Is it better to use copper or aluminum heatsink?

Is it better to use copper or aluminum heatsink?

Why do CPUs need coolers?

What’s the relationship between heat sink material and CPU temperature?

What is thermal resistance?

What’s thermal conductivity ( represents the ability to transfer temperature}?

The common CPU coolers in today’s PCs, whether mainstream or high-end products, almost all adopt this structure: the structure that contacts the CPU/GPU surface is made of copper, which is what we often call the copper base, while the heat pipe The fins or the water cooling row fins are made of aluminum.

The so-called “copper-aluminum combination” refers to such a structure, and the radiator using copper fins or copper water-cooling row, which is what we commonly call “pure copper radiator” can be said to be short-lived in the PC field. Appeared in earlier years.

So why can “copper-aluminum combination” become the mainstream design of PC cooling?

If you search on the Internet, the most common statement is that “the combination of copper and aluminum has the best balance after synthesizing various factors”, which can be simply understood as considering the volume, weight, process, cost, heat dissipation efficiency, etc.

The comprehensive choice after the aspect, there are even claims that the heat dissipation efficiency of aluminum is actually better than that of copper, and copper just transfers heat faster.

Therefore, the combination of copper and aluminum combines the advantages of the two, and the heat dissipation efficiency is higher than that of pure copper structure. will be higher.

These claims all seem plausible, but is that the truth?

The material of the cooling fins has no effect on the heat dissipation

In fact, the Newton cooling formula Q=hAΔT in thermodynamics has already given the answer.

The Newton cooling formula is mostly used for the calculation of convective heat transfer, and the heat dissipation of the CPU is essentially to transfer the heat of the CPU to the air through the radiator.

It does not generate heat itself, nor can it eliminate heat, that is to say, the “heat dissipation” that the radiator needs to undertake is the “heat generation” of the CPU, or the total heat flow.

The energy used to drive the CPU work will basically be converted into heat in the end, so as long as the CPU power consumption remains unchanged, the total heat flow will not change, and the total heat flow is Newton cooling. Q in the formula;

A in the formula is the contact area between the radiator and the air.

As long as the structure and size of the radiator remain unchanged, the heat exchange area will not change, and has nothing to do with the material of the radiator;

h is The heat transfer coefficient of the fluid is basically a fixed value as long as the convection mode and gas type remain unchanged.

Therefore, for convection heat transfer, if the heat is unchanged, the air parameters are unchanged, and the structure design of the radiator is unchanged, then its ΔT should also be unchanged.

ΔT is the temperature difference between the air and the cooling fins.

The constant air temperature means that the fin temperature remains unchanged, and it has nothing to do with the fin material.

In other words, the material used for the radiator has no effect on the total heat flow, that is, the heat dissipation, and the temperature on the radiator will not change.

However, in actual use, heat sinks with the same size and structure but different materials do have a significant impact on the operating temperature of the CPU, especially the full load temperature, which seems to contradict the results shown by the Newton cooling formula.

In fact, this is a misunderstanding that many students have fallen into, which confuses temperature control with heat dissipation.

So why is it that temperature control and heat dissipation are not the same thing? This starts with “why does the CPU need a heat sink”.

Why do CPUs need coolers?

Why do CPUs on PCs need coolers?

In fact, this can also be explained by Newton’s cooling formula Q=hAΔT.

If the CPU directly exchanges heat between the core and the air by convection, then with the core area and the limit on ΔT (the upper limit of the CPU operating temperature), we can quickly calculate the CPU heat dissipation requirement, which is equivalent to the heat transfer rate or the upper limit of the total heat flow.

If the heat generated by the CPU at this time does not exceed this upper limit, then there is no need to install an additional radiator, and it is sufficient to directly convect the air to dissipate heat;

but if the actual power consumption of the CPU will be higher than the upper limit, in order to increase the total heat flow, Basically, there are only two ways, either to increase the temperature of itself in exchange for a higher ΔT, or to increase the heat exchange area A of itself.

Both methods can increase the total heat flow Q to a level that matches the actual power consumption of the CPU. .

