For knowledge, news, and more
Limitations of Air Coolers: Heat Pipe Dry-out Explained
January 17, 2020
“Thanks to AMD packing a huge number of processor cores into their recent consumer CPUs, we are currently undergoing a big leap in the performance of home computers, and consequently, we’re also undergoing a huge leap in the cooling requirements of the aforementioned.”
Linus Tech Tips, December 14, 2019
For CPUs that dissipate large thermal loads, the vote amongst experts is unanimous: you must use water cooling, i.e., you need an all-in-one (AIO) liquid cooling system, especially if you plan to overclock. However, many users don’t want to put water anywhere near their expensive CPU and motherboard due to the inherent risks of leaking/failing, which can permanently damage the electronics.
Until now, the only cooling alternative is using an air cooler. But, as the title of this article suggests, air coolers have a physical thermal limitation that prevents them from maintaining cool enough die temperatures at high thermal loads. Read on to learn about this limitation and how ProSiphon technology overcomes this thermal roadblock with the added advantage of being a no-pump/no-water cooling solution.
To understand the air cooler limitation, you first must understand how air coolers work. Air coolers are made up of a number of heat pipes. Heat pipes are passive heat transfer devices- passive because heat pipes do not require a pump that requires external power (whereas an AIO uses a pump to circulate water in the cooling loop). Figure 1 shows an example of an air cooler (the Noctua NH-U14S, which is an excellent air cooler) and points out where the heat pipes are located.
Figure 1 Example of an air cooler, which contains 6 heat pipes. The evaporator is closest to the CPU and is where the fluid within the heat pipes boils and evaporates. The vapor then moves to the condenser, which is cooled by air flowing over the fins and condenses back into a liquid. The liquid returns to the evaporator via a capillary wick that lines the heat pipes, and the cooling loop begins again.
Heat pipes are excellent heat transfer devices because their heat transfer mechanism relies primarily on evaporative cooling, i.e., a phase change from liquid to gas. The heat from the CPU is absorbed by the liquid in the heat pipe, and subsequently, the liquid boils and evaporates (hence, this area is called the evaporator). This vapor travels through the heat pipe and is cooled, which is aided by the airflow over the fins provided by fans. Subsequently, the vapor cools and condenses back into liquid (hence, this area is called the condenser). But how does the liquid return to the CPU area to continue the cooling loop?
Heat pipes are lined with a wicking material that relies on capillary forces to return the liquid back to the evaporator- the same mechanism occurs when you spill liquid on your shirt; the liquid is wicked away via the shirt material. Thus, the liquid from the condenser can return to the evaporator and begin the cooling loop again.
Okay, so now that we understand the basics of how heat pipes work…what is this thermal limitation of air coolers? When the thermal load reaches a certain point, so much vapor is created that it begins to flood the wicking material, which prevents the liquid from returning to the evaporator. This phenomenon is termed dry-out because there is no longer liquid in the evaporator.
No liquid means no evaporation, which means no cooling. This is a bad, bad situation- with no liquid in the evaporator, there is no way to remove heat, and the die temperature of the CPU will dramatically increase.
It’s easiest to “see” this dry-out phenomenon using a thermal test vehicle (TTV) setup. Simply, a TTV is a way to maintain a constant thermal load and measure temperatures in a repeatable, controlled manner.
A TTV is made up of a copper heater block with cartridge heaters that act as the CPU thermal load. The total thermal load can be accurately determined by measuring the voltage and current supplied to the cartridge heaters. Thermocouples are used to measure the temperature on the surface of the heater block (analogous to the die temperature). Figure 2 shows an image of IceGiant’s TTV (TR4 socket, AMD Threadripper CPUs)
Figure 2 IceGiant’s Thermal Test Vehicle (TTV) for Threadripper. This Threadripper TTV is used to compare cooling solutions in a controlled, repeatable manner. Cartridge heaters are used to mimic the thermal load from a CPU, and 6 thermocouples are placed in the grooves on the copper block (analogous to die temperature); the maximum temperature recorded is then used in the thermal resistance calculation. Note that the setup is allowed to reach steady-state before recording the measurements. Ambient temperature was recorded as the average of 4 thermocouples placed at each cooler’s inlet fan. The TTV was mounted vertically to mimic the actual use condition in a PC case.
Thermal engineers calculate thermal resistance to compare cooling solutions (to account for ambient temperature and thermal load differences), which is defined as
Figure 3 compares the thermal performances of the IceGiant ProSiphon Elite Prototype and two of the top air coolers for AMD's Threadripper CPUs, the Noctua NH-U14S and the Cooler Master Wraith Ripper. Note that this TTV mimics the size and die heat flux of a Threadripper CPU (TR4 socket). For all coolers, their maximum fan speeds were used.
Figure 3 TTV results that compare the IceGiant Elite and two air coolers (Noctua NH-U14S and the Cooler Master Wraith Ripper). The results show that the thermal performance of both air coolers declines after reaching ~200-250 W (due to dry-out), whereas the IceGiant Elite maintains a high thermal performance even as the thermal load increases. Note, this TTV mimics AMD's Threadripper CPUs.
As you can see in Figure 3, the thermal performances of both air coolers begin to decline after reaching ~200-250 W (their thermal resistance increases), whereas the IceGiant Elite maintains its high thermal performance.
To get a physical sense of the performance difference in terms of die temperature, you simply multiply the thermal resistance by the power. For example, at 300 W, the thermal resistance of the IceGiant Elite is ~0.055 ⁰C/W.
Multiplying 0.055 ⁰C/W by 300 W, you get 16.5 ⁰C- this number is the difference between the max. die temperature and the ambient temperature.
You can do a similar calculation for the Noctua air cooler- its thermal resistance at 300 W is 0.077 ⁰C/W- multiplying this value by 300 W, you obtain 23.1 ⁰C.
Thus, the performance difference between the IceGiant Elite and the Noctua at 300 W is 23.1 ⁰C – 16.5 ⁰C = 6.6 ⁰C. From the graph, you can clearly see this performance difference only increases as thermal load increases.
Hopefully, from this article, you’ve learned how heat pipes work and understand why heat pipes become limited at high heat loads. Based on TTV measurements, it is clear that ProSiphon technology overcomes the physical limitation of air coolers and is able to maintain its high cooling performance at higher loads.
If you use high-end CPUs, overclock, do not want water in your case, and/or value high reliability, the IceGiant Elite with its ProSiphon technology is a great choice.