Corsair Hydro Series H100i AIO CPU Cooler

Posted: March 13, 2013 in Kits
Tags: , , ,


I have tinker-itus and despite being a sealed kit, I can’t help but wonder what I could do with the parts and pieces if I take them apart.  Could I replace the tubing with my own custom colored UV tubing, and what about adding another radiator and perhaps a GPU or motherboard block into the loop.  Just because it’s sealed and taking it apart will void my warranty, I have to know the what if’s.  That’s what this section is about.

First I did a simple power consumption test.  I took a multi-meter and measure the pumps power consumption and voltage with all ports unplugged and with the LED turned off.  This is what I measured:



So, the actual power consumption of just the pump and electronics is 12.10V x .17A = 2.2 Watts.

Looking at typical DIY watercooling pumps.  A normal D5/DDC variant is typically in the 12-20 watt range, and smaller kit pumps are in the 6-8 watt range, so 2.2 watts is getting down there in power consumption.  That is perfectly fine and probably a great balance for noise and performance in a CPU only loop, but that’s not a lot of pumping power to consider expansion.

Next I did a quick max flow rate test.  Most of the DIY custom water cooling parts out there are designed for those larger pumps and more importantly bleeding air bubbles is something the larger pumps can do for you.  Generally if you can produce a flow rate of 1GPM or better the flow moves fast enough to purge.  You can run much lower flow rates and many do once bled out and via speed control.  I typically haven’t seen much more than about a 2 degree loss going down to .5GPM levels.

Using Stren’s awesome 2012 Waterblock round-up for flow sensitivity reference: CPU block roundup

You can see that performance in CPU blocks does start to fall off rather quickly at .5GPM and this is with the more restrictive and micro designs of CPU blocks.  GPU blocks and radiators are similar or worse because they don’t have quite the micro channels and such.  As you could see in the internal review the H100i is unusually micro and restrictive which helps it perform well at really low flow rates.  I would say generally, for expansion you need .5GPM or better, so I did a quick little test.

First a visual:

I then did three of what I call volume/time tests using the reservoir to draw from and a 1 Gallon Jug to fill.  I do carefully make sure the inlet level and discharge levels remain constant as to not introduce static head errors.  Then as the gallon jug is being filled, I am also adding water to the draw reservoir.  I took the best of three tests and that was 9minutes 5seconds to fill 1 gallon or 0.11GPM.

Unfortunately, it appears even without the radiator in the loop, the remaining flow rate provided by the H100i pump is far too low to consider expansion.  It’s a good design and certainly adequate flow for the unique high density base, but I wouldn’t recommend trying to modify the loop with DIY parts.  Bleeding even the pump/block by itself was a pain, I couldn’t imagine trying to bleed a motherboard block or open GPU block added in.  You would likely be fighting air pockets getting trapped and may have some real trouble getting good performance.  Even replacing the tubing and adding a custom reservoir will invite some bleeding challenges.  My coil of 3/8″ tubing used in this test had an air pocket that wouldn’t clear by itself.

Update 3-19-13

As a follow up to the previous bucket testing, I also connected up the pump to my normal PQ curve test loop.  I suspected the .11GPM wouldn’t show much on my usual King .5-5GPM flow meter and was right, however this did verify the current measurements and it also allowed me to measure the maximum head.

First the Max Flow Test:



While the flow rate is indeed too small to measure on my normal flow meter, the current  on my cheap multimeter checks out ok and I was able to measure maximum pressure head which came it at 1.14PSI.  Previously I read on Frosty Tech’s review here that the H100 specs were 112cm H2O (1.59PSI), but I was only able to measure an actual 1.14PSI.  The maximum head pressure is at zero flow rate, so despite the CPU block base restriction it should not affect or readuce the maximum head.  I would assume the difference is between H100 vs H100i being slightly different pumps.  I do expect the maximum pump flow rate (without the copper base restriction) to measure higher.

3/26 Update

Did a quick pressure drop test on the cpu block/pump base to measure actual restriction.

This was my test setup.  King Instruments 7520 for flow rate and my usual Dwyer manometer to measure pressure drop fed water by my household tap.



The detailed test results


Comparison to typical DIY parts

Normally I would say something like we have a new king of restriction as the block/pump combo is extremely high in restriction.  It is actually about 3.6X more restrictive than my highest restriction block compared there, but restriction is really only something you are concerned with when designing a custom loop for multiple blocks.  In a CPU only loop and particularly for a sealed loop not designed to ever need bleeding, the consequences of the high restriction are completely moot if used as designed.  It is a good design for a sealed CPU loop, the block is very restrictive however thermal performance is really all that matters for sealed use.

I ran the same restriction test on the radiator:


The radiator picture was much the same compared to other DIY radiators:



While radiators normally are extremely low in restriction, that is not the case with the H100i.  It must have very thin tubes to make it perform well at very low flow rates as the restriction is about 6X more than most radiators I’ve tested.