Now the top cover of the CPU is actually a small-scale heat sink (sourece: overclocking.Guide)

Let us give an example. When the power consumption of a CPU with a core area of 200mm 2 is 200W, the operating temperature is 25°C, and the air with a convection heat transfer coefficient of 200W/m 2 ·K is used for direct heat dissipation, then its operating temperature will be How much is it?

We put the corresponding value into Newton’s cooling formula to get its working temperature.

However, if we calculate under this condition, we will find that its operating temperature will reach the level of 5000 ℃, and no CPU can withstand such a working temperature.

However, when we installed a radiator for it to make its contact area with the air equivalent to 50000mm 2 , the calculated operating temperature of the CPU is only 45°C in an ideal state, which is obviously much more reasonable, which is why now The reason why the CPU on the PC platform needs to be equipped with a radiator.

The relationship between heat sink material and CPU temperature

But so far, we still have not been able to explain, asking what different materials of heat sinks will bring different operating temperatures to the CPU.

In fact, in the last link, we only ideally integrated the CPU and the heat sink for calculation.

The calculated 45°C is actually just the temperature of the contact surface between the radiator and the air, not the CPU core temperature in the true sense.

In fact, the entire cooling process of the CPU is the process in which the core heat is transmitted to the air through the radiator.

There are actually two cooling systems, one is composed of the radiator and the CPU, the other is composed of the radiator and the air, the latter The Newton cooling formula can be used directly for fast calculation, while the former requires formulas for heat conduction, heat diffusion, etc. for calculation, which involves the basis of thermodynamics and heat transfer, as well as the material of the radiator.

Heat transfer is a science based on experiments.

After a lot of experiments, it is found that if the object is a regular shape, such as a cylinder, then the heat transfer rate of the object and the temperature difference between the two ends ΔT=T 1 -T 2 is proportional to the cross-sectional area A of the object, and inversely proportional to the length L of the object, and objects of different materials will follow the above rules, so we can introduce a coefficient λ for objects of different materials, thus obtaining Qx=λAΔT /L formula, and this coefficient λ is the thermal conductivity often mentioned in the field of heat dissipation.

Therefore, if we regard the CPU heat sink as a regular object, the temperature of its contact surface with the CPU is T 1 , and the temperature of the contact surface with the air is T 2 , then it is not difficult for us to see that when Qx total heat flow, A cross-sectional area When , L length, and T 2 temperature remain unchanged, if the λ thermal conductivity is higher, the ΔT will be lower, and the T 1 temperature will be lower, and vice versa.

The thermal conductivity of copper is 401W/(m·Λ), while the thermal conductivity of aluminum is 238W/(m·Λ), the former is 1.7 times that of the latter, and the corresponding temperature difference is 1.7℃ different in value.

On the heatsink, the temperature of the interface between the CPU and the heatsink is lower. In the same way, we can also use this value to further calculate the core temperature of the CPU.

It can also be found that when we use a copper radiator, the core temperature of the CPU will be lower, but the value is not too big compared to using aluminum. difference.

What is thermal resistance?

Of course, this also makes a theoretical calculation, and even so, it is not a convenient thing to directly calculate the working temperature of the CPU and compare the influence of the copper-aluminum radiator on the entire cooling system.

Therefore, we may wish to rewrite the formula of heat transfer rate, Qx=ΔT/(L/λA)=(T 1 -T 2 )/R.

At this point, we compare it with Ohm’s law in the circuit, Ohm’s law I=(U 1 -U 2 )/Re, do we find that the two are so similar?

In fact, the process of thermal diffusion is similar to the process of charge diffusion.

They both occur under the action of potential difference, and in the process of diffusion, there will be resistance.

If the resistance of charge diffusion is resistance, then the heat diffusion Resistance is naturally thermal resistance.

According to the way we rewritten the heat transfer rate before, we can quickly obtain that the thermal resistance in conduction, convection, and radiation are the following three formulas, respectively.

R conduction = L / (λ·A)

R convection = 1/(h · A)

R radiation = 1 / (hr · A)

With the concept of thermal resistance, the analysis and calculation of the heat transfer rate is much simpler, and the heat transfer process can be transformed into a simple series-parallel circuit structure, which is generally called a thermal circuit or a thermal network.

For example, the heat dissipation process of the CPU can simply be understood as a thermal circuit similar to a series circuit.

