What started as a giant build log in the Caselabs TX10-D has become a series of tests and reviews now! This time a test on something most people don't compare- 480 and 560mm radiators- because most of these don't fit in a typical case. But seeing how I can fit ALL 10 rads being tested simultaneously in this one case, I decided to go ahead and check them out. This is going to be a big, continuously updated thread with a lot of pictures (each rad essentially gets a review in here) so I welcome all for the ride.
1) Unboxing and Overview
Let's take a look at the radiators in the arena:
EK-Coolstream PE 480 (Henceforth referred to as EK PE480)
EKWB color coordinates their components so green is radiators, orange is water blocks and so forth.
Inside the outer box (yes, it isn't a sleeve as with their blocks) is another box.
Opening the box we see the bag of screws on top of the radiator unit
Pretty much every radiator (other than the Alphacool Monsta) comes with 4 x number of fans of 30mm and 6mm long screws. So don't expect the fancy arrangement henceforth
EK has chosen to use UNC 6-32 thread size and screws for fan/radiator mounting. To aid in installation, they have provided a small Allen key.
The radiator itself is snugly packed inside a paper sleeve and then in bubble wrap- safe enough to be shipped across the world with no hitches.
The radiator has sharp edges and a clean, minimalist look. The frame itself has a matte finish and is easy to dust off. The unit measured in at 530mm long x 130mm broad x 40mm wide.
Here you can see the mounting holes. If you use the provided screws with a standard 25mm fan or directly mount it to a case/stand then there will be no issues with any radiator. But in case you use your own screws that are too long, then they will pierce the tube channels so be careful and double check the screws before using them.
There are 2 threaded extenders that each accept a male G 1/4" fitting on the radiator and come covered by plastic caps. These are just meant to protect against any dust/contaminants getting in so don't use them as a stop plug.
The core is brass and comes clean from the factory. In fact, as we will see later, all the rads go through a Mayhems Blitz Pro cleaning and the results are documented. The radiator is a dual pass, side to side flow type as with other radiators in this roundup unless specified otherwise.
The fins are uniform across the radiator which is always good to see- no bent or scratched fins in this sample.
Now, let's define a fin (yes, we had to come to that):
Each "/" and "\" is 1 fin each. So in the picture above, there are 16 fins in the enclosed section. This number does not change if you split them up:
Still 16 fins in the enclosed section, just happen to be 16 split fins. So by this terminology that academia and industry typically follows, the EK PE480 has a FPI of 19 and not 38 that they advertize- a true 38 FPI core will be terribly restrictive in terms of air flow. I measured an FPI of 18-19 on average here. Each fin is also ~30 microns in width as measured by a micrometer. While we are measuring, the radiator holds approximately 300 mL of fluid, has 12 flattened tube channels each ~2.19 mm wide. All this will be in a "cheat sheet" later on.
One more feature that the EK PE480 has is a removable core- this makes painting the outer frame a lot easier without having to mask off the core. This is a feature shared by the XSPC AX480 as well and I will demonstrate it in there (slightly easier in that case). As of October 23, 2014 the EK PE480 costs $89.95 in the US.
XSPX AX480 V1
The radiator comes in a nice box again, and has a warning label about using the provided screws and in the right manner.
Inside the box is the bag of screws on top of the radiator.
The screws are UNC 6-32 with a standard Phillips head so no driver comes with this.
The radiator comes protected in bubble wrap.
Here is everything that comes in the box- the radiator and 2 sets of 16 screws each. The trend so far is providing enough fans for push or pull only.
The radiator comes with a built-in gasket if I can call it that. Whether or not it aids in performance will be seen later but it definitely adds to the clean look here in my opinion.
As seen above, it is a similar case as with there being no screw protector (Incidentally, shouldn't these be called tube channel protectors instead?) and longer screws piercing the tube channels. Just use the provided screws properly and no such thing will happen.
The frame has a nice matte finish that is easy to dust off as well. The dimensions were 528mm long x 130mm broad x 40mm wide.
