Heatsink 101: Everything You Ever Wanted To Know
On-demand Web Seminar
During this presentation, we will focus on how heatsinks work and how to design a heatsink while considering all the critical factors such as size, airflow, cost and attachment methods. We will also investigate how, using simulation, the heatsink design could be optimized and validated in the application environment. Engineers involved in board and chassis design would find this session very educational.
As the power density of electronics devices go up, so does the need for chip and system thermal management. One way to cool the components down is to add a heatsink. heatsinks uses conduction, convection and sometimes radiation to enhance the heat transfer from a hot surface to a cooler fluid. Many factors such as cost, manufacturability and weight need to be considered when choosing a heatsink. How the heatsink gets attached to the component is also critical.
What You Will Learn
- How to design a heatsink for a specific electronics cooling application
- To visualize the airflow around the heatsink and identify potential bypass areas
- Estimate heatsink thermal resistance
- Use Response Surface Optimization to come up with the best heatsink design for the considered environment
About the Presenter
Alexandra Francois-Saint-Cyr is the Applications Engineering Manager for North America at Mentor Graphics, Mechanical Analysis Division (Previously Flomerics). For the past 8 years, she has been working on promoting the use of the Mentor Computational Fluid Dynamics (CFD) software by conducting training classes, seminars and software demonstrations.
After graduating from ESSTIN, France’s State Graduate School of Engineering back in 1997, Alex studied to improve passenger comfort in the trains at Alstom Transport by using CFD software FloVENT. In 1999, she moved to the United States where she received her MSME from the University of Central Florida. There she worked on a meso-scale centrifugal compressor project, her research leading to the ‘Best Technical Paper’ award in the Advanced Energy Systems Division at the International ASME show in 2000
Who Should View
- Engineers who have thermal problems with electronic based applications
- Technical Managers
- Thermal Engineers
- Board and Chassis Engineers
10% off your first thermal design consulting project!
Thermal Design – Thermal Modeling – Model Creation Learn More
Parmetric Study of an IGBT Cold Plate Geometry in a Thermal Simulation
The CAD design of a cold plate for automotive power electronics is the key for an efficient cooling. A parametric study of the cold plate design was conducted to find the optimum geometry parameters such...…
Feel the Heat in the Automotive Industry: Thermal Simulation as an integral part of the Design Process
Today’s automotive products have to be in line with the automotive industry's megatrends that make driving safer, more comfortable and more sustainable. High performance products, such as Navigation-,...…
Other Related Resources
How to Predict Component Temperatures for Complex Network and Telecommunications Equipment
The insides of network and telecom equipment consist of a complex assembly of many solid objects (PCBs, electronics packages and devices, cabling, fans, and heatsinks, for example). Airflow is confined...…
11 Top Tips for Energy-Efficient Data Center Design and Operation… A High-Level ‘How To’ Guide
When is Data Center Thermal Design Important? The usage of data centers globally has skyrocketed in the past three to four years in response to growing demand for information storage and transfer from banks,...…
Q and A Transcript
- Q. Wouldn’t weight also be a factor in the heatsink consideration?
- A. Weight is definitely a factor to consider for a heatsink design. It has an impact on the heatsink cost, the component mechanical stress and the heatsink mounting on the top of the component.
- Q. How much better is the heat transfer in a vapor base heatsink compared to just a solid copper base?
- A. A vapor based heatsink uses the same technology as a heat pipe. The heat coming at the bottom of the base vaporizes the fluid stored in the base. The vapor then spreads into the entire base and condenses back into liquid on the inner surface. This creates a base with a much higher equivalent thermal conductivity than an aluminium or a copper base heatsink.
- Q. What data do you need to estimate a component junction temperature?
- A. There are different ways to estimate the component junction temperature. If you have the thermal resistance values (RJC and RJB) from the chip manufacturer, then you can use the same method as explained on slide 6 to compute the component junction temperature. These values can also be entered in a CFD software such as FloTHERM and FloEFD. If these values are not known, you may need to look into the internal details of the package. Mentor Graphics Mechanical Analysis has a package library web tool called FloTHERMPACK, which allows detailed packages along with their 2R or multi-resistor networks to be created and analyzed using FloTHERM.
