The following transcript and video were originally presented at Thermal LIVE 2015 by George Meyer. For access and information on all Thermal LIVE events, both past and present, please visit https://thermal.live/.
Hello and welcome to Thermal LIVE 2015. This presentation is A Practical Guide to Using Two-Phase Heat Sinks. My name is Graham Kilshaw and I’m the Chief Meteor Officer of Electronics Cooling and we are the creators of Thermal LIVE. If this is your first time at Thermal LIVE, this is a three-day event hosted by Electronics Cooling. It features practical educational presentations around all aspects of thermal management. You’ll see round tables, webinars, discussions, videos and our exciting thermal products of the year competition, which is going on right now on the Thermal Live Site. So get on over to the Thermal LIVE site and vote before it closes tomorrow. There’s no cost to attend any of these presentations, so it’s a great opportunity to learn more about thermal management from the comfort of your office chair, so please sign up for more of the presentations this week.
Today’s webinar is a practical guide to using two-phase heat sinks and it’s presented by George Meyer the CEO of Celsia Technologies. George is a thermal industry veteran with over three decades of experience in electronics and thermal management and he currently serves as the CEO and CTO of Celsia and holds over 70 patents in heat sinks and heat pipe technologies. The Celsia company specializes in two-phase heat sink design and they manufacture for Fortune 500 and mid-market enterprise companies. Their company shipped over two million products from their U.S. and Asia-based design and product facilities, and their goal is to provide fast, affordable and reliable thermal solutions using vapor chamber, heat pipe and hybrid designs.
A recording of this webinar and downloadable PDFs of the presentation will be available on the Thermal Live site early next week, so check back there if you want a copy of the presentation. This webinar is interactive, so you will be able to ask questions throughout George’s presentation and we encourage you to do so. Just take a look at the GoToWebinar screen now and look at the box on the right-hand side of the screen which is labeled question. If at any time you want to pose a question, just fill out the box and hit send, that simple. We’ll receive your questions during the course of the presentation and then we’ll try to answer most or all of those questions at the very end. If you want to minimize or maximize your screen just use the arrow button, and if you have an issue during the presentation, just click the hand icon to raise your hand.
George’s presentation is about 35 minutes today and it will be followed by a 10-minute question and answer session. So, over to Mr. George Meyer, who will begin the presentation now. Over to you George.
Thanks, Graham. Thanks everybody for showing up today and I just wanted to double-check and make sure that everything, you can see my screen properly and we’re good to go. On that note we’ll continue then. All right. So thanks so everybody for showing up and Graham for hosting this. We’re just going to get right into it. During the presentation today we’re going to go through these items. The questions we’re going to answer is: Do I need two-phase for my thermal design? Are vapor chambers just flat heat pipes? We’re going to talk about the sizes and quantities you might need for an application. How do I integrate them into my thermal solution? What does the heat exchanger look like, in other words, how do I get rid of the heat? And then finally, we’ll talk a little bit about how I do thermal modeling and performance modeling of these things.
The first slide that we’re going to look at looks at three different designs, an aluminum design, a copper design and a design using a vapor chamber. But the question is when to use two-phase devices, and the short answer is, you really only need to use those is when conduction is limiting the performance of your thermal solution. So if you have high conduction losses in your solution, that’s what two-phase devices do. There are other reasons you’ll want to take a look at these and other benefits that they will give you such as weight, size, thermal profile, helping to minimize hot spots. So those are ancillary benefits that you get from them. In looking at these three solutions, we look at the aluminum one and we’ll just call that baseline one. The weight of it is one, the performance is one and the cost is one.
So we’ll start with an aluminum heat sink. These are 1U heat sinks. The normal progression is to move to a copper heat sink, and you can see, we kept the base thickness the same. The weight has gone up by a factor of three, just because copper weighs three times of aluminum. Cost is generally about somewhere between 1.6 and two times the cost of an aluminum one. If you look at the two-phase device, you can make the bases thinner using the two-phase device, because it’s got such high conductivity and that then generally brings the weight down to underneath copper, so it’s about two times aluminum, but 30% less than copper and the cost is just a little bit over what the copper will be. From a performance standpoint, the copper one will generally give you about five-degree performance improvement over the aluminum one and the two-phase one will generally give you about a five-degree performance improvement over the copper one.