The fins are uniform across the board in consistency.
There are again 2 extenders on the ports that accept any G 1/4" fitting. These extenders come with plastic caps on them to stop physical contamination during storage and transportation.
In order to remove the core for painting the frame, the extenders need to be removed first. I had to use pliers on one of these, so if that's the case then be sure to use a piece of cloth or paper in between so as to not scratch the fittings. Note that disassembly was done after all tests were completed to prevent any possibly installation factors.
Next remove the 4 Torx T10 screws on either side with an appropriate driver (I did it the other way round as seen in the pictures):
The radiator core is help by tape but once the two extenders and one side plate is removed, it should easily slide right out:
Once the frame is ready to go, reverse the steps and assemble it back. XSPC also includes a desk stand to take advantage of this feature. The radiator has 30 micron wide fins at 18-19 FPI, 12 channels 2.2 mm wide, was clean out of the factory and holds 300 mL of fluid. At this point, I contacted both EK and XSPC because these two radiators were very similar in specs. Here is a direct quote from EK:
There are very few radiator OEMs available currently. It is common for radiator cores from different companies to overlap and have the possibility to be manufactured in the same factory. What makes radiators stand apart, both in performance and aesthetics, are their FPI and external casing.
So if you see similarities in radiators across brands in terms of packaging or otherwise, now you know. The XSPC AX480 V1 costs $114.95 as of October 23, 2014 in the US.
XSPC AX480 V2
As far as I know, this is the first public mention of the AX V2 series. You can find out if your sample is a V2 by seeing if there is a sticker "V2" by the barcode on the box. Mine did not have the sticker but I have been told this is definitely a V2. The only difference between AX V1 and AX V2 is with the fin design so this will be a short section.
Let's look at fins:
These are louvered fins on the XSPC EX560 (Yes, I jumped a bit here) with the cuts opened up. The louvering introduces local turbulence in the air flow, and this can be useful or even worsen performance depending on the situation. By contrast, these are unlouvered fins on the XSPC RX480 V3 (Another jump):
As you can see, there are no cuts at all. Louvering fins ends up only hurting performance at low fan speeds/air flow and this is something you will see being a feature of the rads excelling at such conditions. Now let's look at the AX480 V1:
The AX480 V1 has louvered fins, and I expected that the AX480 V2 would have no louvering. But:
There are still cuts, but not fully opened up. I am not sure if this is enough to have an effect but we will see later. Everything else is the same, and the AX V2 will replace the AX V1 (or already has depending on the region) at the same price point. In fact, when you have them together makes it impossible to distinguish one from the other unless looking very closely:
XSPC RX480 V3
No confusion about the version here, the RX480 V3 comes in a white box and has the same warning label as before:
Inside is the bag of screws on top of the radiator as well 2 XSPC stop plugs.
These screws are UNC 6-32 with a Phillips head and the stop plugs are nickel plated and low profile.
The radiator itself is packaged in a paper sleeve which is then inside bubble wrap.
The frame has a nice matte finish and the radiator dimensions are 515mm length x 130mm breadth x 56mm width.
The fins are once again very consistent across the board.
There are 4 ports here- two on the top/bottom and two on the side- that come covered in plastic caps. Hence the reason why 2 stop plugs are included. I like having multiple ports to aid in filling, bleeding, draining and also giving options with tube routing so this is a welcome move. One of the 4 ports was slightly cross-threaded and I had to use a G 1/4" tap to get the stop plug to screw in properly.
There are no screw protectors but screws that are longer than what should be used will not piece the channels- they will only bend in some fins with minimal performance effects at best.
The RX480 V3 came in clean out of the factory, holds 500 mL of fluid, has 30 micron wide fins at 13 FPI, 9 tube channels that are 2.1mm wide and costs $122.95 in the US as of October 23, 2014.
The radiator is packaged in a similar manner to the RX480 V3 so I will post pictures here interjecting in only when need be:
The frame has a matte finish and the radiator dimensions are 595mm length x 143mm breadth x 35.5mm wide.
