Table of Contents
Up for review today is Corsair’s newest entry into the Low Cost Liquid Cooling (LCLC) category, the Hydro Series H70. This is the second offering from Corsair in the LCLC category, being preceded by the Hydro Series H50, which was reviewed twice by us here at Overclockers.com; first by hokiealumnus shortly after it was introduced right here and then by myself when I compared it to high end air cooling. What I found in that review was that while the H50 was a great first effort by Corsair, it still lacked a little in performance compared to the best in high end air cooling. In particular, I found that the fan choice for their cooling solution to be the biggest let down on the H50, with it not having enough static pressure or CFM to efficiently remove the heat of an overclocked LGA1366 system. Substituting different fans and setting it up with a push-pull fan configuration helped a lot, but it still lacked just a little bit to meet or beat the best air coolers.
Now, we have Corsair’s answer to the shortcomings of their original LCLC offering, which comes with a thicker radiator for more area to move heat from it and they have given this new solution a true push-pull dual fan setup for more airflow through the new, thicker radiator. We will see if these changes bring the performance of this cooling solution to where it matches or even beats high end air on a 200 watt heat load. I also would like to thank Corsair for supplying this cooling solution for review.
Specifications (Courtesy Corsair)
- Pre-filled, closed-loop system is easy to install
- Copper CPU cooling plate for maximum cooling performance
- Integrated pump and reservoir is sealed for zero maintenance and improved leakage protection
- Large, double-thick 120 mm radiator for outstanding heat dissipation
- High-performance, twin 120 mm fans offer outrageous airflow in push-pull configuration
- Included step-down adapters let you customize your fan speed for low noise or high performance.
- Two-year warranty
- Model – CHWH70
- Cold plate material – Copper
- Radiator specifications – 120 mm X 50 mm
- Fan dimensions – 120 mm X 25 mm per fan
- Fan specifications – 1600 or 2000 rpm, Air Flow 50.35 – 61.2 cfm (per fan), Static Pressure 1.8 – 2.2 mm H2O (per fan), Noise level 26 – 31.5 dBA (per fan)
- Radiator Material – Aluminum
- Tubing – Low-permeability for near-zero evaporation
I also asked Corsair for some other information that isn’t listed on their website:
- Akasa is the OEM of the fans and the fans draw around 2.4 watts of power each, in case you are worried about overloading a motherboard fan header while using the two into one fan power adapter on a mobo fan header.
- The Thermal Interface Material applied to the cold plate of the waterblock/pump assembly is made by Shin-Etsu, just like with the H50.
- The OEM of the radiator and waterblock/pump assembly is Asetek once again, but the design is Corsair’s. Asetek originally didn’t have a double thick radiator solution.
- Like the H50, this system uses deionized water with propylene glycol, to prevent corrosion, as its cooling fluid.
The box the H70 comes in is sized at 265 mm X 221 mm X 137 mm. The box is shorter, but wider and taller than the H50 box. Like the H50 packaging, the box this unit comes in is sturdily constructed and well-packed to cushion it through all the bumps of shipping the unit to the store. I will let the pictures posted below show all of the marketing literature printed on the box.
Upon removing the radiator/pump assembly from the box, I was struck by how much more this assembly weighs as compared to the H50. Also immediately noticeable is that the pump/waterblock assembly is drastically shorter than the unit used on the H50. This should make for an even tidier install with potentially better cooling for the motherboard components, since the pump assembly doesn’t protrude as far, blocking airflow around the socket area. It will also give more clearance for the radiator/fan assembly, since this assembly is around 4 inches (101 mm) long when using the stock fans and is around 5 inches (127 mm) long with 38 mm thick fans installed on the radiator.
Additionally, Corsair drastically shortened the hoses from the pump/waterblock to the radiator. It is likely that Corsair figured they could shorten the hoses since this unit was designed to run in a push/pull configuration with a radiator twice as thick or something of that nature. The H50 has hoses of around 312 mm long whereas this unit’s hoses are around 230 mm long. This could potentially cause a problem in some cases where the 120 mm exhaust fan cutout is far away from the cpu area.
The radiator itself measured out at 120 X 150 X 48 mm and the weight of the radiator/pump assembly is 723.5 grams. There is also about a 4 mm plenum area between the top of the fan mounts and the radiator fins themselves, which should help in dispersing the air across the radiator. The fin count is just about as dense as the H50 radiator at around 19-20 fins per inch. The fans themselves are Corsair labeled and weigh approximately 116 grams each. Another change that Corsair made to this unit as compared to the H50 is in the electrical cabling. On the H50 they used conventional sleeved cabling, but this unit uses cabling that looks and feels much like a miniature version of the FlexForce cabling that they use with their HX series of power supplies. It looks decent and isn’t hard to work with, so I have no problem with this switch in cabling.