So, let’s take a look at some rules of thumb. We all know that two-phase devices are incredible heat conductors, generally from five to 50 times better than what you’ll see in copper. Numbers that the people like to use are somewhere between 1,000 and 50,000 watts per meter-K. The exact figure is something you could calculate, but it’s primarily dependent upon the distance that you’re moving the heat. Ideally, if you’re looking at using two-phase devices, you really need to be moving the heat more than we like to say an inch or two. Put those in millimeters, it’s 30 to 50 millimeters. Remote fin stacks heat exchangers are a perfect example of that.
If you’re interested in using these devices to reduce hot spot or attache to a local heat exchanger like the one 1U heat sinks we looked at, there’s a rule of thumb that I like to use and that is that the area ratio of the heat input device, your semiconductor, and the area of the vapor chamber ought to be a 30 to one ratio. What do I mean by that? If you have a one square centimeter heat source, then your heat sink or your vapor chamber should be at least the size of 5.5 centimeters by 5.5 centimeters, so roughly 6 X 6. That size will start to give you performance improvements over using just metal. The final rule of thumb, when you’re looking at designing these things, if you need to move 100 watts, don’t choose a device that’s rated at 100 watts. We generally like to design in about a 30 to 40% thermal headroom.
Let’s take a look at the inner workings of these things. Everybody has seen these slides before, we’re not going to spend a whole lot of time on this. The only thing I want to point out on this slide is if you look at the internals, the cross-section of the vapor chamber, in the lower right-hand corner, shows you the support structure that’s used in these things. It’s used for two reasons. One, when you do clamping on these things, the clamping pressure is going to be quite high. Typical things in the consumer market are like 20 PSI. In the server market you can get 50, 60 pounds of pressure clamping against these things. The support structure is there to transfer the clamping pressures across the open area of the vapor chamber.
The first thing we want to look at is the performance of these. What are the different wick structures and how do they compare to each other? There are three main wick structures, a grooved, a screen and a sintered powder. What you will see is the power densities that these different wick structures can handle, the typical thermal resistances that you’ll see with those wick structures and the orientations that they can operate in. Sintered wick structures are what are 99% of what’s used in the marketplace and you can see why. They can handle up to 500 watts a square centimeter. They have the lowest thermal resistance and then generally, they’re pretty much useful in most orientations as long as the length is reasonable. The graph just shows you sort of the thermal resistances. It also shows you where the power densities you’ll start to see dry out.
Now what is the performance limit of these devices? In most cases, the performance limit is the capillary limit. So what we’re really going to do it just talk about the capillary limit for a minute. The chart will show you a particular design, what the performance of that would be. That’s a particular wick structure, thickness, orientation, but the capillary limit, as you can see is the limit. If you’re looking at say the 100-degree mark, this particular design is good for about 60 watts.
I’ve thrown together a chart to give you an idea of these limits based on the diameter of the heat pipe. So the thing to look at is the wick limit is the red line and that is by far the controlling limit. The other two lines are the vapor limits. The blue line is the vapor limit if the heat pipe is round. The green line shows you what happens to the vapor limit when you flatten that heat pipe. In other words, you make the vapor space smaller, so the limit becomes lower. But in both of these cases, the capillary limit, the wick limit is the limit.
So what are the differences between all of these things? We’ve got heat pipes, we’ve got hybrid one-piece vapor chambers and we have two-piece traditional vapor chambers. Hybrid heat pipes, everybody’s familiar with those. Small diameter tubes from three to 10 millimeters. They can be bent, they can be flattened. Typical dimensions are three to eight millimeters flattened. The key thing with the standard heat pipes, between say three and six millimeters, is they’re made by the millions, so they’re very cost-effective.
