Introduction

Optimal design of a heatsink, meeting program targets for cost, weight, size, and performance, is one of the more challenging activities within most electronics engineering teams. Without a dedicated solver, designers or thermal engineers can be involved in a game of ‘opinioneering’, which typically involves overdesign, or initiate an expensive and time-consuming physical design of experiments that provides limited results. In the case study below, we worked to optimize an extruded aluminum heat sink for an IGBT module in a low-cost, high-volume design. Reliability requirements (10-year life), harsh environments (vibration, elevated temperature), limited ability to perform maintenance (consumer household), and cost constraints eliminated forced air cooling as a practical solution. The focus was instead to optimize the design within the dimensional constraints provided by the end-user.

IGBT Module

Optimization was requested because the IGBT module starts to behave intermittently at 60°C and shuts down completely at 65°C. The module shutdowns due to a thermal cutoff limit of 100°C at the internal thermistor. The data sheet for the IGBT module indicates that the cutoff temperature of the IGBT junction temperature is 150°C. The module thermal resistances are supplied by the manufacturer and are shown in Figure 1. The goal of this analysis was to reduce the thermistor temperature during operation such that the module achieves stable operation at 60°C and shutdown at or above 65°C. To achieve this goal, the case temperature of the IGBT must be decreased. This increase in the operating margin will be achieved by modifying the relevant heatsink parameters. The baseline heatsink dimensions are displayed in Figure 2.

Figure 1: Thermal resistance of IGBT module

Model Calibration

A thermal analysis of the original heatsink design was conducted to baseline the thermal model. A thermal image of the heatsink while the IGBT module is dissipating 40 watts at 20°C is shown in Figure 3 (left image). The maximum heat sink temperature is measured as 90.1C, or a 70.1C rise above ambient. The results of thermal simulation at 40 watts at 20C are also shown in Figure 3 (right image). The maximum temperature rise was predicted to be 75.7C. The difference between measured and predicted is within 6%, which is a reasonable margin of error.

Figure 3: Heatsink temperatures at 20°C ambient

Design of Experiments

The heatsink parameters assessed in this design optimization study are presented in Table 1. Number of fins, fin height, location of heat source, and surface treatment were all assessed in terms of their ability to lower IGBT case temperature.

Table 1: Heatsink DoE Parameters

Results

When the calibrated model is run at the desired use condition of 20 watts at 60°C ambient, the case temperature rise is 43.1°C over ambient (see Figure 4). This clearly explains the intermittent operation of the IGBT under these conditions and indicates that for the heatsink optimization to be successful, the new heatsink design must reduce case temperatures by at least 5 and preferably 10C.  Interesting finding within the DoE: At 20C and 40W, the optimum number of fins is six (6). However, when the ambient temperature is increased to 60C, but the power dissipation is lowered to 20W, the optimum number of fins drops to five (5).

Figure 4: Baseline heatsink analysis at 20 watts, 60°C ambient, five fins optimal, 103°C

Heat source location 
The effect of moving the heat source was simulated at 60°C ambient conditions. Moving the heat source to the heatsink center had little effect on the performance (<1°C temperature change), as shown in Figure 5. Moving the heat source above the centerline increased the temperature by almost 3°C, as shown in Figure 6.

Figure 5: Heat source located at center

Figure 6: Heat source located near top

Heatsink Fin Height

The effect of changing the fin height was modeled at 60°C ambient conditions. Increasing the fin height by 0.25” decreased the temperatures by 5.5°C, as shown in Figure 7. Increasing the fin height by 0.5” decreased the operating temperature by 10°C. This indicates that the module will be running at 93.6°C. Increasing the ambient temperature to 65°C has a minimal impact on the temperature rise. At 65°C the IGBT module will be operating at 98.7°C and should be at the limitations of its operating range. Further modifications of the heatsink should be done to increase this margin.