The mounting system is the exact same as they used in the H50 cooler. This isn’t bad by any means as this mounting system seems to be very good, was easy to work with, and gives good repeatability in mounting the pump/waterblock to the CPU. Corsair probably figured there were better ways to spend their R & D dollar. Why try to fix something that isn’t broken anyways? For controlling the fan speed, Corsair went with some inline resistors that you can install in between the fans and the motherboard header. They also include a handy two to one fan adapter, which will let you feed both fans from just one motherboard header. While this is very handy and it should be safe to use with the fans included with this unit, I would recommend that if you upgrade the fans to something with a higher power draw you will want to power each fan independently so that you don’t burn out a motherboard fan header. Another neat feature of the fan adapter cable is that it passes through only one fan’s rpm sensor, so that the motherboard isn’t confused about sensor pulses from two different fans.
Installation of the H70 was basically the same as with the H50, except that due to the shorter tubing I had to mount the Swiftech Radbox I use to mount the radiator much closer to the case because of the lack of a 120 mm exhaust hole on the old case. I was able to mount it on the Radbox with no problems though and had enough room to even test with 38 mm thick fans. But another 25-30 mm more tubing sure would have been welcome. The pump/waterblock assembly mounted on the CPU exactly like the H50 mounted its pump/waterblock assembly. That was totally expected, since they both use the exact same mounting system. I also love the fact that Corsair gives you all the mounting equipment to mount this cooling system on all current single processor desktop systems with the unit.
Fans used for testing were as follows:
|Fan(s)||Size||RPM||cfm||dBA||H2O Static Pressure||Wattage||Weight|
|2x Stock Corsair H70||120 X 25 mm PWM fan||1600 & 2000 rpm (1600 rpm using inline resistors)||50.35-61.2 cfm||26-31.5 dBA||.0709-.0866 inches||2.4 watts||116 g each|
|2x Scythe S-Flex SFF21G||120 X 25 mm||1900 RPM||75 cfm||35 dBA||N/A||2.88 watts||182 g each|
|2x Scythe S-Flex SFF21F||120 X 25 mm||1600 RPM||63.7 cfm||28 dBA||N/A||2.4 watts||178 g each|
|2x NMB-Mat (Panaflo) FBA12G12L-1BX||120 X 38 mm||1700 RPM||68.9 cfm||30 dBA||.130 inches||2.88 watts||255 g each|
|2x Sanyo Denki San Ace 109R1212H1011||120 X 38 mm||2600 RPM||102.4 cfm||39 dBA||.26 inches||6.24 watts||243 g each|
|1x Sanyo Denki SanAce 9CR1212P0G03||120 X 76 mm (for extreme CFM testing)||6200/2700 RPM (compound fan)||300 cfm||70 dBA||1.93 inches||86.4 watts||753 g (with fan guard installed)|
The testbed system consists of the following components:
- Case – Chieftech clone of the Antec 1040 case series, with the original 80 mm exhaust fans being removed and the holes enlarged to mount two 92 mm fans externally on the outside of the rear case bulkhead instead of internally. For testing the H70, I had to remove one of the exhaust fans and mount a Swiftech Radbox externally to mount the H70 radiator/fan components on because this old case doesn’t have a 120 mm sized exhaust fan. This worked out well and while Corsair shortened the hoses up significantly, I was still able to mount the radiator and fans externally on the Radbox. The waterblock/pump unit fits through the hole left by removing one of the exhaust fans with no problems and the tubing to the radiator is long enough to mount this way and is run through that hole also.
- Motherboard – Asus P6T
- Processor – Intel Core i7 Extreme 980X, overclocked to 4000 MHz @ 1.304 v.
- RAM – Corsair XMS3 DDR3 1600
- Video Card – eVGA 7900GTX
- Power Supply – HEC Cougar series (German HEC, not US model) S700
- Hard Drive – Western Digital Caviar 250 GB SATA hard drive
- Optical Drive – Lite On DVD-RW drive
- OS – Windows Vista Ultimate 64 Service Pack 2
- The initial mount runs were made with the pre-applied TIM that this unit came with. For the other two remounts, Arctic Cooling MX2 thermal paste was used for testing as I have found it to give good consistent results with no appreciable break in and it applies and is cleaned up easily. This also gave me a good baseline to see how effective the pre-applied TIM was compared to a known good performance thermal paste. After running the remounts and comparing the results I found no discernible difference given the test environment.
Test Methodology and Results
The testing methodology used is the same as I used with my previous reviews of the six high-end heatsinks and the H50. All energy saving features of the motherboard and processor were turned off to keep it from down clocking the processor speed and vcore. All fan control functions were turned off in bios to keep the fan at maximum speed. For processor temperature monitoring purposes, I am using Real Temp 3.40, with logging enabled at 2 second intervals.