Of course, as we mentioned before, what we are analyzing here is more of an idealized model.

In reality, the heat of the CPU will not only be conducted into the air through the radiator, but also through the CPU substrate, base, and motherboard PCB.

In addition, there are other heat dissipation media such as silicone grease in the entire heat dissipation system , so the heat circuit of CPU heat dissipation is actually more complicated, and it should be a circuit that coexists in series and parallel.

Here we have made an idealized model for easy understanding. Interested students can further study and understand by themselves.

Through the above heat circuit, we can see that the heat of the CPU comes from the core, and is conducted to the radiator through the top cover, and then the radiator and the air conduct convection heat exchange.

If the internal thermal resistance of the core and the contact thermal resistance between the core and the top cover, the top cover and the radiator are not considered, then the thermal resistance in the entire thermal circuit is composed of the thermal resistance of the top cover, the radiator and the air. This leads to the following formula.

Qx = (T 1 -T 2 ) / (R top cover + R radiator + R air )

In this formula, the thermal resistance between the top cover and the radiator can be calculated using the thermal resistance calculation formula of conduction heat transfer, while the air can use the thermal resistance calculation formula of convection heat transfer, T1 is the CPU temperature, T2 is the temperature of the air.

Therefore, if we only change the material of the radiator, for example, from an aluminum radiator to a copper radiator, without changing its size structure, that is, the volume remains unchanged, then the higher the thermal conductivity of the material used for the radiator, the better its performance.

The resulting thermal resistance will be lower. Since the thermal resistance between the top cover and the air is also constant, the CPU core temperature can be calculated according to the following formula:

T 1 = T 2 + Qx · (R top cover + R radiator + R air )

Therefore, the higher the thermal resistance of the heat sink, the higher the operating temperature of the CPU, so that there is enough temperature difference between it and the air to compensate for the impact of the higher thermal resistance. Therefore, when we test the radiator, we essentially measure the thermal resistance.

In order to accurately show the thermal resistance, we need to control the variables in the test environment.

It is especially important that the total heat flow of Qx cannot be changed.

This is why we believe that when using the actual platform for heat dissipation testing, only when the CPU power consumption is locked and the room temperature is guaranteed to be the same, the full-load operating temperature of the CPU can be used as the performance of different radiators.

Strictly speaking, it is the basis for the comparison of temperature control performance.

The reason is also why we use a fixed power heating platform to test the radiator and evaluate the radiator according to the temperature difference.

Thermal conductivity: represents the ability to transfer temperature

According to the above discussion, it is not difficult for us to draw the following conclusion, that is, under the condition that the CPU temperature is variable and the total heat flow is constant, as long as the structure, size, air temperature and convection method of the radiator do not change.

Then the “Heat dissipation performance” does not actually change with the change of material, but the ability to control the CPU temperature is indeed closely related to the material.

The aluminum material is worse under the same conditions. The so-called “aluminum material is more conducive to heat dissipation” is just an imprecise and incorrect statement.

There is almost no trace of pure copper heat sinks now

So why are the radiators almost never made of pure copper?

First of all, we can see from the previous calculation that in fact, the difference in CPU temperature between the copper radiator and the aluminum radiator in actual use is not very large.

Basically, only those who pursue the ultimate heat dissipation effect need to use a pure copper structure.

In this case, the aluminum radiator of the same structure can also meet the needs.

The second is the radiator of the two materials. In actual use, the time from the start of work to the temperature stabilization is almost the same.

In heat transfer, when the temperature of the entire system stabilizes and does not change, it is generally called steady-state heat transfer, which is equivalent to what we often call “maximizing heat dissipation efficiency”.

Therefore, for the heat dissipation system, the more The sooner you enter the steady state heat transfer, the more favorable it is for heat dissipation.

However, the thermal conductivity only expresses the ability of the material to transfer heat, and the temperature is related to the heat capacity of the material, while the heat capacity is related to the specific heat capacity and mass of the material.

However, in the cooling system, the radiator is more of the same volume rather than the same weight, so here we can introduce a coefficient, called the thermal conductivity coefficient, also known as the thermal diffusivity.

If the thermal conductivity shows the ability of different materials to transfer heat, then the thermal conductivity shows the ability of different materials to transfer temperature.