If we look at our one-piece vapor chamber, those are made similar to heat pipes, but they’re made using very large diameter tubes and they’re designed then to be made into a flat structure. So they can be flattened into a rectangle. The surfaces can be embossed. The typical thicknesses are somewhere between 1.5 millimeters and four millimeters thick, and we can get up to 100 millimeters wide by 400 millimeters long. Those cover most of the uses for vapor chambers, direct contact pressure up to 90 PSI.
Now, if you look at a traditional two-piece vapor chamber, where you actually use two stamped parts, an upper and lower stamped plate, the big advantage of those is you can stamp out complex shapes. So the key advantage of going to a two-piece vapor chamber is simply to be able to manufacture these things in unusual shapes. Most of the other specs are very similar as far as widths and lengths and mounting pressures.
So what is the difference when we talk about moving heat or spreading heat? Well, when you’re moving heat, you’re moving heat from point A to point B. Typically, that’s from a heat device to a remote heat exchanger located somewhere else in the system. If you’re spreading heat, that means that your heat sink is located right adjacent to your heat source. The way to look at that is the heat pipes will give you sort of a linear flow moving heat from point A to point B, where spreading heat using a vapor chamber will spread the heat in all directions.
So let’s talk about moving heat for a minute. 99% of all applications use heat pipes. If you look at notebooks and desktops, and often complex shapes are required, they need to be bent. The photo shows you sort of a typical notebook application. You can see that these are not only bent but flattened and joined together and multiple heat sources. They’re readily available in volume and they’ll work against gravity pretty well, say plus or minus 45 degrees. Ideal for consumer application. And this just shows you an example of that.
Here are two other examples of moving heat. Example in the upper left-hand corner is a heat sink for a small form factor PC. The processor cooler that uses the system fan, it uses four, 6-millimeter heat pipes and move between 100 to 135 watts into the air stream created by the system fan. And the lower right-hand picture is showing some copper blocks and heat pipes integrated into a casting for natural convection to help spread the heat.
This slide takes a look at spreading heat to a local heat sink. So we’ve talked about moving heat, now we’re going to talk about spreading heat. Heat pipes are a good choice, if you’ve got pretty good airflow, you’ve got lots of room for fins, you’re nominal power densities of let’s say less than 25 watts a square centimeter, you have normal ambient conditions and generally, if cost is a concern. The picture shows a telecom equipment application. For moderate performance applications, where spreading needs to be augmented, the use of several heat pipes embedded in the base may be sufficient. So you can see this, using flattened heat pipes does a reasonable job of spreading within the base of a heat sink.
Now, if you have performance limits, for example, if I have a limit to my Z height direction, if I have high power densities, if I have high ambience or low air flows, if you have to meet NIBS requirements, 55-degree ambients or even worse case military requirements where every degree counts, then you take a look at using vapor chambers. This example shows you a GPU cooler, so graphic chips cooler. The original design used two eight-millimeter heat pipes and that’s the photograph on the left. In looking for better performance, we took the heat pipes out, replaced it with a vapor chamber and did a direct contact to the chip, and we picked up six degrees better cooling. So two exact, same form factor solutions, six degrees difference in performance.
Here are a couple of other examples of spreading heat to a heat sink that’s locally to your device. The second example, we saw that before, the heat pipe version used four flattened heat pipes, whereas the vapor chamber version uses a single large vapor chamber. The difference is about four degrees C. The vapor chamber has a four degree C better performing than the heat pipes. That’s pretty typical for what we see when we’re doing these designs and comparing heat pipes to vapor chambers, it’s somewhere between four and eight degrees. We like to say between 10 and 30%. That’s between three and nine degrees normally.
Example number three. We had an application that had six ASICs. The six ASICs had to be within two degrees in temperature of each other and there was a weight constraint of 1,000 grams. What we designed was a two-piece vapor chamber that had an opening in the middle to get rid of the weight, so all six of the ASICS sat on the periphery of this vapor chamber. This design met the specification of keeping each of the six ASICS within two degrees of each other.