- Q. What’s the difference between a laminar and a turbulent flow?
- A. A laminar flow is a fluid which moves in parallel layers, which no disruption between the layers. A turbulent flow is a fluid which flows in a chaotic manner, creating high variations in pressure and velocities in space and time. Any fluid mechanics books listed on slide 23 will contain more details about this. A heatsink exposed to a turbulent flow will have a higher heat transfer rate and a higher pressure drop than if it was exposed to a laminar flow.
- Q. In the CARMA board example, did you investigate other heatsink designs?
- A. The other heatsink type investigated was a monolithic heatsink with a base covering all four components. It actually gave better results than the individual component heatsink mounting because the base acted as a nice heat spreader for all the components.
- Q. The Integra-Luxtec light model analysis seems to have been done on the actual CAD geometry. How did you do it?
- A. FloEFD is a CFD software which is embedded into major CAD tools such as ProEngineer. You can then define your thermal analysis boundary conditions on the original CAD design, solve the model and explore the results such as temperature distribution against the CAD geometry.
- Q. So you mean that circular shape of your 2nd example is with FloEFD not FloTHERM?
- A. That’s correct. The Integra-Luxtec model was done with FloEFD.
- Q. Is there a minimum temperature where radiation always should be considered?
- A. Radiative heat transfer between surface 1 and surface 2 is as follow: Q12 = e1sA1 F12 (T14-T24). As it can be seen from the formula, radiation will become more significant if surface 1 is at a higher temperature than surface 2. Other important factors to consider is the 2 surfaces’ emissivities and the view factor, which is how surface 1 and surface 2 see each other. If the heatsink radiates to ambient air, radiation should help dissipate more heat away as long as the ambient air temperature is lower than the heatsink surface temperature.
- Q. How easy is it to change the gas properties of the fluid?
- A. Fluid properties such as density, specific heat, viscosity and thermal conductivity are fully defined in pre-existing libraries. In case you need to modify them, you can just edit the libraries and create new entries for your fluid of choice.
- Q. Can I input the theta SA value to FloTHERM without modeling the heatsink?
- A. Yes you can. There is an option to attach a predefined profile of the heatsink thermal resistance versus velocity to any solid surfaces. Note that you will also need to define a flow resistance in order to take the heatsink pressure drop into account.
- Q. If junction-to-case and junction-to-board values are not available on chip manufacturer's datasheet, can we make some reasonable assumptions on how much heat is dissipated to the heatsink and how much heat is dissipated to the PCB?
- A. When a heatsink is placed on a component, people will typically assume that 100% of the heat will go up into the heat sink. Some may consider 90% to the heatsink and 10% to the board. When using a CFD tool like FloTHERM, note that you can derive the component’s thermal resistances from the package library called FloTHERMPACK. It contains JEDEC standards libraries such as BGAs, PQFPs, etc. Most of today’s components are part of FloTHERMPACK.
- Q. How much better is a turbulent flow compared to laminar?
- A. Let’s take a look at a simple example such as flow over a flat plate from slide 10. Let’s assume that the plate is 1 meter long, that the Prandtl number Pr equals 0.7 and that air properties are for an air temperature of 30°C. For a Reynolds number Re of 63120 (laminar flow; velocity = 1 m/s), the Nusselt number Nu equals 150. For a Re of 126240 (turbulent; velocity = 2 m/s), Nu equals 396. That’s a 264% increase. Keep in mind though that the pressure drop will also increase.
- Q. How do you go about estimating the CASE to PCB resistance and heat transfer/flow?
- A. Case to PCB resistance for a typical BGA package is equal to the sum of Junction to Case and Junction to Board thermal resistances. For a package such as a TO220, which gets its bottom side mounted to a heatsink, Case to PCB is the heat transfer path from the Case of the package to the leads connecting to the PCB. If the internal details of the package are known, a 1D conduction network could be defined. A detailed package placed in the system environment can also be built and analyzed using a CFD tool such as FloTHERM for a full 3D picture of the heat transfer path.