The fins are not uniform throughout in this case, with quite a few bent out of the box itself.
No extender fittings here just like with the RX480, and the 2 ports on the radiator come with plastic caps on.
There are no screw protectors here, and there is enough wiggle room for the screws to pierce the channels if (a) too long a screw is used improperly and (b) it is screwed in at an angle. Again, nothing that a normal usage with the provided screws will experience.
The EX560 had minor residue left inside after the soldering/brazing process, holds 285 mL of fluid, has 30 micron wide fins at 17-18 FPI, 13 tube channels that are each 2.05mm wide and costs $89.99 in the US as of October 23, 2014.
Alphacool NexXxos Monsta 480 (Henceforth referred to as Alphacool Monsta 480)
In my case the radiator actually had shipped with no accessories in the box other than a few black screws so I had to contact the retailer and get the accessories separately. The actual screws are supposed to be copper to go along with the provided copper stop plugs and logo sticker. But less than 1 month of usage had resulted in discoloring:
Assuming nothing's wrong with your unit, there will be a box with 8 pouches of screws in it. Each pouch will have 4 screws- either 30mm long for mounting a standard 25mm wide fan into the radiator or 35mm long for going through case/gasket->fan->radiator. Based on this, I would believe the manufacturer intends for this radiator to be used in a push-pull fan configuration as opposed to every other radiator in the roundup. The screws are type M3 and have a hex head- this may require the use of a washer if your case fan holes are bigger than the screw head.
There 7 ports on the radiator- 2 each of the top, side and bottom on one end and a single port at the bottom on the other. This again is very handy to fill, bleed, drain a loop and route tubing easily.
There are no logos anywhere and instead you get a few stickers for you to place accordingly once installed so as to have the logo oriented right side up. I like that they give us a choice here! The radiator frame has a glossy finish that picks up stuff easily but can be wiped away as well- not a dealbreaker by any means unless the radiator comes with dings and scratches out of the box:
The radiator dimensions are 522mm length x 125mm breadth x 86mm width. Note that these are not 80mm wide as many believe and can result in an incompatibility if you haven't accounted for the large size.
The fins are pretty inconsistent to be honest with even the FPI having a broad range as we will see soon.
On the plus side, even if you use the 35mm screws by mistake instead of the 30mm long ones, there is a screw protector plate throughout to make sure they hit the plate first and not the tube channels.
On the negative side though, the radiator was quite dirty out of the box- something Alphacool/Phobya really need to address before more and more customers decide to go with other options. The radiator holds 870 mL of fluid, the fins are 25-35 microns in width at an FPI of 8-10, there are 10 channels each of 2mm width and the radiator costs $139.95 in the US as of October 23, 2014.
HardwareLabs Black Ice Nemesis 480GTS U-Flow(Henceforth referred to as Black Ice Nemesis 480GTS)
The GTS (S for Stealth) comes in a nice box with the Nemesis writing in silver (as opposed to gold in the case of the GTX versions).
No, that isn't a Corsair Gaming logo
Inside is another box with a warning label about proper screw usage.
Opening it up, we are greeted to the radiator immediately. No paper sleeve, no bubble wrap. Normally I would be ok with this seeing how snug everything is and the rad being in perfect condition to prove that point but those flaps at the top are just asking for trouble- in fact a few people have already mistakenly had the flaps close down at an angle down on the fins immediately below resulting in bent fins with paint scratched off. Upon talking to HWL about this, I was informed it was an issue with a small batch of products due to the packaging providers and steps had already been done to rectify this (and they had too, as we will see very soon).
The screws are in the side compartment with 16 each in the M4 type, 30 and 6mm long. The radiator has a "Dark matter" finish for most customers (A matte finish for the European market, not due to any laws/regulations but simply from a market poll):
It is a very unique look by far (In fact, I am told HWL came out with the matte finish first as well) and definitely something one needs to see in person before deciding whether they like it or not. The finish means dust and other contaminants get retained but on the other hand they are also not easy to spot:
The radiator dimensions are 520mm length, 132mm breadth and 29.5mm width.
