Choosing the Best 120mm Radiator Fan: Testing Eight Fans with Corsair's H80
by Dustin Sklavos on August 23, 2012 12:01 AM ESTTesting Methodology
If you've been keeping up with our case reviews, our testing methodology for the fans here is going to seem relatively similar in some ways. Our test system may seem a bit unusual in more than a few ways, but stick with me and I'll explain why I put it together and tested it the way I did.
| Fan and Radiator Testing Configuration | |
| CPU | Intel Core i7-2700K overclocked to 4.4GHz @ 1.4V |
| Motherboard | Zotac Z77-ITX WiFi |
| Graphics | Intel HD 3000 IGP |
| Memory | 2x4GB Corsair Value Select DDR3-1333 |
| Drives | Kingston SSDNow V+ 100 64GB SSD |
| CPU Cooler | Corsair H80 |
| Power Supply | Corsair CX500 |
| Enclosure | BitFenix Prodigy with 200mm BitFenix Spectre Pro intake @ 5V |
The processor, with its healthy voltage boost and overclock, throws a pretty substantial amount of heat at our cooling system. Testing with an i7-2700K at stock speeds would defeat the purpose; Intel's own stock cooler can handle that, we want to "separate the men from the boys" so to speak.
So why use a closed enclosure, and a Mini-ITX one no doubt? As it turns out, my experience in testing Origin's Chronos LAN box suggested that this might actually be ideal. Removing the middle drive cage allows for a straight shot between the Prodigy's intake and the radiator fan, allowing us the opportunity to test how quietly and efficiently the fans can run in a closed system with no real acoustic baffling, while the 200mm Spectre Pro attenuated to 5V runs both quietly enough to not significantly impact results while providing enough airflow to ensure the radiator fans can do their job. Using a larger enclosure felt like it might complicate things with too many variables; the small and wonderfully efficient BitFenix Prodigy felt perfect for the job.
Since a dedicated GPU wasn't needed, one wasn't used. This prevents a graphics card from generating additional heat or noise or deflecting airflow.
.jpg)
Finally, for the closed-loop cooler we used Corsair's H80. Our own testing proved this was a solid performer and fairly representative of 120mm closed-loop units. The H80 includes a thick, beefy 120mm radiator as well as having dual fan headers built into the waterblock that run non-PWM fans at a constant 12V. I elected against testing in a push-pull configuration, though, to isolate individual fan performance; test results are in a push configuration only.
Thermal and acoustic test cycles were done the same way as our case reviews. First, the system is left powered and idle for fifteen minutes. At this point the sound level is tested, room ambient temperature is recorded, and idle temperatures are recorded. Then eight threads of small FFTs in Prime95 are run for fifteen minutes, and load temperatures are recorded; since the block runs the fans at a constant 12V, the only fan that changes speed (and thus noise) is the stock H80 fan, so the noise level for that fan is recorded again during the Prime95 run.
Thank You!
Before moving on, we'd like to thank the following vendors for providing us with the hardware used in our roundup.
- Thank you to iBuyPower for providing us with the Intel Core i7-2700K.
- Thank you to Zotac for providing us with the Z77-ITX WiFi motherboard.
- Thank you to Kingston for providing us with the SSDNow V+ 100 SSD.
- Thank you to Corsair for providing us with the H80, the SP120 fans, and CX500 power supply.
- Thank you to SilverStone for providing us with the Air Penetrator AP121 120mm fan.
- Thank you to BitFenix for providing us with the Prodigy enclosure and Spectre Pro 120mm fan.
Facebook
Twitter
RSS
113 Comments
View All Comments
Hummingbird - Thursday, August 23, 2012 - link
I don´t think that this fan roundup really helps to decide which fan is the best out there as long as the Corsair H80´s radiator is far away in terms of it´s dimensions (depth and fin gaps/distance) from the standard radiators which are normally used for H2O setups.eric appla - Thursday, August 23, 2012 - link
Thank you for efforts with the review Dustin. When I seen Anandtech doing fan review I was very excited that I'll see very detailed and scientific approach as I'm used to from AT articles. Unfortunately this review was not fully what I expectedWhat I'd find useful in the Fan reviewes is also some CFM mesurements and noise measurements.
Noise measurements are very tricky as even A weighted dBA noise level is not telling full story. I would suggest to provide noise spectral picture as it can show at what frequencies are the dominant components of the noise. Some fans do hum, some squeek.
On the CFM and static pressure front it will be interesting to measure them in few scenarios.
1) open air (case fan scenario)
2) dust filter (case fan scenario)
3) low restriction radiator for example XSPC RX series
4) medium restriction radiator for example XSPC RS series
5) high restriction radiator for example XSPC EX series
Pretty much every radiator manufacturer makes few different radiators optimized for different fans based on the static pressure and airflow they can deliver.
Full review like this will show people that there is no single best fan in the world, there are just best fans for particular usage scenarios.
Challange is on :)
For Fan reviews to use with radiators I usually go to websites specialized on this narrow subject such as martinsliquidlab or skineelab as they have the required test equipment. Have a look at what these guys do, I think you'll find it exciting and inspirational and as I know you based on your approach to reviewes you'll surely find a ways to improve their methodologies even further.
Thanks again for all the efforts
krumme - Thursday, August 23, 2012 - link
A spectral plot wont bring us any closer, as there is dynamics involved to, and the charateristics of that dynamics. Besides that its expensive to do.Just add a colum with subjective impression/valuation and notes, fx. clicking, high picth tone..., even distributed...
Its the subjective that matters in the end.
Robalov - Thursday, August 23, 2012 - link
Nice article and get's right to the point, however I'd like to see 20+ fans tested.They're are so many touting that their 'special' design knocks the socks off the others.
Fans don't seem to get the air time (ba dum tiss) of other components are are left entirely up to personal recommendations on the forums, as the specs are just massaged or even outright lies.
As an aside, CoolerMaster really are a horrible brand.
BlueReason - Thursday, August 23, 2012 - link
First off, it's nice to see Anandtech making an effort to evaluate fans, a component sorely under-prioritized in rig setups (not to mention extremely difficult to review).However, as stated by others, the NF-F12 really should have been tested. I don't even like any of Noctua's other fans, but the F12 is possibly the most specialized, single-purpose consumer grade fan on the market, engineered ground-up for heatsink performance. Whether it delivers on that performance is debatable (in my experience, it does), but its exclusion from the comparison is rather odd.
Are any of the reviewed fans even PWM?
On a side note: some F12 units seem to have acoustic issues, hence the occasional "loud" comments. My first F12 suffered from this and was replaced by Noctua, to good effect.
khanov - Thursday, August 23, 2012 - link
No no no. You can't say that those Corsair fans win, since they only manage to equal the performance of Noctua's "old gentleman" NF-P12. That is not a win from Corsair as those Noctua's have been available for 6+ years now (but still it is a good first effort from Corsair).None of the fans you tested can actually best the performance of a Noiseblocker M-12P at the same or lower sound pressure level on a low profile, high-fin-count rad. (and that is what these high pressure fans are for, right?). So the real winner is Noiseblocker, but you decided to exclude them along with Noctua. Shenanigans, Dustin.
Noctua's NF-P12 are a good second best for many peeps into water cooling and are usually available at a reasonable price. If you are planning to push/pull a 3x120 or 4x120 rad for example, then the cost of 6x or 8x your chosen fan adds up very quickly. So to be reasonable about it the Noctua's should realistically win simply because the Corsairs are WAY overpriced right now (flavour of the month?).
Noiseblocker's M-12P's are still the real performance winner (cooling vs perceptible noise) but may cost you several arms and legs to import depending where you live.
I like that you have a WC setup and are now testing fans vs. rads. I look forward to future fan/rad. tests but just do it properly please. Excluding the real players is like saying the new Ford Fiesta is the fastest car on the Nürburgring because you deliberately excluded any real competition!
kg4icg - Thursday, August 23, 2012 - link
Nice review. I'm using the H100 instead of the H80 and yes I am using the stock fans. The thing is, I also have the Corsair Link hooked up inside my system which is controlling fan and pump speed instead of 1 button on the cpu block. Ironically I have more power hookup's for fans than I have room for fans in my case.Iketh - Thursday, August 23, 2012 - link
Can't believe this fan was not included.http://www.newegg.com/Product/Product.aspx?Item=N8...
It has winglets designed specifically for static pressure. The bearing it uses is also 100% silent. At max RPM, the fan will get loud though because of the winglets, but completely silent mid-low.
Onus - Thursday, August 23, 2012 - link
So, once again, Crappermaster is caught lying about their products. Amazing.thralloforcus - Thursday, August 23, 2012 - link
I would have loved to see some Noctua, Noiseblocker, and Cougar fans tested. These seems to be pretty popular for radiators. I'm using the stock H100 fans right now in my H100, with two Noiseblocker M12-P fans for pull.Of course I'm always looking for better performance! I was using two Panaflo FBA12G12H1BX fans with the stock fans, but it seemed that the voltages given out by the H100 were too low, and the fans kept shutting off.