The formula for thermal conductivity is α=λ/ρc, where λ is the thermal conductivity, ρ is the density, c is the specific heat capacity, and the product of ρc represents the heat required to raise the temperature of an object per unit volume by 1°C.

These parameters are actually known values, so we can calculate the thermal conductivity of copper and aluminum, the former is 115mm 2 /s, the latter is 100mm 2 /s, that is to say, in the case of the same structure, copper The speed at which the material radiator reaches temperature stability is only about 15% faster than that of the aluminum material, and the thermal conductivity of the former is 1.7 times that of the latter.

Compared with the aluminum material, in terms of the current common CPU radiator, the full calculation is a gap of about 1 minute, which can be ignored in actual use.

Summary:

It is not only the heat dissipation capacity that determines the structure of the radiator

In fact, this time we only discuss the steady-state heat conduction in thermodynamics and heat transfer.

In fact, the heat dissipation of PC will be a more complex process, and more parameters need to be considered in the actual calculation.

For example, in our discussion The radiator is designed according to an idealized model, but the actual design of the radiator will be more complicated in structure, and the composition of thermal resistance also has more items.

Even the convective heat transfer coefficient of air is actually It is not a fixed value, but needs to be calculated and processed in different positions.

The knowledge involved here is quite wide, and it is not something that our super class can fully explain. We can only make ideals. explain.

The actual structure of the radiator is far more complicated than the theoretical calculation

Returning to the question of whether it is better to use copper or aluminum for CPU heat sinks, in fact, many students will subconsciously think that when the CPU is fully loaded, the temperature is higher, and the heat generation will be greater.

In fact, as long as the power consumption of the CPU remains unchanged, when its temperature is stable, its “heat generation” or total heat flow is actually basically the same.

If it is explained in terms of a circuit, it is the entire composition of the CPU and its cooling system. In fact, it is equivalent to a “variable voltage constant current circuit”.

The current is equivalent to the total heat flow, and the temperature difference between the CPU and room temperature is equivalent to the voltage.

In other words, the potential difference, the thermal resistance is naturally the resistance in the circuit.

When the thermal resistance increases, since the heat remains unchanged, the temperature difference naturally needs to increase, that is to say, if the room temperature does not decrease at this time, it can only be that the temperature of the chip increases.

If you want to reduce the temperature of the chip, you can only reduce the thermal resistance of the cooling system without changing the heat generation of the chip and the room temperature.

If restrictions are added under this premise, for example, the size and structure of the heat sink are not changed, then a material with a higher thermal conductivity can only be used, such as changing from aluminum to copper.

In this way, the heat generation and heat dissipation of the entire system remain unchanged, but the temperature of the chip can indeed be lowered.

Therefore, the statement that “aluminum dissipates heat better than copper” is not correct in a strict sense. Whether copper or aluminum is used for the radiator, or even other materials, only affects the temperature of the heat source. For “heat dissipation” is unchanged.

Of course, if you just look at whether the temperature of the CPU is low enough, then in the case of the same size and structure of the heat sink, the copper material can indeed keep the temperature of the chip lower than the aluminum material.

So why do most of the heat sinks on PCs today use a “copper-aluminum combination” instead of a pure copper structure with more ideal temperature control?

According to the formula of thermal diffusivity, we can also know that the thermal diffusion rate of aluminum material is not much different than that of copper material, that is to say, the time for the two to enter the temperature equilibrium state is not much different.

Therefore, the pure copper radiator does have a better performance on the CPU temperature, but after considering the cost, weight, processing difficulty and other factors, compared with the aluminum radiator, the investment is not proportional to the income.

The pure aluminum radiator does not have an advantage in temperature performance. Therefore, after synthesizing various factors, it occupies the bulk of the radiator.

The fins that are mainly used to expand the heat dissipation area are made of aluminum and are in direct contact with the heat source.

The structure, such as the base, heat pipe, etc., is made of copper.

This structure forms an optimal balance point, and it gradually evolves into the basic structure of today’s mainstream and even high-end radiators.

DIY a Router

DIY a PBX

DISCLAIMER OF COSFONE.COM

Is it better to use copper or aluminum heatsink?