Not the easiest to photograph but hopefully you can see that there are screw protectors here in case someone uses the wrong screws.
The fins were extremely consistent across the board with no visible damages at all!
There are two ports that come in uncovered but this was one of the cleanest radiators out of the box.
The radiator holds 190 mL of fluid, has 25 micron thick fins at 16-17 FPI, and 12 tube channels at 1.22mm width each. Definitely the thinnest fins and tube channels out of any other brand in this test. It costs $98.95 in the US as of October 23, 2014 and comes with a limited lifetime warranty- something no one else offers!
HardwareLabs Black Ice Nemesis 480GTX (Henceforth referred to as Black Ice Nemesis 480GTX)
The "Nemesis" wording on the GTX (X for Xtreme) is in gold as opposed to silver with the GTS variant.
Here we see what I was referring to above with the updated packaging- I am content with the packaging now but I can see where others would prefer at least a bubble wrap sleeve on top of this.
The screws (same as with the 480GTS) are in the side compartment again.
The Nemesis 480GTX has the same dark matter finish, just in a bigger frame. Unlike the Alphacool formula of having the same FPI but increasing the core thickness/number of rows, here the thicker GTX frame also means a differently optimized core and fin structure. The dimensions are 518mm x 133mm x 54mm.
Again, extremely consistent fins that were flawless out of the box.
No screw protector here, but the screws can only hit the fins as the tube channels are offset. I still only recommend using the provided screws of course!
The radiator has 25 micron thick fins at 13 FPI, and 12 tube channels at 1.22mm width each. It costs $142 in the US as of December 23, 2014 and comes with a limited lifetime warranty. Oh, and it was the cleanest radiator of all the ones tested in terms of the Mayhems Blitz Pro treatment.
One important thing to note is that the Nemesis radiators (only the GTX, not the GTS variant) follow the front to back flow scheme and not side to side. When the two ports are facing you, the port on the right has a lower travel distance to the core compared to the port on the left. So depending on the way you orient the fans, choose the inlet and outlet accordingly. If you have the inlet on the right then the side with the barbs is the hot side and the other side is the cold side.
HardwareLabs Black Ice Nemesis 560GTX (Henceforth referred to as Black Ice Nemesis 560GTX)
This is the 560mm (4x140) version of the Nemesis 480GTX and shares all the similar features except with these changes: Dimensions are 592mm x 153mm x 54mm, fluid volume 510 mL, 16 tube channels each 1.22mm wide and costs $207.95 in the US as of October 23, 2014. One thing to note here is that HWL has now moved to a more standard 15mm fan spacing on their 140x mm rads which is good to see.
HardwareLabs Black Ice SR-1 560 (Henceforth referred to as Black Ice SR-1 560)
This was purchased used from the original owner who had had minimal usage on this with just distilled water for a couple of weeks. So other than missing the original packaging and screws, the radiator itself was is pristine shape (after having used a Datavac blower to remove any remnants of dust on the fins).
The frame has a matte finish and the radiator dimensions are 605mm x 152mm x 56mm.
The radiator has only 2 ports and follows a standard side to side flow direction.
One of the sides has a larger gap to the core than the other to act as a shroud and help eliminate dead spots in air flow.
The fins are consistent across the radiator.
No screw protectors here but the tube channels are offset and so longer screws will only hit the fins.
The rad holds 530 mL of fluid, has 30 microns thick fins at 9 FPI , 14 tube channels each of 2.02mm width and costs $197.95 in the US as of October 23, 2014. The SR-1 has 20mm fan spacing which may or may not be conducive to you.
Mayhems Havoc 480
Mayhems recently released their Havoc series of radiators to the public for sale. Boasting a quad pass, triple tube flow one would expect a longer mean residence time of the fluid inside the radiator at the expense of potentially liquid flow restriction.
