Testing the SilverStone HE02 Fanless CPU Cooler

The SilverStone Heligon HE02 fanless CPU cooler is designed for Intel or AMD processors, and is made of a copper base, six heatpipes, and 30 aluminum fins.  It is 160 mm tall, and has a mass of nearly 1 kg.  The cooler has an asymmetric design that can allow it to fit into small cases, avoid blocking the first PCIe slot on the motherboard, and/or avoid interfering with tall memory heatspreaders.  It can be fitted with 120 mm fans, although since I build Fully Silent PCs, I will not be adding any fans to it.  SilverStone claims that the HE02 can cool up to 95 watts fanlessly, though I will be testing that claim here.  The real-world cooling capacity will depend on several different things, including the case chosen, the contact between the CPU die and the heatspreader, the contact between the CPU heatspreader and cooler, and the ambient temperature in the room it is placed in.

I paired it with the SilverStone Lucid LD03, which has a sideways motherboard layout and plenty of openings to promote unrestricted natural passive airflow.  The fans were removed from the case, as well as any dust filters that might have restricted airflow.  My own testing showed the LD03 to perform in this way similar to an open-air setup, which is as good as it gets for passive cooling.  For testing, I also chose to use the Gigabyte H370N mini-ITX motherboard, Intel i9-9900 and i3-9100 processors, 16 GB of T-Force Vulcan Z memory from Team Group, a 1 TB Crucial P1 NVMe solid state drive, and a SilverStone NightJar NJ450 fanless power supply.  The build process was documented in the video below.  The ambient room temperature for these tests was about 70 degrees Fahrenheit or 22 degrees Celsius.  Thermal Grizzly Kryonaut thermal paste was used between the CPU and cooler.



My test procedure included running Prime95’s torture test with small FFTs for half an hour while recording the core temperatures and core power use with HWiNFO64.  I did this a total of 14 times, waiting for the CPU to completely cool between iterations.  I started with all CPU power limits set to 10 watts in Gigabyte’s UEFI, and increased power limits each time by 10 watts until I ran into some kind of thermal throttling, for both the i9 and i3 CPUs.  I did not have any AMD processors on hand to test, and AMD’s UEFI settings do not allow for as much control of power limits, but results should be applicable to AMD CPUs nevertheless.



I plotted the average core temperature against time.  Starting with the i9 CPU, the 30 minute CPU temperatures went up an average of 8.7 degrees Celsius for each 10 watt increase in power use.  These power values for each curve are the average actual power use during the tests, not the power limits set in UEFI.  Using 80 watt power limits, at the 15 minute mark the motherboard’s VRM MOSFETs actually began to thermally throttle, though a couple of the cores were very close to throttling as well, so I knew that this was above the cooler’s limit and I stopped the i9 test here.



Moving on to the i3, the 30 minute CPU temperatures went up an average of almost exactly 10 degrees Celsius for each 10 W increase in power use.  Something interesting happened here; two of the i3’s cores were reading 5 to 8 degrees hotter than the other two, and one of the cores began throttling at 21 minutes during the 60 watt power limit test.  I kept the test going as the throttling was not severe, and two of the cores never throttled even after the 30 minute mark.  So the i9 made it to the 80 watt test, while the i3 only made it to the 60 watt test.



This was unexpected, although I believe I know what caused this difference.  Intel’s i9 CPUs have solder between the die and heatspreader, while the i3 CPUs have thermal paste there instead.  Solder is much more thermally conductive then any thermal paste.  Most of Intel’s 9th generation i5, i7, and i9 CPUs include solder, but their i3, Pentium, and Celeron CPUs still include thermal paste.  You might think that a cooler that is able to cool an i9-9900 at 70 watts would have no problem with an i3 at 60 watts, but that turned out not be the case.  We can blame Intel for that, and not SilverStone or any other cooler company.  I’ve heard it said that certain lower performance coolers should only be used with less expensive CPUs, but what really matters is the power use and the die to heatspreader interface.  The optimal CPUs for this cooler wouldn’t actually be an i3, but a soldered i5 or i7 CPU.



I then decided to take a more scientific approach to finding the true passive limits of the HE02, so I plotted the 30 minute average core temperatures against the sustained power use during each test, and calculated the formulas for a linear trendline for each CPU.  For the i3-9100 the expected stress temperature in Celsius is 0.98 times the power use plus 38.  For the i9-9900, and I assume other soldered CPUs as well, the expected stress temperature is 0.85 times the power use plus 33.  The R squared value for both trendlines was above .99.  If we target a maximum stressed temperature of 100 degrees, we get a power target of 63 W for the i3, or 79 watts for the i9.



If you are concerned about performance loss while applying a CPU power limit, I also performed PassMark’s CPU Mark performance test after each iteration of Prime95.  For the i3 at a safe 50 watts, the CPU Mark was 9,149, and for the i9 at 70 watts, the CPU Mark was 16,442.  For comparison, I plotted the CPU Mark scores against the power limits applied.  The i3’s performance flat-lined after 40 watts, but the i9’s performance kept increasing and most likely would continue to increase with higher power limits.  That means that you won’t get the most out of an i9-9900 with power limited applied as appropriate for passive cooling with the HE02.  Again, an i5 or i7 would be more appropriate in this configuration.



So, is SilverStone’s recommendation of 95 watts too high?  Yes, in my opinion, CPU TDP should not be higher than 65 watts when using this cooler passively.  Yes, you can and should use power limits when available, but there would be little reason to use a 95 watt CPU and limit power to 80 watts or less.  For the average CPU, I would put a reasonable limit at 70 watts, right between the i3 result and i9 result.  If you use a different case that is not as passive cooling friendly as the LD03, the recommended power limit might be even lower.



That said, there are not a lot of options for effective stand-alone passive CPU coolers, and the SilverStone HE02 is actually the best fanless cooler I’ve tested that is still in production.  The now discontinued NoFan CR-95C is even better, but it is not easy to find.  As of November 2019, the CR-95C is still available at Quiet PC USA, or here at Fully Silent PCs as part of a complete custom fanless system, while supplies last.  The HE02 on the other hand has wide long term availability.


Leave a Reply