Temps in my computer room were maintained between 20.5 to 21.1 °C (69-70 °F), measured at the front of the case. If room temperatures exceeded these parameters I re-ran that individual test run.
For loading the CPU, I used Prime95 version 25.8 using in-place large FFT’s and ran it for 30 minutes to stabilize temps. After 30 minutes I would exit Prime95 and let the CPU idle for 5-10 minutes. The highest recorded temperature from the hottest core for each run was then recorded off of the Real Temp log, the lowest temperature on any core was recorded and the average temperature on the hottest core was calculated during the load portion of each run.
Each fan configuration was tested with three remounts of the pump/waterblock and the lowest average temperature run recorded, to minimize any problems between mount to mount installations. One difference between this review and my tests with the H50 is that I didn’t even attempt to test with just one fan installed. This cooling system was designed to use two fans and that is how I tested, with one exception. When testing in the extreme performance realm with the SanAce 9CR1212P0G03, which has enough static pressure and cfm to run on any radiator efficiently with just one fan installed. The SanAce 9CR1212P0G03 was installed in a pull configuration, just like my tests with it on the H50.
This first chart gives the results with the various fan combinations I tested:
As seen in the chart, it isn’t worth upgrading the fans with an equivalent 120 X 25 mm fans such as the S Flex F fans. There was a noticeable gain going with the Panaflo and Sanyo Denki fans though, due to their higher static pressure.
This second chart compares all the heatsinks and LCLC cooling systems that are delivered as an out-of-the-box cooling solution that I have tested on this platform. These results might get your attention too.
As this chart shows, the Hydro Series H70 is fully competitive with the best of high end air heatsinks that come as an out-of-the-box solution. And it absolutely trounces its older brother, the H50 too.
As a followup to the Hydro series H50, the H70 addresses the weaknesses of their first LCLC offering very well. They changed the fan to a better performing one and doubled the fan count to two, so that the radiator can run with a push-pull fan setup for better performance. Along with the upgraded fan setup, the H70 features a better radiator with much more cooling fin volume to better shed the heat. These changes have brought the performance of the H70 cooling system squarely in competition with the best of high end air cooling. And my fan substitution tests also show that the radiator’s ability to get rid of heat isn’t maxed out with the included fans, as performance steadily improved with the substitution of higher cfm, higher static pressure fans.
The only problem (if you could call it a problem) is that Corsair increased the MSRP on this product to over $100. And that makes it significantly more expensive than the two high end air cooling solutions that I consider to perform as equal to the H70. The cheapest I can find the Hydro series H70 is $109.99 with free shipping from Amazon.com. I found the Noctua NH-D14 from Provantage selling for $77.26 and around $10 shipping and you can find the TRUE with a Scythe SFF21F fan (plus an extra set of fan clips too and functionally the same as the TRUE 1366RT which looks like it’s being phased out) for $56.99 plus around $11 shipping from CrazyPC. This puts the Hydro Series H70 at a bit of price disadvantage compared to those two heatsinks for comparable performance.
Other than the price issue though, the out-of-the-box performance is there and its cooling capabilities can be enhanced with the simple substitution of higher performance fans on the radiator. Another good feature of this system is that the low profile waterblock/pump assembly will fit inside narrower cases that might have problems with a tall tower style cooler and it has no interference or clearance problems with ram slots that might be placed close to the CPU socket on some motherboards. It also comes with mounting solutions for all current single processor sockets on the market.
- Performance fully competitive with high end air cooling solutions.
- All current AMD and Intel mounting equipment included with the package.
- Easy to upgrade performance by substituting higher performance fans.
- Keeps ram slots totally clear and eliminating any clearance issues with your ram.
- CPU socket area kept pretty clear, which should help with motherboard component cooling by not impeding airflow around the socket area.
The Not So Good:
- Shortened the hoses between the waterblock/pump assembly and the radiator. This might give some people problems mounting the unit if their 120 mm fan cutout in the case is farther away from the cpu area than normal.
- The included pre-applied TIM is great stuff and works well, but is single use only. I would much prefer they include a small tube of the same thermal material instead so that the user can get a few mounts on his system without having to buy some more thermal paste. Better yet would be to keep the pre-applied TIM on the waterblock/pump assembly and include a small tube of paste. For the price they are selling this at, I can’t imagine that adding a quarter’s worth of TIM in a small tube will kill their profitability.
- Pricing on this unit is higher than comparable high end air solutions. It’s priced around $22 higher shipped compared to the Noctua NH-D14 and over $40 more expensive than a TRUE with an SFF21F fan delivered to the door.
If the price differential doesn’t bother you, I heartily recommend the Corsair Hydro series H70 cooling solution for your high end cooling needs. Basically, if you need better performance than this system you will need to go with a true custom watercooling loop costing several times this system’s cost. Again, I would like to thank Corsair for sending this unit to us to test.