- Q. For pure convective cooling does the fin thickness play an important role as fin height and fin spacing?
- A. The fin thickness typically plays a minimum role compared to fin height and fin spacing. Fins are usually not thick enough to create a high pressure drop effect at the entrance of the heatsink. It will become a more important factor when the fin spacing is small though.
- Q. How does pressure affect temperature drop?
- A. If you look at slide 11 of the presentation, it shows how the heat transfer coefficient varies with velocity. Since Pressure drop varies with velocity at the power 2, such that DeltaP = f(V2), the heat transfer coefficient will be a function of: h = f(DeltaP0.25). You can then see how the heat transfer coefficient increases with pressure drop.
- Q. In computers, heatsinks tend to get really dirty, reducing their efficiency. Can you tell us of some methods to make this problem smaller?
- A. The method to avoid this situation is to put air filters at the different air inlets. If it’s not possible, then you may need to open the computer on a regular basis to vacuum the dust.
- Q. Do you think if it used pin-fin? Do you have result of pin-fin? In addition, circular shape of your 2nd example goes well with the grid?
- A. Pin fins were not considered for the 2 examples shown in this presentation. Regarding the second example, the Integra-Luxtec model, it was done using FloEFD. FloEFD is a CFD code which uses an Octree based mesh that handles any complex geometrical shapes.
- Q. Is there a good ratio for the heatsink base and fin?
- A. A fin to base ratio has to be considered mainly for manufacturing issues, for example for extrusions. Fin height-to-gap aspect ratio up to 6 and minimum fin thickness of 1.3 mm are attainable with standard extrusion. A 10 to 1 aspect ratio and a fin thickness of 0.8 mm can also be achieved with special die design features.
- Q. Can you define pressure?
- A. Pressure can be defined along with temperature and velocities.
- Q. Do you have the reference for the minimum fin spacing formula?
- A. The formula for the minimum fin spacing on slide 11 is derived from the Laminar Entrance Region formula: Le = 0.05Re.D with Re = (r.V.D)/m . This formula can be found in many Flow and Heat transfer books such as Burmeister, L.C, Convective Heat Transfer, 1983.
- Q. What about junction-board thermal resistance? Does that get entered?
- A. The junction to board thermal resistance can definitely be considered in the thermal analysis. The software will then take it into account when computing how heat gets transferred from the junction to the PCB. If you would like to know how this value can be calculated, please have a look at question 3.
- Q. Can heat be transferred downward?
- A. When using a CFD tool such as FloTHERM or FloEFD, heat conduction, convection and radiation are solved in 3 dimensions so you will be able to see if any heat goes downward or not. Typically, even if there is a heatsink placed on the top of a component, some heat will go towards the PCB.
- Q. What is the range of the surface finish recommended...is high polish the best? If the surface finish is not smooth will the AIR passing through swirl around in the eddies? Is that good?
- A. It’s typically better if the surface is not highly polished for 2 reasons. 1. It will increase the turbulence of the flow and therefore improve the heat transfer rate. If you look at question 4 and 12 in the Questions and Answers document, you will see the difference in performance between a laminar and a turbulent flow. 2. If the heatsink is designed for a natural convection application, radiation is more significant. The emissivity of a rough surface is higher than the one from a polished surface. You will therefore have more heat dissipated by radiation from a rough surface than from a polished surface.
- Q. Do fins have an ideal configuration? Round, square, rectangle.
- A. There is no ideal configuration as it will depend on many factors such as flow rate, orientation, etc.Round fins are good if you do not know which direction the airflow will be from.Most of the time, rectangular fins give the highest heat transfer rate compared to round or square fins but they also tend to give the highest pressure drop.Please have a look at slide 12 in the presentation.
- Q. Sharp corners? Round corners?
- A. It’s hard to answer this question as is but in general, Round corners will give a lower pressure drop than sharp corners.