The package that a customer gets is not the radiator alone. In this case, it came with a 480mm "Mayhems Tech" fan grill and a Mayhems Blitz lite cleaning kit (Blitz Part 2 solution and a set of large sized gloves). Everything came in nicely secured in a mother box with a lot of protection- good to see, especially when liquids are part of the package. Speaking of liquids:
Blitz Part 2 (with an instruction sheet and MSDS included) gives you 4 liters of effective cleaning solution for the entire loop. The Blitz cleaning kit (Pro version) has a Part 1 acid bath meant for radiators only and the absence of it would suggest confidence in the cleanliness of the radiator. One would expect nothing less here given they advocate and produce the cleaning kit themselves.
The fan grill (3mm thick, and length depends on the size of the radiator accordingly) actually pleasantly surprised me. I have never been a fan (heh) of grills before but this is well made and the logo is well incorporated- not at all a sore sight in my personal opinion. Usage of the grill is a variable that could affect airflow through the radiator and so I have not included it in the testing process here to keep everything consistent.
The radiator is one of the first 60 retail samples, and we see a spec list on the box- 9 FPI, quad reflow, triple tubed, split fin, anti piercing plates. So far so good.
Inside the box you are greeted with a warning/instruction sheet for the Blitz solution and also an MSDS- transporting liquids into a foreign country isn't easy as Mayhems found out themselves! I appreciate everything being made available as a hard copy for reference. After all, one is building a PC here presumably and expecting people to have another device on which to view this information is a bit much.
On the side of the radiator and inside the box is the set of mounting screws.
You get 16 x 30mm long UNC 6-32 screws and 4 x 6mm long UNC 6-32 screws. There should have been 16 of the short screws as well, and I have been told more of them will be included in the packages henceforth. You thus get enough screws for a single set of standard 25mm thick fans.
The radiator comes bubble wrapped on top of all the exterior packaging so I am confident it will withstand most delivery services around the world. Mine arrived all the way from the UK as it is without any issues.
There are 2 ports on the radiator, which come with a dust cover on them. The covers are actually handy if you end up needing to flush out other rads with an acid bath- these will stay screwed in there no problem.
On the side is the Mayhems logo which is mirror symmetric and thus works out irrespective of which side you mount it.
So this is actually a unique 1+2 tube flow, and not what I expected when I read triple tube, quad flow on the box. We will see soon how this affects things. The radiator dimensions came in at 520 mm x 124 mm x 63mm.
The fins are mostly consistent across the board and do come in at the advertized 9 FPI. These are louvered fins, and not splitter type as seen. By "Split fin", I believe Mayhems was referring to the cuts in the fin (louvering) and it is a matter of terminology here. Irrespective of what terminology one goes by, I would like to see the terms being clearly explained in case people think otherwise. This could lead of cases of potential misleading even if the intention was otherwise. Given that this is still not settled as of Nov 21, 2014 I can only state this much at this point.
The louver angle is one thing I noticed was inconsistent. Talking to people in the heat transfer industry and looking up a few reference books, for every 1º of louver angle with respect to the fin plane there is ~1% increase in performance up to a certain point. Now this is because of turbulence created by the louvering and is mostly affected at mid-high fan speeds/airflow through the rad. At low fan speeds/airflow, this may not help out and potentially even negate performance. To be completely sure, one needs to have modeled and tested out airflow through the particular core in a wind tunnel. But looking at the other radiators that claim to be superior at low fan speeds/airflow (Black Ice Nemesis, XSPC RX V3) they all have non louvered fins. We shall see how this works out as well.
There are indeed screw protector plates to make sure one has a last line of defense before piercing the tube channels if using longer screws than what should be used. The radiator holds 580 mL of fluid, has 30 micron thick fins at 9 FPI and 12 tube channels each of which is 2mm thick. As of November 21, 2014 this is not yet available for purchase via a retailer in the USA and the only way is to buy it directly via Mayhems at £79.16 ($124) before any applicable duty/VAT.