Now here’s an unusual example where we’re moving heat to a remote heat sink using vapor chambers. You generally would only do this if thermal performance is absolutely critical and the ultra-thin vapor chambers can allow for more fin area, it gives you some more packaging options. In other words, in this case the vapor chambers just took up less room than the equivalent heat pipe and gave you a little better performance.
And here’s some oddball stuff. Spreading and moving. Example number one shows some flattened heat pipes that are put into a base and then machined to simulate making direct contact with the heat source. Generally that works pretty good. You have to be careful about hot spots when you do that, because you have stripes of copper heat pipes and you have stripes of aluminum. These are good for applications where the ASIC is lidded, so you have some spreading within the chip. It’s not recommended for bare die or flip chip applications, just because of the hot spot issues.
The second example shows the heat sink we talked about earlier for a small form factor PC. They wanted to upgrade the processor to the next generation Intel and AMD processor, and as usual, there’s a bump in power. So instead of changing the design, all that was done was simply a vapor chamber was put between the heat pipes and the device and the chip instead of using just a piece of copper, and that picked up a five-degree performance improvement.
So let’s talk about sizing for a minute. This slide will show you typical heat pipe sizes, diameters from three to 12 millimeters. It’s typical power ranges, for example, the three millimeter will do between five and 20 watts. Most people will ask why such a wide range. The five watts if you’re using that three millimeter against gravity and the 20 watts is if you’re using that three millimeter in a gravity aided position. Remember capillary is always the limit on these things. Usually the limit. So you can go through and see what size do I need, what size do I have space for and roughly how much power it will carry. If we know the exact details of the application, we can tell you the exact performance bumpers.
Also, shown here are the typical thicknesses that these things can be flattened to and then the resulting width once you do the flattening. Then it also shows you some options. In other words, I have a for the same application, a 95 watt ASIC. I can either use three five millimeter heat pipes, because they’ll carry between 11 and 50 watts and if this is a gravity aided or horizontal application, we can use the 50 number. So using three five millimeters, I can get up to a potential 150 watts of heat transfer, so that covers my 95 watts. Or you can use two six-millimeter flattened heat pipes. In either case, it gives you enough thermal capacity margin on the heat pipes and the main difference between these two is the amount of Z-direction that you need above your device. If you’ve got lots of room, then you can stick with the round shape. The evaporator stack up is seven millimeters. If you have a space constraint you can go with two flattened heat pipes the six-millimeter ones and that would give you a stack up of about four and a half millimeters.
So we’ve done the same thing in looking at vapor chambers. The top row shows you the tube size that we start with. The second row shows you the powers that we’re capable of handling with those sizes and just pick one of the sixteen and a half millimeters, you can do from 50 to 150 watts, up to 70 millimeters where you can do 360 to 800 watts. In most cases, when you use a vapor chamber, the capacity of the vapor chambers is many times what’s generally required. Not too many applications for 70-millimeter vapor chamber are going to be between 360 and 800 watts. Generally, they’re in the couple hundred-watt range. Then again, it shows you what they can be flattened to, what the resulting width is. And here again, we’ve shown two examples for that 95 watt ASIC.
If you’ve got a generally a rectangular area, very much like a 1U heat sink, we’ve taken a 43-millimeter tube and we’ve flattened it to three and a half millimeters and that gives us a 67 millimeter and we made it square, 67 x 67 and that’s how you generally cool a 1U heat sink. If you have a more rectangular application, this second example shows a 32 millimeter flattened to three and a half millimeter vapor chamber and it results in a 50-millimeter wide vapor chamber and to get the required performance it’s 90 millimeters long. There again, we go through the comparisons, what the headroom is, how much heat capacity is. And as I mentioned before, the VC heat capacity here, the two examples show 490 watts and 360 watts, and we’re cooling a 95 watt ASIC, so for vapor chambers the thermal limit is generally not the vapor chamber.
Effective conductivity. Everybody says, what’s effective conductivity? If you look out in the literature, there’s numbers anywhere from 10,000 to as high as 50,000. If we know the exact configuration of the heat sink, what’s the cross-sectional area of the T pipe or the vapor chamber, what’s the power, what distance is it moving the heat, we can calculate an effective conductivity. Here’s an example that shows that. The exact same heat sink, one with the heat source at the end and one with the heat source in the middle. And you can see the estimated conductivity of 13,000 watts per meter-K for the heat sink with the heat source at the end and six, roughly 7,000 watts per meter-K with the heat source in the middle. Here again, we can do this calculation very quickly for almost any application.
