to shift all the hot air out the back of your case. Curing this problem certainly isn’t easy – with the amount of heat that modern day GPU’s and CPU’s can expel, standard air-cooling systems begin to struggle.
So, how do you solve the problem? Well, this is where watercooling comes into play. As the heat can be transferred away from the source (in the coolant), you can massively increase the surface area with which you transfer the heat to the air. Instead of being limited to a cooler that can only hold one or two 120mm fans, you can use a radiator that holds pretty much as many as you desire, and even then you can use more than one radiator. This massive increase in heat transferring surface area allows you to dramatically reduce the speed of the fans used, or even remove them entirely – reducing the noise of the system enormously.
But, which route is best? Are you better off going down the conventional route by combining a radiator with fans, or going all out with a purpose built passive radiator? With a conventional radiator you’ll obviously need to run the fans as slowly as possible to reduce the noise, but how much of a hit will you take in terms of performance? Well, today I hope to answer those tricky questions.
Over the next few pages I will explain which major system components need liquid treatment, and give you a quick guide on how I went about cooling them for today’s test setup…
CPU Watercooling
Although modern processing cores are becoming more and more power efficient, they’re also being clocked faster and more of them are being crammed onto a single chip, which inevitably leads to a higher heat output. Although conventional air coolers can tame them, few can do it quietly; those that can tend to be very large, leaving limited space for other components within the system, such as RAM. It’s probably no surprise then that the CPU is the most common component to watercool, with a massive range of blocks and kits on the market for this very purpose (check out my review of some of the latest kits here).
For todays review, I decided to go with the EK Surpreme HF water block. Widely accepted amongst the watercooling community as one of the best blocks out there, it should be more than capable of taming the AMD Phenom test chip. It was used in the Custom Kit in our recent round-up, combined with other components it produced the best results on test – nice!
GPU Watercooling
An increasingly popular component to watercool is the GPU. With Nvidia and ATI cramming more and more cores into their cards in a bid to take the performance crown, cooling performance has been compromised. With lots of heat to expel and little space to do it within, modern cards have to rely on high revving fans to shift the vast quantity of heat they produce.
Watercooling won’t just silence a noisy graphics card, it will also reduce the temperatures massively – sometimes by as much as 50%!
The most common way to go about it is to use a full cover block – it takes care of watercooling the card’s core, RAM, VRMs and any other toasty components. Unfortunately, blocks tend to be expensive, and a block designed for a 5870 will only fit a 5870, so with every GPU upgrade you need to buy a new block.
The other method uses a core only block, which has movable mounts, so will work with most cards currently available. You then use a number of small heatsinks dotted around the card to cool all the other components (Ram, VRM’s etc). Whilst this may save you money, it doesn’t quite deliver the visual appeal of a full cover block, nor the cooling performance – with all the heatsinks used you still need a reasonable level of airflow to cool the card.
Therefore, I decided to go for a full cover waterblock as I needed to ensure the card would still work without airflow over it. EK’s 5870 v2 block is the only one available that fits my card, so a nickel-plated, plexi topped version was ordered (to be in keeping with the theme of the build). Despite taking my time (making sure not to break anything), installation took just under an hour, which isn’t bad at all. If anyone is considering watercooling their graphics card, all I can say is read the instructions! Instructions typically end up in the bin, but they were absolutely essential in the installation of the waterblock.
Motherboard Watercooling
With more and more features being crammed into motherboards, it’s no surprise that these too are getting hotter and hotter. Gone are the small heatsink/fan combinations of yesteryear, now motherboards use complex heatpipe arrays with large heatsinks to dissipate the heat. Sometimes these elaborate solutions still aren’t enough though – I had problems with my motherboard in the recent watercooling review I put together. Pointing a 120mm at the Northbridge area solved my stability problems, but the Northbridge still ran hot – testament to just how much heat they kick out these days.
Early motherboard water blocks came in 3-4 pieces: one for the Northbridge, Southbridge and then a final block or two to cover the VRM’s. This method meant that many blocks worked on more than one motherboard, but it was often confusing knowing what block to go for. It was also difficult to plumb it all neatly, leading to messy systems – something us enthusiasts hate. Block manufacturers obviously took note of this, and we now have gorgeous, single piece waterblocks. These only require a single inlet and outlet, also making them much neater to install. As with GPU blocks this does make each block unique to the motherboard it was designed for, making upgrades expensive.
For today’s review I will be using a nickel-plated, single piece waterblock from EK. Installation was simple, but it does take some time. Getting all the fittings lined up can be a right pain, but once that’s done, you just bolt it down and you’re away!
Other System Components
The only other major heat contributors within a high end system are; PSU, RAM and Hard disks. Although hard disks can be watercooled, modern drives simply don’t need it, largely thanks to the performance available from SSD’s – hot running drives such as the Western Digital Raptors just aren’t used anymore.