This section was intended to explore the “Expansion and Modification” capabilities and it has become very clear to me that you should not attempt that.  The high restriction block & radiator design, smaller pump, and resulting very low flow rate is not something that will bleed well at all and I would question how well your normal DIY parts would work since they generally are design for much higher flow rates.

I would not recommend expansion or modification to the sealed loop.  Keep your warranty and leave it alone, any expansion will require  lower restriction CPU block and likely more pumping power.  The very low 0.11GPM max flow rate is far too slow to do much expansion with.

Pages: 1 2 3 4 5 6 7 8 9 10

  1. Pedro says:

    Thanks Martin,
    I’ll use your tips. I will cut the tubes to the size I need them to be and then re-assemble and bleed the kit out of the case.
    Do you know that voltage is the pump using? if it is using the 5v rail maybe I can mod a cable to connect it to the 12v to speed it up a litle.. humm I can alreasy smell the smoke..

    Using a reservoir is out of the question as there is no space for one on the case. It’s a mini-itx build with a ncases M1 case.

  2. Hi Martin,

    Hope you had a great Christmas ! And hope you have a safe and joyous New Year also!

    Quick question, have you heard of that Kraken g10, the CPU AIO adapter for a GPU? It looks pretty cool and read some impressive reviews. Just wondering your thoughts on it. Also, do you know someone who sells a heatsink kit for a graphics card to use on the VRam and VRMs? Perhaps an assorted package of small heatsinks you could glue to the chips? I talked to a guy that was going to use 2 of these with H90s to cool 2 780s. He said he would get back to me when he has it up and running.

    Take care

    • Martinm210 says:

      No personal experience, but this review:

      Seems to indicate the VRM need more than the fan and noise is really not better than the stock HSF. I had a GPU only block on an old 8800GTX with passive ramsinks and a 120mm fan which seemed to do ok, but I do think you generally need the little Heatsinks and a fan close by to keep the rest of the card cool. In the end it doesn’t look nearly as nice as a full cover GPU block.

      By the time you buy two AIO units and all the other bits, you could nearly buy lower end DIY watercooling that looks a lot better. might be fun to play and try, but that’s my limited experience take on it so far.

      • Thanks Martin, that is a good write up, liked the thermal images, they really tell a story. I sort of figured the same as you, the NZXT info says the fan will take care of VRAM and the VRMs, but it seems they are not. That is why I asked about a heatsink kit, to let this guy know.

        Your right, after adding up the cost for everything, a full cover would be a lot better, but this guy got most of the stuff free so that is the way he is going. I will drop you a line when he gets back to me about how things went.

        Take care

        • Martinm210 says:

          Nothing wrong with putting together a bunch of free or cheap parts and tinkering, it’s all good!

          • vChris says:

            Hey Martin, first off I love your reviews. Always very thorough and well-controlled testing.

            But I was wondering if you collected any data on the h100i built in fan outputs. I have a watercooling project in the works and I’m wanting to see if I can just plug my NZXT grid fan splitter (30 watt max) into one of the h100i’s 4 available fan headers and use that to control a 2nd radiators fans.

            I couldn’t seem to find any specs on the max fan wattage so I was hoping you tested it.

          • Martinm210 says:

            No i didt’t but…you could probably make or buy one of the power supply fed splitters like the swiftech pwm splitter or one of the other brands that use only the pwm signal from the device and draw power direct from the power supply. The h100i newer models do use 4 pin pwm fans so the output should include PWM signal. You could make your own out of a spare fan molex adapter and a normal 4 pin Y splitter too. You just need to bypass the power and ground and draw them direct from molex or sata power to the PSU.

            I would just recommend testing the H100i output voltage. If power out remains a constant 12v while reducing speeds, then you can be sure it is sending a regular PWM signal. Pretty sure it does, but I never checked. I also had no luck with link on windows 8 so I have no way of testing and bypassed the pump fan controller completely.

            Of coarse you could also bypass the h100i fan controller and run PWM off the motherboard cpu header using speedfan as well on to multiple PWM 4 pin fans.

  3. Kaah says:

    I like your scientific approach and trust your judgement
    I’m owner of the h100 coupled with Noctuas (NF-F12) and confident with that
    I’d like to move my hardware to a new home (namely a Lian Li TYR-X2000(F(N)))
    therefore I’m looking for reviews about Coolermasters Nepton 140XL – ideally in Tandem with 7mm shroud/anti-vibration adapter and Arctic F14 PWM fans …
    any plans to do something along this line? 😉

    • Martinm210 says:

      Sorry, no testing in the near future. I’m in the middle of my buggy building project. I also tend to feel you really need at least a 240 rad to gain the benefits of watercooling. I’ve been able to bring even the 240s to their knees and throttle my 3930 at lower fan speeds. A lot of high end air coolers are fairly competitive I think with many of the 120/140 CLCs. A noctua NH-D14 with a pair of Gentle Typhoon AP 45s running with PWM mods would probably outperform most of the CLCs of similar size and with lower noise.

      • Karsten says:

        thanks for your input *goes back to the drawing board^^*
        New idea:
        Swiftech H320 + Nexxos 180mm monsta rad + EK GPU full cover (GTX 680)