Bending. Some guidelines around bending. Heat pipe radius three times the diameter of the heat pipe, flattened to roughly one-third of the diameter. Those are general guidelines. If you’re going to machine the heat pipe, make sure that you’ve got sufficient wall thickness to do that. Generally, you need about at least .5 millimeters to do any machining. If you look at the one-piece vapor chamber, bend radius is 10 millimeters along the narrow plane and we flatten these things, from one-tenth to one-twentieth of the original pipe. We can do pedestals, embossments, generally between a half and one millimeter high, if you have to access something that recessed slightly. So the two-piece vapor chambers, you can do a stamped bend in these things as tight as 1-X the material thickness. In other words, if I have a one-millimeter material thickness, I can do a one-millimeter bend radius. You have to be careful about the performance loss when you go around those tight corners, but the upper and lower plates are stamped so you have a lot of flexibility in your designs using the two-piece vapor chamber.
Okay, so bending rules of thumb. So you spec a straight heat pipe or a vapor chamber, make sure you have some thermal margin on that specification. If you’ve got a 75-watt load, make sure you’ve got at least 100 watts of carrying capability. If you’re going to do some bending on the heat pipe, you can do some quick estimates as to what effect that will have on the performance, and we use an example of one 90 degree bend and a 45 degree bend, which gives you basically 135 degrees of bend. Then for each 10 degrees of bend, the Qmax will decline by .56% for each degree. Then we run through an example there, 135 degrees of total bend divided by 10 multiplied by .56 gives you 7.6 degrees in Qmax. So if you’re doing a bunch of bending, you have to be careful. It does affect the Qmax. So if you take a look at this for a 70-watt heat source with two bends totaling 135 degrees, you’ll need a heat pipe with a Qmax of 108 watts.
Let’s take a look at how do we mount these things. The top section of this slide shows what we call clip style attachments. These are generally used in consumer products. Cost is very low. The downside of these is that the clamping pressures are often fairly low, so you have to be careful if you have TIM thermal resistances when you’re using low clamping pressures. The reason people use these is because they’re cheap and they’re easy to install. Pushpins are another attachment method. We generally don’t see those used for heat pipe assemblies, primarily just because of their strength. They’re fairly low pressure. They’re generally only good for smaller heat sinks. By far the most common attachment method used for higher powered loads are spring-loaded screws or some fastening mechanism where you can get higher pressures, 40, 50, 60 PSI. Those are quite often spring-loaded screws as are shown.
The heat exchanger fin design, everybody’s seen these pictures. You have extruded heat sinks. You have form skived heat sinks or forged heat sinks. You have stamped and folded heat sinks. The nice thing about the stamp and the folded design’s technology is it gives you a lot of flexibility. Tooling cost is often modest, a few thousand dollars. And obviously then, for the most flexibility, widest design flexibility, simply just machine the heat sinks out of solid or relatively close-knit shaped extrusion. A lot of high-end medical, military test equipment, low volumes, they just simply use machined heat sinks. But we find that the most flexible is the stamped fin, what everybody would call a fin pack or a zipper fin they call them sometimes.
A quick look at attachment methods. 90% of the time we just solder these things together. You take your aluminum if it’s an aluminum heat sink, you nickle plate it. Nickel and copper solder very nicely together, so the parts are quite often just soldered together. There are occasions, maybe 10% of the time where we’ll use a thermal epoxy to do the assembly. And here’s some examples showing the two different … We’ve got three pictures of soldered assemblies and some photographs of assemblies that have been epoxied together. The downside of an epoxy is, you’re going to get a little bit higher thermal resistance in that epoxy joint than you would a solder joint. Generally, it’s on the backside of the vapor chamber or heat pipe so it’s not a big hit.