PSUs are also becoming more and more efficient, with most high end units producing very little noise at all (if any). They’re also impossible to watercool, with the only watercool-able units being pre-built by Koolance and costing a small fortune. It probably comes as no surprise to you then that we won’t be covering this aspect in today’s review – which just leaves RAM…
Watercooling your RAM used to be incredibly complicated, and still can be, but there is an easier route. The difficulty of watercooling RAM depends immensely on which brand and series of RAM you use. A few Mushkin lines, along with the newer ranges of Corsair Dominator RAM are all very easy to watercool, as they have mounting holes already on top, which allow a waterblock to tap into the heat-spreaders cooling ability. Most RAM however, doesn’t have this, so you have to remove the heat-spreaders before applying all new ones purpose built for watercooling – which is why it’s so messy. When building a silent system though it will be worth considering – I’ve had a number of problems in the past with RAM stability due to temperature, and operating in a case with limited airflow they could begin to get very warm.
I therefore decided to watercool the RAM for todays review. Thankfully, I have some of the newer Corsair Dominator RAM modules, so it was an incredibly simple installation (literally just removing, then inserting a few bolts). Once again, an EK Nickel Plated waterblock was used to be in-keeping with the overall theme of the build.
Active Radiator(s)
Only one more source of noise remains – the radiator. With all that heat to remove the radiator/fan combination is an essential part of the watercooling loop. However, watercooling all the internal components to reduce noise would be pointless if you then went and used some ultra-noisy fans to expel the heat from the radiator. There are two ways around this – either use a number of very slow spinning fans (and I mean slow), or use a completely passive radiator. But which route is best? Well, I’m hoping to answer that very question.
I decided to go with a range of Phobya G-Changer Radiators, testing a number of combinations to find which would be best for a full on test. I wanted to keep the setups realistic – ones which could actually be installed into a case. Linking them all together for some uber cooling then was out of the question.
Some cases will accept dual radiators, but bearing in mind how many components I was going to watercool I was pretty certain this wouldn’t be enough. Therefore I decided to start with a triple radiator – most decent watercooling cases will happily accept these in the roof or floor. These cases often have a 120mm mount at the back as well, and so my second active setup was decided upon – a triple 120mm radiator along with a single. I also decided to use the Phobya Nano 2G fans – not only do they look great but they also have some impressive specs. Here are the setups I decided upon:
Phobya G-Changer 360mm Radiator with 3x Phobya Nano 2G Fans -£92.54
Phobya G-Changer 360mm & 120mm Radiators with 4x Phobya Nano 2G Fans -£135.18
Before testing anything though I needed to find out what sort of voltage the fans required to operate silently (or as close to silent as you can get). I started out with 7.5v, with impressive results. From around 60 centimeters away they were completely inaudible, but I wanted more, especially since the other setup is completely silent. With this in mind, I dropped the voltage further to 6v (yes, that is half their rated input voltage). The result was fantastic, with the fans being completely inaudible unless I put my ear right up against them – perfect!
Passive Radiator
The final radiator setup for testing will be Alphacool’s Cape Coras, a solution built from the ground up to be completely silent.Removing heat without any forced airflow is incredibly difficult, which is why the Cape Cora kit is so big – it really is a monster. Coolant enters into one of the black tubular pieces, traveling down one side of the rad. It then crosses over the aluminium heatsinks, transferring its heat along the way. It then exits out the other side of the radiator – simple.
The Cape Cora arrived in two pieces, but the whole thing fastens together with relative ease. My only slight gripe with its construction is the way everything just slots together. The stop barbs on the ends are just pushed in – there’s no screw thread to hold them in. In principal this should be fine, but in testing I had some issues with pressure. I actually had the end of the radiator fly off, spilling coolant all over my desk/monitors and keyboard in the process. After a re-think and a careful rebuild (removing some of the restriction from the loop), I managed to get it all working.
The Cape Cora’s are fairly expensive at £106.18, which I suspect is largely down to the cost of all the raw materials. When you consider the fact you don’t need to buy any fans though, its price looks quite favorable compared with the active radiator setup.
The Loop & Test Setup
Of course, a watercooling loop consists of more than just waterblocks and radiators. You also need a reservoir, pump, liquid, fittings and tubing. Here’s a quick run down of the other components used:Phobya Balancer 150 Black Nickel,
Phobya Zuper Zero UV Red Coolant,
Generic Grey tubing & a range of 3/8” fittings,
2x Laing DDC 10w Pumps w/ Aquacomputer tops.
Although I have gone for two pumps I’d like to point out that this isn’t essential. I decided to go down this route as I had a number of extra components in the loop, such as flow meters, temperature sensors and quick release fittings, all of which add extra restriction. It therefore only seemed fair to give it a boost by adding in an extra pump.
