When I first saw the cooling design, I thought to myself, is this really going to cool the components that well? I decided to put the cooling to the test. I had to devise a way to show performance without using the PCH, NF200, or mosfets to heat up the cooling apparatus, so I did the next best thing, I took apart my array of 10W 10omh resistors and used a few to heat up the heatsink. Artificial testing here we come!
know this picture is scary but no worries I am not modding the motherboard, I just felt like scaring you! The motherboard is probably perfect and doesn’t need any voltage modifications. I am using everything in the picture, including three digital temperature probes. They are pretty cheap but do the job well, my thermocouple was down, and so I used these thermal probes instead; they do the job very well. As you can see in the next picture both meters together show the same exact temperature which is ambient temp (it’s hot in here). Each meter has 3 probes attached to it and is wired to use each when switched to the certain probe. I will use only 3 probes for this experiment, as the other 3 are already integrated into my main pc.
Experiment #1 Heat dispersion:
This test shows me how well heat is dispersed from one heatsink to another as well as from the bottom of the heatsink to the top. One temperature probe for the resistor is placed between the resistor and the NF200 heatsink, another temperature probe is placed on top of NF200 heatsink, and then the last probe is placed on the bottom of the PCH heatsink to see how well the heat pipe transfers heat. Then I moved the probe from the top of the NF200 heatsink over to the top of the PCH Heatsink. Two 10 watt resistors are used, they get very hot to the touch and cannot be touched, they are powered by 12v with no limit on amperage, and they are held together by their soldering, then I add an extra bit of thermal past to the lapped surface of the NF200 heatsink, and tie them down with zip ties, the probe is in-between the resistors and HS.
Test #1 results:
As you can see the difference between the bottom of NF200 HS/top of resistor is a few degrees difference, about 3C in the beginning. When the heatsink starts to heat up that difference increases a small amount. The fact that it is within 10C is amazing, most high-end air coolers have a problem keeping within 10 of surface temp of the chip, but I believe because these blocks are one solid piece they are able to transfer heat much more effectively. When we move over to the PCH, which is heated through the heatpipe we can see heat exchange is very good, in the beginning there is a 3-6C difference. Once heat gets up to a point it seems that the vaporization that takes place inside the copper heatpipe transfers heat much more effectively when heated. We see the 6C difference decrease to only a few degrees, this indicated that the clamp and connection between aluminum block and copper heatpipe was done correctly. In the final moments of the test, I moved the probe from the top of the NF200 heatsink to the top of the PCH heatsink. While the heat is less on the PCH, the transfer of heat is about the same as from bottom of the NF200 HS to the top of the NF200 heatsink, this is most likely due to the fact that both are solid blocks and specific heat is the same for both blocks, so heat transfer would be about the same which we see is the case.
Experiment #2: Heat Transfer. In this experiment I will add 3 more 10 watt resistors to the mix, and zip-tie them down to the mosfet heatsinks. I will use the stock thermal pad that is already there, because I want to test the system as is. In the past GIGABYTE used very thin heat pads, as is the case here. One thermometer is placed on the bottom of the NB block as usual, then one is placed on the PCH HS on top, and the third is placed on top of the mosfets HS on top. In the beginning I will let the system heat up to 50C (These resistors go up to 70C when on the original HS I had them on before testing), and then place a fan (silent 1500 rpm fan), to mimic case airflow. Please do understand that I am generating 50 watts, the NF200 is supposedly rated at 12W and the driver mosfets will be about 1 watt (at 5 amps a piece(24x5=120amps) which is more than enough) each under load conditions with all 24 activated, so that is 36 watts + PCH which I do not know. This 50 watt array of resistors is to mimic a heavily overclocked system, please take into consideration that not all 50 watts will be concentrated in the areas which they are in this test.
Test #2 results:
