Thermal Design in Military Embedded Computing Applications

John: Welcome to today’s webcast brought to you by Military and Aerospace Electronics Online. Today’s event focuses on thermal design in military-embedded computing applications.

I’m John Keller, chief editor of Military and Aerospace Electronics magazine and I’ll be your host and moderator. Today we have two distinguished presenters Brian Muzyka, Sales Manager at Advanced Cooling Technologies in Lancaster, Pennsylvania, and Scott Garner, Vice President and Manager of the Electronics Products Group at Advanced Cooling Technologies who will join us for questions and answers.

This presentation is live and interactive so you can submit a question at any time by clicking on the “Ask a Question” button in the presentation window. After our panelists give their presentations I’ll open it up to questions from our online audience and for discussion among the panelists.

First, a couple of housekeeping details. If you are running pop-up blocking software such as the Google toolbar you will need to disable it to view this webcast. In addition, we recommend that you close down all other applications for better performance.

For your convenience this presentation will be available on demand within 24 hours of this live event. Additionally, the audio portion of this presentation could be available as an MP3 download. A reminder email message will be sent to all registrants with a link to the archive and will also be accessible from the homepage at www.militaryaerospace.com. So let’s get started with Brian Muzyka of Advanced Cooling Technologies. Brian?

Brian: Thanks, John, and thanks everyone for joining us. As John mentioned please feel free to type in questions at any time during the presentation. We’ll get to a couple afterwards and any that we don’t get to during the presentation we will answer in the next couple days via email.

We’re going to get started with a quick overview of the presentation. We’ll start with an overview of military embedded computing followed by a brief background on heat pipes describing the technology and the enhancements they can have to the military systems. We’ll then look at an overview of the thermal pack and hit on each major point along the path which includes heat spreaders for electronics cards, the attachment of the card to the chassis, and the card guide or the chassis itself. Finally we’ll wrap up the presentation and take any questions you might have.

So in general military electronics takes a lot of demanding challenges. Many of these challenges are coupled with the thermal design of the system. The most notable thermal challenge is the temperature requirement. Military systems must be able to operate in very hot ambient temperatures up to or even exceeding 70 degrees C. This is coupled with the limiting factor of electronics that typically have a maximum case temperature between 85 and 100 degrees C. This creates a very small thermal budget for design engineers.

Harsh environments create additional challenges including the need for ruggedization. Systems must be able to handle shock, vibration, and other various elements regarding harsh environments. Military standard 8-10 outlines a lot of thermal requirements in military systems. These systems can be deployed in mission-critical jobs and need to be durable, effective, and reliable. Also because most systems are on mobile platforms there’s a need for lightweight solutions.

The popular SWP acronym is vital in thermal systems as size, weight, and power are all key areas of design. Today we’ll talk mostly about embedded computer systems although the principles and design considerations are applicable across any electronic system. In embedded VME or VPX systems the electronics board that pass directly to heat spreaders which move the heat outward toward the chassis. The edge of the card is attached using a card retainer usually known as a wedge lock. This allows for easy assembly and quick servicing or replacing of the electronics boards. This connection transfers heat to the card guide or chassis, which then dissipates the heat with either liquid or air to the environment.

For liquid cooled systems, the liquid is typically located at the base of the chassis while air cooled systems have thin stacks surrounding any hot areas of the chassis. The thermal load, ambient conditions, and available air flow or liquid flow will typically determine the dissipation method.

Now in this figure we’ll look at the thermal challenges of a typical system. The figure here shows an air cooled chassis with a single card being slid into the card guide. The first thermal challenge we outline is spreading the heat along the card frame. Typically the card frames are aluminum for its weight and thermal properties.

If improved heat spreading is needed C\copper is an option but this will add weight to your system. As we continue to explore options that don’t add weight, we’ll get into high K plates which stick to the similar weight to aluminum but thermally out-perform them greatly.

The next critical point is the wedge lock. At this location we’ll discuss in more detail coming up. It becomes a thermal bottleneck of the system due to the large thermal resistance network. Heat is disproportionately spread into the chassis which causes a great temperature rise. A standard wedge lock will have nearly 80 percent go through the card frame directly to the chassis and only about 20 percent of the heat go through the wedge lock. This creates large temperature rises and a relatively short thermal path.