This was also my first time using the Phobya Balancer range of reservoirs, and boy am I impressed. The thing has a really solid feel and weight to it, and the nickel finish is just stunning.
The Phobya coolant on the other hand wasn’t such a success. The colour faded after just a few days, turning to more of a pink than a red. Performance wise though, it seemed to perform pretty much as expected.
As I mentioned earlier I also placed a number of temperature sensors in the loop. These will let me see the temperature of the coolant itself which, along with all the other readings I will be taking (CPU, GPU, NB, and RAM) should give a very good overall picture of how well each setup performed.
The rig used for torturing the radiators is as follows:
AMD Phenom X6 1090T @ 3.9ghz (EK Supreme HF),
8GB Corsair Dominator @ 1600mhz 8-8-8-24(EK-RAM Dominator),
Asus Crosshair IV Formula(EK -FB Asus Crosshair IV),
Asus Radeon 5870 v2 (EK 5870 Asus V2),
OCZ Vertex 120gb,
Enermax Modu87+ 500w PSU
I thought about including the temperatures I was getting with the stock coolers, but it’s not really fair to make comparisons with them. Not only do the fans produce a fair amount of noise, but there’s also no way to isolate each component to ensure it doesn’t get some sort of airflow from another, making the results misleading and false.
Temperatures shown are the Delta values – the temperatures recorded minus the ambient temperature (all in Degrees Celsius). This means that the error you inevitably get by using absolute temperatures can be removed, effectively laying out a level playing field so to speak.
Test Results
The graphs speak for themselves really – the active setups comprehensively outperformed the passive radiator setup. Considering the size of the Cape Cora’s I really was expecting more, but when my CPU went above 70 degrees (actual temperature – temperature listed is Delta T) I decided to pull the plug before something broke (AMD official spec lists 62 as the max running temperature).
As expected, reducing the fans down to 6v had a noticeable effect on the performance of the active setups, but it was nowhere near as bad as I thought it would be – considering the fans were completely inaudible, I think the performance was actually very good. I’d be interested to see how well other brands of radiator cope with such restricted airflow – the low fin density of the Phobya radiators helped them with this enormously.
At 12v the fans are much quieter than the stock fan of the 5870 and the performance is just leagues ahead. Even at 6v performance was still exceptional – having a modern GPU loading with a Delta T under 32 degrees is incredibly impressive, but the fact that it was performed in complete silence makes it even more so.
Similarly the Delta T temperatures for the rest of the system were all great, only the CPU temperature caused concern when running the fans at 6v. With a Delta T of 40 and an ambient temperature of 22, it was riding on the 62 degree maximum temperature of my CPU – fortunately it didn’t rock over and the tests were completed.
Conclusion
The benefits of watercooling are clear to see, not only can it silence your system but it can also drastically improve cooling performance. I can’t tell you how nice it was to turn my system on without actually knowing it was on – a number of times I had to actually check the power LED to see that it was powering up.
It’s not all good news though, as the passive radiator setup failed today’s testing, showing it’s not just a case of throwing together a load of components. The Phobya radiators and fans really were the stars of testing, showing off what watercooling is truly capable of, and that it can be done in complete silence. This setup has more than one plus point as well – there’s a flexibility to its thermal capacity.
With the passive setup what you get is all you have – there’s no way to adjust how much heat the kit can expel, which could become a problem. For instance, you design and overclock your system to perform on the limits of its cooling capacity, but what happens when say, it’s a hot summer’s day and your ambient temperature rises by ~10 degrees? With the active setup, you can sacrifice silence a little by ramping up the fans, adding in room for error, but with passive you’re stuck – you can’t increase the cooling capacity one teeny tiny bit.
Bearing this in mind, and the fact that the passive kit failed when put under load, I’d have to recommend all of you looking to build a silent rig, forget the passive kit and take a long, hard look at the Phobya radiator range. The flexibility, performance and build quality they offer is beyond anything I’ve tested before, which is why they’ve earnt themselves the Vortez.net Gold Award.
Obviously building a passive rig takes a lot of time and money. You’re looking at a good 4+ hours for setting it all up and bleeding the loop, along with around £500 for the all components. However, if you treasure silence, or want to push your rig further without it turning into a hair dryer it’s the ultimate solution.
Phobya G-Changer Radiator Series
Pros:
+ Excellent Performance even with little airflow
+ Great Build Quality
+ Various mounting options thanks to two inlet/outlet threads
Cons:
– Zilch
Pros:
+ Excellent Performance even with little airflow
+ Great Build Quality
+ Various mounting options thanks to two inlet/outlet threads
Cons:
– Zilch
I would also like to thank Aquatuning for providing the test kit for today’s review.