We have to keep this thing moving, so let’s talk about performance modeling. Here we show a performance modeling accuracy curve. Generally people would say that if I’m doing a generic heat sink ballpark that my accuracy is going to be, let’s say, loose. Then as you move up in complexity on your modeling, you go to an Excel model or a CFD model, you’re going to get closer and closer to be able to predict the performance. Let’s talk about those methods for a minute. We like to do a heat sink ballpark calculation, and what I mean by that is if you’re sitting in a meeting and someone’s proposing a new design for a piece of equipment and they tell you how much power and how much space you get for your heat sink, you can within a couple of minutes find out how much trouble you’re in.
What we would normally do in a case like that is that we simply take the space that’s available and we calculate just the heat exchanger performance for that space. So for a given heat source, power maximum delta-t, the airflow, so what we do is we calculate the effective air temperature rise and the fin to air resistance, we add these two together and then we add what we call a fudge factor, just to see how close we are. Let’s give you an example.
So here’s an example of a ballpark calculation. This particular application at 130-watt heat load, maximum delta-t are 45 degrees. We knew what the airflow was, we were told it was 16 degrees or 16 CFM. If it was natural convection, that same application would turn out to be about three CFM. There’s the heat sink dimensions. So what we do is, we do a rough calculation of the effective air temperature rise based on that 16 CFM and it gives you 8.8 and a half degrees delta-t. So that’s simply due to your airflow.
Then we say, okay, if I’ve got this much space for my heat sink, we can do a rough calculation of forced convection, say 10 fins to the inch on the heat sink, calculate the amount of square fin of surface area, 2.78 square fin of surface area. You do your calculations on that surface area, it gives you 17 delta-t. So if I look at my air temperature rise and my fin to air delta-t, I’ve got 36 degrees C of temperature rise for this application. Now, I’ve added 10 degrees to that. That’s my fudge factor. So real quick, ballpark calculation says 36 degrees, my spec is 45 degrees. I’m okay, we don’t have to go back and take a look at changing the entire design of the piece of equipment.
We move onto an Excel model where you calculate almost everything in the whole thing. You can calculate your TIM two resistance, you can calculate your evaporation resistance, your condensation resistance, you calculate your fin delta-t, your fin to air delta-t and your air temperature rise. So almost every resistance that’s in there. That will generally get you to a much closer estimate of the performance. Using this model, we have a 39 degree C temperature rise.
I would say the next step is to throw it into one of the CFD. CFD packages are really nice, because it sort of gives you a flexibility to optimization, it also allows you to put other things in there, system-related things that may affect the performance. In this CFD analysis, it came out to be 37 degrees. And then we did a real test. The real test data turned out to be 31 degrees C. And there’s the part that we did a test. Now, so how did all of our estimates turn out to be, well, the ballpark came out to be 36 degrees, our CFD said 37 degrees and our Excel model said 39 degrees and the part itself was at 41 degrees. From an estimating standpoint all of these numbers are well within what we would normally see as far as considering it a good estimate.
And that’s all we have for today. Do we have questions?
Thank you so much George. Great presentation.
I hope everybody found that useful. So let’s move on to question time. We’ve got about seven or eight minutes available. We’ve had quite a lot of questions George. So I’ll try to-
Uh-oh. That’s good. I like questions.
… help everybody out and say if we don’t get to your particular question today, we will get all of your questions over to George and his team and we’ll post all of the questions with the answers on the Thermal Live site at some point next week. But let’s pick out a few here now. George, what’s the reliability of two-phase devices?
Oh, that’s an interesting question, since I just made a post on LinkedIn about that. There’s good news and there’s bad news when we talk about reliability. The good news is, these things last a really long time. We’ve got millions and millions and millions of them in the field without any significant signs of any failures. We have a product that’s been on live test 30, 40 years showing now change in performance. So they are reliable. The problem comes when someone in a big organization says, “I need a number. I need a reliability number.” That number doesn’t exist today, and we’re working on that. I can send you a whole bunch of information on reliability and what we’re doing and I’ll include that on what we send out or post later.