The final area we’ll look at is once heat enters the chassis. Again here it relies on conductivity of the chassis to move the heat to the ultimate heat sink. In liquid cooled systems you need to move the heat to the liquid which is typically isolated at the base of the chassis. It’s usually located here to prevent any leaks isolated from your electronics. An air cooled system, the idea is to spread heat to maximize your fin efficiency on those fin stacks.

Now that we’ve outlined the system, we’ll investigate improvements from normal systems. The focus in these solutions are heat pipes. We’ll begin with some background on technology and then continue demonstrate their effectiveness in these type of military systems. Heat pipes are a passive two-phase heat transfer device. They utilize the latent heat of a fluid to very effectively transfer heat across the lane.

Looking at the figure in the top left portion of the slide, the evaporator area will be placed beneath your heat generating electronic component. Heat is gathered and input into a heat pipe which causes the fluid to vaporize. The vapor then moves along the center of the heat pipe to a cooler region passively through the inherited pressure gradient.

At the colder region, the fluid condenses back to a liquid. The liquid is then pumped to the evaporated using a capillary action provided by a wick structure. Overall a heat pipe will have a 2 to 5 degree temperate rise across the length of the heat pipe. They can be used by itself or in conjunction with other metal components to increase system performance.

Any time conduction is a limiting factor in a system a heat pipe should be considered. In the case of embedded computing, moving heat to an external heat sink is a great example. Due to the densely packed cards in these systems there is no available room for local heat sinks which creates fairly long conduction paths. The increased conductivity with the heat path solution leads to isothermal surfaces which allow for lower electronics temperatures.

The biggest benefits are exactly what military electronics systems are looking for, SWP. Heat pipes can decrease the size and weight of the system with more effective heat transfer. Also by lowering electronic case temperature you can pack more power into your system for increased capability.

The final benefit we have listed there is also a fairly important one. Heat pipes are very versatile. They can be bent and flattened within your system. This allows for retrofitable designs which save a lot of cost and time that a full system redesign may incur. This is a critical part of embedded computing systems because the electronics layout and the stacking of the boards and the dissipation method is typically unchangeable once you go down a certain path.

The figures we have shown here represent good examples of the benefits. In the top figure heat pipes are laid into available real estate of a 6U VPX card to move heat from the electronics to the edge. We’ll go deeper into card cooling a little later but it’s easy to see how a heat pipe can easily be integrated for large thermal benefits.

The bottom figure has a heat pipe taking waste heat off the hot side of the thermal electric cooler. There wasn’t room right at that junction for a large enough heat sink so the heat pipe moves the heat away from that area to a space where you have more volume.

Now you’ve got a little background on heat pipes. We’ll look at utilizing the technology in an embedded computing system to increase performance. We’re going to start with a quick review of the overall thermal path.

The example here, again using the same example in the top corner, will serve as an illustration of the thermal path. The primary areas of concern are labeled ‘A’ which is the conduction cooled cards, ‘B’ which is the card chassis interface, and ‘C’ which is the dissipation off the chassis.

In this case it’s using convection to air. The first resistance within the card is moving heat to the edge with conduction. This is a fairly straight forward, but also critical component none the less. At the board chassis interface the card has a parallel resistance network moving heat into the chassis. In one path, heat travels directly to the chassis from the card with a simple interface resistance labeled here as R2.

In the other direction the resistance network consists of R3, which is the card and wedge lock interface, R4 which is conduction through the wedge lock, and R5 which is the wedge lock and chassis interface resistance. Because of this resistance network, it’s a big reason why you have maldistribution of heat going into your chassis and also have that thermal bottle neck at this interface.

Finally at the location C, the heat must conduct through the chassis as shown in R6 and it must be dissipated off the sidewall for air cooling in R7. If the chassis were liquid cool, R6 would be even more vital of a component as well as it comes a little later. You’d be basically moving the heat all the way down to the base and then R7 would be along the bottom surface there.