One more thing- a few hours before I began updating the thread to include in the Mayhems Havoc, Mick informed everyone that they had received feedback from users and reviewers alike and in the next revision (a much larger batch), the following items will be included/continue to be included:
1) 100 mL of Mayhems XT1 Clear concentrate (2 liters of coolant effectively with distilled water)
2) Fan grill (previously stated to be free only for first 60 buyers)
3) Blitz lite kit (100 mL of Part 2 concentration to give 4 liters of cleaning solution effectively with distilled water)
4) A thicker coat of paint (this sample was quite fine in my opinion, but others did notice discrepancy of paint coverage on the frame)
So the overall package you get is now looking better for the money than before. Without these items, would a lower cost (say $100-110) for the radiator alone be a better option? Perhaps not for those looking to include this as part of a new rig. But for those upgrading radiators or just switching things around, it is not as easy to justify. Oh, and did I say this was one of the cleanest rads in my inventory? I expected that would be the case but always good to know.
Testing methodology:- I used a Swiftech MCP50X pump with a FrozenQ 400mL cylindrical reservoir. The pump was powered by a direct SATA connection to an EVGA 1300G2 PSU, and was controlled by an Aquacomputer Aquaero 6 XT. There was an in-line flow meter previously calibrated, as well as a Dwyer 490 Series 1 wet-wet manometer to measure the pressure drop of the component under test- in this case each radiator. Every component was connected by 1/2" x 3/4" tubing, compression fittings and 2 T-firings with the manometer.
The lower the pressure drop from the component, the better it is in terms of being liquid flowrate optimized. Most rads are not as restrictive as, say, a CPU waterblock so this won't affect most people unless you plan to have a lot of them (especially the Nemesis GTS radiators where the thin channels and the core optimized for performance has a clear effect). The EK PE and XSPC AX radiators are indistinguishable here which, knowing about the common OEM, does make sense.
3) Air flow restrictions
Testing methodology: For the 480mm radiators, a Noiseblocker-eLoop B12-4 was mounted in "push" and an Extech 45158 thermo anemometer was placed 6" away from the rad/fan assembly and in the same spot each time to eliminate the effect of deadspots or variation in the X/Y axes. For the 560mm radiators, a Noiseblock Blacksilent Pro PK-3 was mounted in "push". Both fans were powered and controlled by the Aquaero 6 XT.
I realize linear aiflow isn't as useful a metric as volumetric airflow. But at this point, without a sealed box with a duct of controlled size/shape or even a closed adapter of sorts, I can't do much better. Knowing that these numbers also vary from fan to fan, I can only display these to used relatively to each other and not with any other reviewer/tester's results.
While this is a bit of a jumble, somethings are pretty clear- the Alphacool Monsta predictably suffers from least airflow through it despite the low FPI. The XSPC RX480 V3 and the two HWL Nemesis rads have the most linear airflow through them consistently. The Mayhems Havoc is in the middle of the pack consistently along with the EK PE and XSPC AX rads.
Not much of a surprise in here given what we know of the fin and tube channels, the SR-1 560 has the most airflow compared to the other two. But how would the liquid and air flow restrictions effect thermal performance?
4) Thermal tests
Testing methodology: Everything needed (monitor, peripherals, motherboard w/CPU and GPU, radiator, PSU and so forth) was placed in a sealed climate controlled box at 25 +/- 0.05 ºC. Each radiator was connected by Koolance QD3's for easier changing of fans and radiator. The flowrate was held at 1 GPM constant (because liquid restriction tests have already been done) and I believe people generally try to maintain the same flowrate and not hold a pump at a fixed power if controlling them is an options- that's how I do it personally anyway. The CPU, an Intel i7 4770k at 4.6 GHz and 1.3 Vcore, was held at a constant load using a custom XTU profile and the GPU, an EVGA GTX 780 Ti Classified KPE under load from Unigine Heaven 4.0 at 1080p/extreme HD settings, was overclocked and overvolted such that the total system power draw was 850w as measured by a Kill-A-Watt unit inside the hot box. Of course this doesn't mean the rads were attempting to dissipate 850w of heat, especially after PSU efficiency and power consumption from the other components (RAM, fans, pump, SSD) are taken into consideration. My point here was to maintain a near constant heat load into the liquid loop and this helped achieve stable liquid loop temperatures (As measured by 2 separate in-line temperature sensors hooked up to the AQ6) pretty quickly. The tubing and fittings were insulated by a heater sleeve with the heat function not being utilized (doesn't provide anywhere enough heat to get to the 700+ watts I wanted anyway). Every single measurement was done twice to be sure.
