THERMAL TESTING OF A 3D PRINTED NON-UNIFORM HEATSINK AGAINST STATE-OF- THE-ART FINNED GEOMETRY Robert Smith, P.E., Chief Technologist FEBRUARY 13, 2016 QRP 6764 Airport Road West Jordan, UT 84081 www.qualifiedrapidproducts.com 2/13/2016 copyright 2016
Abstract Two 3D printed heat sinks were tested against each other. One had the geometry of a typical finned heatsink and the other had a geometry that makes slight modifications that take advantage of the flexibility of 3D printing (called non-uniform here). The surface area for both heat sinks was essentially the same; however, the non-uniform heat sink gradually changed from thick to thin in a way that improved airflow and fin efficiency. The results showed that the non-uniform heat sink had significantly lower thermal resistance (30%) in spite of a slight pressure drop increase. The non-uniform heat sink shape was selected using best engineering judgment with a basic understanding of heat sink theory. More work could be done to find the optimal shape. Experimentation combined with CFD could likely iterate to find an even better shape. The AlSi10Mg is not as conductive as 6063 aluminum but perhaps a shape can be identified that counterbalances this weakness, or future research may identify a more conductive 3d printable material.
Introduction A test was conducted on a non-uniform heat sink to compare it against a leading commercially available geometry from Alpha heat sinks (Z40-12.7B). The Alpha heat sink is generally build of 6063 Aluminum and is bonded together. The purpose of this test was just to compare the geometry differences so both the Alpha geometry and the non-uniform geometry were both printed and compared. The Alpha printed heat sink was also compared against the Alpha vendor datasheet to compare 3D printing vs. the 6063 bonded version. The objective was to compare pressure drop as well as thermal resistance over a function of airflow. Figure 1: 3D printed heat sinks printed for both Alpha geometry and mesh fin geometry Approach Concept Laser CL31 aluminum was used to print the geometry. Both parts met the geometric requirements of the Alpha vendor datasheet including the smooth bottom finish. The unit under test consisted of a finned heat sink similar to the Alpha geometry but with subtle differences. Since the heat load was a small.5 x.5 square load in the center, the base plate was made thicker in the center and gradually got thinner toward the edges. The fins were made slightly thicker than the Alpha fins at the base (.020 ), but tapered much thinner near the top (.009 ). In addition, the base of the fins was even thicker in the center of the heat sink and each fin gradually got thinner towards the exterior. With the extra air space gained from the thinner fins at the top, an extra fin was added, totally 27 fins). The leading and trailing edges of the fins are also tapered somewhat. QRP followed the test procedure used by Alpha and published on http://www.micforg.co.jp/en/temeasuree.html. A test plenum was printed on an FDM printer to channel the flow through the heatsink. A tiny type T thermocouple was placed in a tiny hole on the bottom of the heatsink. (Omega part number 5TC-TT-T-40-36). A foil heater was applied on top of the thermocouple on the bottom surface of the heat sink that was 0.5 inches square (Birk i part number BD3546 53.0-L24-03) with a resistance of 53 ohms.
Figure 2: Subtle variations in heat sinks are difficult to spot. Figure 3: Test apparatus used
Figure 4: Alpha Z40 heat sink datasheet
Figure 5: Test set-up with no heat sink installed in test fixture. (Insulation not applied yet) (Other sample parts included in this figure were not part of the scope of this test). One test was conducted with no heat sink installed and the bottom plugged up with a smooth surface mimicking the rest of the walls so that the pressure of the system could be subtracted from that of the heat sink. The first objective was to try and get results that approximated the Alpha datasheet for the Alpha shaped heatsink. The pressure curves were assigned a second order polynomial trendline to identify any potential bias in the pressure gage. This Y-intercept value was subtracted from the values measured in the test. Results Figure 6 shows the performance of the Alpha geometry compared against the vendor datasheet. The pressure drop ended up being lower. The test set-up was inspected to see if the gaps on the side of the heatsink were too large and they appeared to be very close to the spacing of the fins as specified in the original test procedure. Higher altitude of Utah testing (4600 ft) may also account for some of the lower pressure drop. The baseline printed heatsink was slightly bowed when post machining so there is a little more air gap at the top of the sink. This may explain some of the lower pressure drop and lower performance.
Static Pressure (mm w.g.) The thermal resistance was higher for the printed heat sink. Part of this may be due to the material thermal conductivity but maybe not all. The values are within the same order of magnitude of the vendor sheets and the two sinks were tested in the same way. Therefore, the comparison of the two geometries is deemed valid. Figure 7 shows the pressure drop across the non-uniform heatsink compared against the Alpha sink. It was only slightly higher but the resulting flowrate is very close to the same. Figure 8 shows the thermal resistance comparison between the two heat sinks. The difference was significant. Assuming a standard off-the-shelf 40mm fan, even with the slightly higher pressure drop the non-uniform heat sink had a 22% improvement over the Alpha geometry. 5 4.5 4 3.5 3 2.5 2 1.5 1 Printed Alpha pressure Z40-12.7B spec mm H.G. Z40-12.7B spec C/W Printed Alpha Resistance 0.5 0 0.0000 0.5000 1.0000 1.5000 2.0000 2.5000 3.0000 3.5000 AIRFLOW-(l/s) Figure 6: Performance of Alpha geometry against vendor datasheet
Thermal resistance ( C/W) Static Pressure (mm w.g.) 6 5 4 Variable fin geometry Baseline Geometry 3 2 1 0 0.0000 0.5000 1.0000 1.5000 2.0000 2.5000 3.0000 3.5000 4.0000 4.5000 AIRFLOW-(l/s) Figure 7: Pressure drop across 32 fin heatsink 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 Variable fin geometry 0.0 0.0000 0.5000 1.0000 1.5000 2.0000 2.5000 3.0000 3.5000 4.0000 4.5000 AIRFLOW-(l/s) Figure 8: Thermal resistance comparison Conclusions The modifications made to the non-uniform heat sink were very subtle. Looking at the final parts, one can barely distinguish the difference. However, these subtle differences had a significant impact on thermal performance. Flexibility with slight transitions in surfaces opens the door for a lot of research to find the optimal heat sink. It is recommended that CFD be used to find a more optimal non-uniform geometry and then further testing be conducted on a hand-full of ideas. It is also recommended that more work be done to identify 3D printable materials that have higher thermal conductivity.
i Purchasing from Birk may have been the most painful purchasing experience I have ever had. They were extremely slow to respond, I had to call many times and get passed around from person to person, their communication was poor and part numbers are confusing, they asked me to reformat my email to make it more convenient for them, they made the payment method difficult, and they took a long time to deliver. Order attempts sent to their sales email seem to go into a black hole.