Okay. How about this one. Are these two-phase devices damaged by the freeze-thaw cycle?
They can be. In other words, if you look at any of the designs, we’re generally talking about water heat pipes, water freezes. When water freezes, it expands. If you have a puddle of any kind in any of these devices, that puddle will expand when it freezes, after about 10 cycles you have the possibility of rupturing the tube. On a centered wick, if you minimize the fluid charts, in other words, you don’t put in excess fluid, the fluid is contained in the wick structure. We generally test those things to 2,000 freeze-thaw cycles and we don’t see any change in performance. So they can be, but there are ways to design around it.
Got it. Great. Thanks. I don’t know if there’s an answer to this one. Can you say what are the minimum and maximum operating temperatures of heat pipes?
Sure. Here again, we’re going to stick to talking about water heat pipes just because that’s what’s used out there for these things. Obviously, water freezes at zero, so you’re not going to get much in the way of two-phase heat transfer at zero. It’s a function of the vapor pressure. In other words, at somewhere around five or 10 degrees C, you’ll get, it’ll be liquid. You’ll get a little bit of vapor pressure, but not much. You need vapor pressure to move the vapor. For most applications, we generally say you’ll get good performance starting at about 30 degrees C. So between zero and 30, you’ll get performance, it’ll just be degraded. And then 30 degrees C up to 150 or 200 degrees C you’ll get good performance. I don’t know if that was a clear answer, but 30 to 100.
Yeah, sounds good. Let’s move on to the next one. This attendee is asking why and when do I need to use CFD if my Excel appears to be fairly accurate?
What normally you do is you use CFD for a couple of reasons. One, you want to really play around with fin thickness and different alloys of aluminum and so you want to do some optimization. CFD will do a better job at that. CFD is also really good if you’ve got now other things before the heat sink, after the heat sink, other parts of the system that might affect the performance of the heat sink. You can’t really do that in an Excel model, so you really need to throw all that stuff into a CFD.
Okay. We’ve got time for just a couple more. This question comes from Viva [inaudible 00:41:21], who’s asking, in general is spreading more efficient than moving?
That’s a good question. I think no. The reason is, your biggest resistance in these devices is at the evaporator. So once you get the heat into the vapor, the vapor doesn’t know whether it’s spreading or moving. In other words, probably 80% of your thermal resistance is right at the evaporation and the rest of it, it has to do with sort of lengths and distances and bends. So no, not much difference between spreading and moving, as far as efficiency goes.
Thanks, George. The last question today comes from Sanjay Upasani, who’s asking, what type of thermal interface material is typically used between the heat pipe/vapor chamber and the device and what pressure is required for good thermal contact with the vapor chamber?
It’d depend on the market that you’re in and what you’re designing for. If you look at the tablets and notebooks and things like that, they generally use grease, but they generally have fairly low clamping pressures. So they have to use a good TIM, because they have low clamping pressures. If you look at applications where they have the ability to provide higher clamping, telecom, servers, that sort of thing where the pressures are 30, 40, 50 PSI, then generally, they hate grease. So they’ll go with some kind of pad. Quite often, it’s what they call a phase change pad, so it’s a piece of graphite with a phase change material on it or things like that. Still, they try to keep the pads relatively thin.
Well, we’re almost out of time. George, thanks for a great presentation. Thanks for answering as many questions as we could. Sorry-
Well thanks for the opportunity.
… if we did not answer everybody’s question, but if you did pose a question that we did not answer, look on the Thermal Live site next week and George and his team will attempt to answer them all. Attendees, if you have any more questions that we didn’t get to here or if you didn’t have a chance to pose a question, simply send us an email to firstname.lastname@example.org and we’ll include that in the Q&A. Coming up next at 1:30 PM Eastern Time in the United States, the presentation is Liquid Dispensed Thermal Interfaced Materials and this will be presented by Bergquist, and you can register for that right now at thermallive2015.com along with all the rest of the presentations coming later today and for the rest of this week.
To download this presentation, please visit https://thermal.live/2015/a-practical-guide-to-using-two-phase-heat-sinks/.