I would love to get a water heater with a variable controller instead to make things simpler. But the in-line variants seem to top off at ~300-400w which really don't test out quad rads enough, and the bulk heaters are simply not safe enough in my opinion to use in an isolated, sealed chamber. If there are safer options that don't break the bank, I would love to re-test so I can then accurately map out the watts dissipated at a given delta T (loop - ambient) value for example.
Finally, each fan speed is indicated in the bar graphs below as well as the fans used. I tried my best to not undervolt any fan as much as possible to remove that slightest bit of possible variation from run to run when undervolting and that's why you will see the use of several different fans- all in "push" at this point. To put things in perspective in terms of noise level, the B12-1 at 600 RPM registered 20 dBA and was barely audible even when run outside in the open, and the B12-4 at 2200 RPM registered 39 dBA and was sufficiently loud so that I could hear them through the hotbox (which isn't acoustically sealed, so acts as a typical case I would say).
Again, I have to say these results should be taken relative to each other only at this point due to the different testing methodologies and fans used. I would love to be able to put out watts dissipated vs fan CFM and accompany it with a library of information to get CFM vs RPM values for as many fans as possible. But that's not something I can do realistically as a hobby/out of interest.
Having said that, we can see a few things here:
1) The XSPC RX480 V3, the Mayhems Havoc and the two HWL Nemesis rads are great at the lower fan speeds/air flow. The optimized fins and cores, the channels and rows all work together.
2) The EK PE480, XSPC AX480 V1 and XSPC AX480 V2 are pretty much the same for all intents and purposes- buy as per your preference of looks/brand/warranty/price. They also only are in favorable conditions at the higher fan speeds/air flow.
3) The Alphacool Monsta is a monster at higher fan speeds/air flow. I will test this and another radiator in push-pull as well where I think this would fare better given the extreme size of the core.
4) The HWL rads scale great with airflow while the XSPC RX V3 and Mayhems Havoc do not- goes to show that liquid and air flow restrictions don't tell everything and the entire package can affect things. The Nemesis series was optimized for scaling and it has showed to be the case in here.
Now for the 560s:
Not much of a surprise here given the smaller sample size of radiators, and also knowing that the SR-1 has a successor coming in very shortly. The EX560 also costs a lot lower but all that being said if you can afford it then I don't see anything really beating out the Nemesis 560GTX consistently across various fan speeds.
Performance should never be the sole metric to base a radiator choice on. In fact, never buy a radiator separately- get the radiator and appropriately chosen fans in terms of airflow, static pressure and noise levels you are comfortable with. Those bar graphs above also don't tell the entire story, especially when the difference is a couple of ºC in some cases. Look at the entire package- accessories, build quality, warranty, radiator ports and other features and definitely the price point. The above results will hopefully help people make decisions but if you quote these anywhere (and you are of course free to do so) then please mention that the restriction and thermal results should be taken relatively within each set only. Whenever I see people quote results from Martin or Bundy saying "xxx" rad is better than "yyy" rad at "zzz" RPM, it frustrates me a little knowing that the testing methodologies are different and the fans play such an important role. If I were to run the 480mm rads with a fan that moves less air through the rads compared to the eloop, then the XSPC RX480 V3 would have probably been the leader at higher fan speeds also. Ditto with if I used fans better optimized for rads than the eloops where the Monsta would have started leading at lower fan speeds! That's why I placed the fan details in each thermal test as well in order to minimize this potential issue.