Now we’ll look at heat spreaders for electronics cards, and we’ll look at some typical options and then we’ll talk about high K or embedded heat pipe solutions to increase performance. Again to reiterate the thermal path in this section we’ll be looking at the spreading along the electronic card frame. These card frames rely solely on conduction to remove the heat.

The basic challenges, as most are aware, is the increasing power in the boards such as CPUs of FTGAs. Cards need to have a thin frame to fin to tightly packaged chassis which makes the thermal solution even more vital. These frames typically attach to the card guides with the wedge lock.

One of the ways to increase heat transfer into the chassis is making sure that the card edge has fairly even heat distribution. Having a concentrated heat load on one edge can cause issues down the thermal path.

As I mentioned before aluminum is most common for it’s combination of weight, cost, and thermal and mechanical properties. It has a thermal conductivity of about 180 W/mK. The next option that most designers look at is copper. In this option you sacrifice weight for thermal conductivity. Copper has a thermal conductivity of about 400 W/mK, but is over 3 times more dense than aluminum. The solution we’ll be discussing in detail is embedded heat pipe frames or HIK frames which has a thermal conductivity ranging from 500 to 1200 depending on geometry. It also has weight similar to aluminum as we’ll get into.

So continuing with HIK plates. HIK is a term for high conductivity, talking about plates embedded with heat pipes. They take the isothermal properties of heat pipes and embed them into standard aluminum plates with either an epoxy or a solder joint to increase the overall conductivity of the plates.

The heat pipes are strategically placed to get good thermal results while not effecting current geometry or mounting services that you might have in your system. Overall the heat pipes with the solder is similar to the weight of aluminum which makes the overall plate weight similar to aluminum.

So the result of the plate that is the same size, same geometry, same weight as aluminum but has a conductivity nearly three to five times greater. These plates can be used as structural components as well within the system. To illustrate the conductivity value we’ll look at the design in the bottom right corner. This plate had various high power electronic components mounted to it resulting in several hot spots.

So as you can see here with this thermal image the original aluminum plate had hot spots up to 91 degrees C. This was obviously greater than the desired case temperature of the system design so we looked at embedding heat pipes and coming up with a high-case solution to reduce those hot spots significantly. In this case the ultimate cooling was liquid cooling along the rail and the board was attached with wedge lock into that cold rail.

Looking at the figure to the right you can see that the overall result, the design reduced the temperature by over 20 degrees. You can also see the benefit of routing heat pipes along the rail which is a big driver in these conduction cooled boards. In the figure to the left you can see concentrated hot spots on the top that are fairly close to liquid cool badge. By simply using heat pipes to move the heat horizontally to the edge wouldn’t cause a large benefit, but by routing the heat pipes along the cold edge, the heat blocks out the condenser end is lowered which results in large thermal gains.

So now going back to the application of conduction cooled cards, the challenge here is to design the layout to create a cost effective, thermally enhanced design within a thin frame. With ACT designs we have gone as thin as 1.83 millimeters with a proprietary embedding technique that still allows for a transfer of significant levels of power. Design engineers should be cognizant of the importance of spreading heat not only to the edge, but along the edge as we discussed in the last slide.

A final note is the electronic standoff and mounting features can be avoided when routing heat pipes. Heat pipes can be placed from either side which will not affect those critical standoffs to get your heat into the frame and avoid large resistances with pads or greases. So the next section we’ll be looking at is the card to chassis interface. What we’ll do here is investigate the current methods and the new ACT dual condenser design to enhance the card to chassis interface. We’ll wrap up this section by showing some comparative tests of several approaches to illustrate the performance gain.

Again looking at the thermal resistance network you can see there’s a large number of resistances for a short thermal path associated with the wedge lock connection. The card is attached into the chassis with a wedge lock. The thermal resistance network through the wedge lock shown on the bottom of the parallel resistance network is much larger which creates that poor distribution as we’ve discussed before.

The resistances R3 and R4 are attributed to thermal resistance at the card to wedge lock metal to metal interface and conduction through the wedge lock. The proposed solution is to eliminate these resistances by moving heat directly above and below the wedge lock. This will allow for a near 50 to 50 split of the heat entering the chassis and lowers the overall temperature rise of this interface.