Figure 7: Heatsink fin height +0.25”

Figure 8: Heatsink fin height +0.50”

Effect of Anodized Surface

The effect of radiation heat transfer is very important in natural convection, as it can be responsible of up to 25% of the total heat dissipation2. The capability of a material to radiate heat is given by its emissivity. Extruded low-cost aluminum has a relatively low emissivity (0.02 to 0.2), which can impede its thermal performance. One way to improve emissivity of aluminum is through an anodization treatment. Anodization is an electrochemical treatment process that introduces a relatively thin layer of oxide. When the treatment is combined with a black dye3, it can increase emissivity to almost 0.9. To assess the impact of anodization, the emissivity was increased from 0.4 to 0.9 for the model with the as designed fin height. The model was then run with the 0.5” added to the fin height. The results are shown in Figure 9 and Figure 10. Increasing the emissivity alone on the heatsink is not sufficient to drop the operating temperature below 100°C when operating at 65°C and should be combined with additional fin height which should place the operating temperature at about 94.3°C at 65°C ambient.

Figure 10: Emissivity increase with +0.5” fins, 13.8°C drop

Conclusions

Using more standardized optimization techniques, it was determined that fin height and anodization had the greatest ability to drop case temperatures below 100C when the IGBT was dissipating 20 watts at 65°C ambient. Orientation of the heat source and fin count had minimal effects.

Delivered to you by P&A International

Advantages of using copper over aluminium for heat sinks are often discussed. If you look at high end CPU coolers for personal computers market for example, you will see extended use of copper. Coper thermal conductance is 4.01 W/cmK and aluminium is 2.37 W/cmK. With copper thermal conductivity so much higher that aluminium one would expect significant performance difference in real world. Let’s run some thermal tests.

We have set up a test rig consisting of 200mm X 200mm X 2mm copper and aluminium plates. A heat source was chosen as power mosfet and was mounted in the middle of the plate. A continuos thermal load of 30 Watts was dissipated via heat sink. Temperature was measured at heat source (mosfet mounting pad). Ambient air temperature was measured at 24 Celsius and air was stationary.

During first 10 minutes there was no significant temperature difference between copper and aluminium heat sinks. We recorded 1 degree C and temperatures have reached 60C and 61C with cooper heatsink running cooler. Such a small difference can be accounted for the higher thermal inertia of the copper heat sink, as the mass is higher it takes longer to reach same temperature at fixed thermal input.

On 15^th minute, mosfet mounted on to the copper heatsink reached 65C while mounted on aluminium heatsink was running at 66C. At about this point advantage of copper heat sinking becomes evident.  By transmitting heat more efficiently, it allows components to operate cooler.

After 120 minutes, the 1 degree C thermal advantage remained valid. Temperatures reached 84C for coper and 85C for aluminium heatsink. There was no further improvement with coper versus aluminium, both plates have saturated at about 150 minutes maintaining same temperature difference.

Is the coper heatsink better than aluminium? Sure, thermal component will run cooler mounted on the coper. Is it wort additional cost as coper is so much more expensive and punishment of extra weigh? Maybe, depends on application. If you already exhausted all the possibilities of using cheap and light aluminium heatsink and chasing small improvement for the critical thermal system, then definitely coper is a great candidate.

Now, at the same time we had a third test rig with the extruded aluminium heat sink with finns and same mosfet dissipating 30 Watts of heat on it. It was cooled via conventional air flow. Heatsink was 1/6 the size of the plate, but had 50 mm fins on the back. It is old Pentium III cooler. Made out of extruded aluminium, about the same weight as our 200mm aluminium plate it performed great. After 120 min run, it reached only 66 degrees Celsius.  Compare this to the 85C of the copper plate.

So by using same amount of material, copper or aluminium, you can have a dramatic improvement in heat dissipation just by altering the design of the heatsink. This is the most important message from the test result. Only when heat sink design has exhausted itself for particular application, then material choices worth considering. Or fan forced cooling, as been activated on 121 minute and resulted in finned aluminium heat sink temperature drop of 17C.

As tested by engineers at P&A International