I realize this is not a conclusive roundup by any means, heck it isn't even an absolute one. As I mentioned before, I would have definitely preferred to be able to post Watts dissipated vs fan CFM for each radiator. This would be a much better scientific report, but that also doesn't help most people without providing (a) an accurate means of estimating how many watts each particular rad in each particular system would have to dissipate for a given delta T value and (b) having a compendium to get CFM vs RPM for each particular fan being used. This is definitely not a 1 person job even if he/she is paid to do nothing but this. So I want to encourage people to start sharing their own results- even small tests are better than nothing! Just make sure you do it properly and rigorously. Ok that was my small rant, and I am done now! Thanks for reading
Thanks! So far everything I've tested has been for the big TX10-D build so if nothing else, I know now what to use where with the specific components I have. Depending on my time and money (and of course manufacturer interest), I may well add to this with other rads also.
Edit: Can you guys read all the test result plots correctly? I am not sure why but the images seem to have been downsized literally on my monitor here despite it being ok on a few other places..
Yeah I am not sure what's going on with the graphs. It's fine in other places but here and LTT where they host the images themselves, it's the same issue. I will try again or just put up links to the full images soon.
Hah you have no idea. Stren pointed out if I had done multiple runs and that's when I realized I hadn't mentioned that I had done each measurement twice just to be sure. There was no change but always good to be sure!
LOL ya the 900D just became too small for me, so bumped up to a Gunmetal STH10. I will be housing 2x 560's, 1x 480, 1x 360, and 1x240. I really wish the top part could fit 560 rads without modding, but oh well, already have 2 480s in my 900D.
I am really hapy to see they do great with Range, as no matter what RPM speed you are at, they perform very well. This was a great roundup, and I think it is time you take a vacation
I think it's both the Nemesis rads being optimized to scale with fan speeds and the RX V3 being quite optimized for lower fan speeds. The GTS has a higher fin density also, and uses thinner fins as well. As I said, there are multiple factors that contribute to thermal performance and one cannot base it on thickness and FPI alone anymore.
Note that the results I have at these fan speeds may not be indicative of what you get unless everything is the same (especially the fans). Don't quote performance and fan speeds alone especially when different fans have different airflow and noise profiles. Also remember that the RX has 2 extra ports which is something that the Nemesis rads don't. Judge the entire package together
Edit: Another thing is that the temperature sensors have their own limitations. If there are two rads within 0.5 C of each other, I really wouldn't take that as a legitimate argument to go with a particular rad just based on thermal performance.
The second would help in reviewing the results as both fin density and rad depth **can** add cooling capability but the fans need more "oomph" to push it thru.
Question..... I first saw the matte finish discussed in one of Bill Owen's reviews and I both liked (appearance) and disliked (thermals) it at the same time. Looking at the numbers, it becomes readily obvious that the cooling by the rad, and by that I mean the fins, takes care of the majority of the heat load, but that leaves a substantial amount that is tossed off my radiation from rad shroud,s block surfaces, tubing, component surfaces, backplates, etc..... seems to me that this thick coating would "insulate" the transfer of heat in this manner. Did you notice or measure lower surface temps on the rads with the matte finish ?
Good idea. I am not sure if rad depth alone is sufficient. One thing I haven't mentioned in there is once I opened up the AX480 and saw the core by itself, I realized that the frame dimensions don't indicate how much of it the actual core is. If you look at the dimensions above, then some rads have done a better job of fitting in a core in a smaller space but then again others may well have more length and breadth of the core for the same outer dimensions. It's hard to explain this in just writing, but without having access to the actual cores in each case, it's mostly just a hypothesis backed by the visible core sizes.
Radiation through air or even the rad frame really shouldn't be a major factor at these temps. Forced convection is by far the dominant heat transfer mechanism. But no, I haven't tested this so I can't rule it out completely.
Thanks for sharing, I think that is first testing Ive seen with new rads...and really nice job with data and testing, that is a lot of work. And the black ice nemesis does look like a good all around rad at all fan speeds.
A forum community dedicated to overclocking enthusiasts and testing the limits of computing. Come join the discussion about computing, builds, collections, displays, models, styles, scales, specifications, reviews, accessories, classifieds, and more!