So to accomplish this goal the proposed concept is referred to as a dual condenser HIK plate. This is similar to a HIK conduction cooled card except the heat pipe extends out of the plate and integrates the condenser end of the heat pipe into a movable condenser block. The nature of thin wall heat pipes allow for this repetitive movement, the small repetitive movement of attaching the condenser block to the chassis. So what will happen is the wedge lock will then slide in between the condenser block and card and it will still be used as the mechanical force that presses the condenser block tight against the chassis.

Since the heat pipes are transferring the heat above the wedge lock there is no need for a heat transfer through the wedge lock which creates a large thermal benefit. A similar mechanical attachment to the chassis is achieved with this designed as compared to a normal wedge lock design.

So now to validate this concept we set up a repeatable test using a standard aluminum conduction frame, a HIK conduction frame, and the dual condenser HIK conduction frame. You can see the heat pipe layout in both the HIK and the dual condenser HIK. The goal again was not only to move the heat to the card edge but also isothermalize that area for better transfer into the chassis.

This slide looks at the test set up that we ran. The set up was basically designed to mimic how a card will result in a real system. We laid heaters in towards the center of the card and we took overall thermal resistance measurements that measured the delta T from the points shown in red here. Thermal couples were placed beneath the heater and in the heat sink itself.

So once you get those temperate readings you divide the delta T by the total power and you solve for the thermal resistance. So we ran one at this set up and we also looked at the interface resistance, which are shown in blue here. These will account for basically the interface resistance through that wedge lock interface.

So here is the graph for overall thermal resistance. You can see there’s a big jump just by going to a HIK solution and spreading that heat along the edge of the card. This creates a uniform interface for the wedge lock and drives down overall thermal performance. If that’s not enough going to a dual condenser design will shave even more temperature rise off your system by further reducing that interface resistance.

Now if we’re just looking at the card to chassis interface you can see the benefit of dual condenser design. There were large gains at this interface by going to a HIK but even much more significant by going to the dual condenser HIK. So going from a standard aluminum plate to a dual condenser can save over 7 degrees at 80 watts and that’s just this interface alone so it’s a very small portion of the overall thermal path. Sorry about that.

So the final step is looking at the card guides or the chassis itself. So here representing the C circled in this figure is where we’re going to be looking at, and we’ll be looking at both air cooled systems and liquid cooled systems for the chassis.

The two primary methods that we’ve been going over are air and liquid cooled chassis. In the liquid-cooled chassis the conduction was in the chassis down to the liquid base is the primary design consideration. When your current framework is not thermally conductive enough there are a couple options to allow for quick fixes which include implementing heat pipes directly into your chassis or card guide, or using an external bolt-on HIK pipe to increase conduction with no changes to your chassis.

The second method is air-cooled chassis. Here the main design consideration is spreading heat to gain fin efficiency and add surface area. Just as important in this design is making sure your fins are properly designed.

As a general rule of thumb for this a first step is making sure you have enough fin volume to meet your temperate rise goal. From there you can optimize the sizing and spacing by looking at your air flow or if you’re using natural convection. Then you can look at the conduction gradients and see if there’s a need for a HIK plate or any further integration to increase your conduction along the surface. When going to a HIK plate you get better fin efficiency by getting a more isothermal surface underneath the fins.

So I mentioned this briefly but this is a look at a bolt-on design HIK plate. So what we’re doing here is the bolt-on HIK plate will take heat move it rapidly in the direction of the heat pipes. So when you’re using this for a liquid base cooled application, for example, it will greatly reduce your conduction gradients and is a relatively easy thing to integrate onto your system.

It requires no changes or redesign of the chassis itself and it’s a relatively straightforward process because you’re basically using straight heat pipes to move the heat down. To demonstrate the additional benefit of a bolt-on HIK plate we simulated a 6-slot system, a 6-slot card guide. The breakdown of power is shown to your left and if you take a look at that you can see there’s a cluster of high power components in slots 3, 4, and 5.

For this analysis we used a liquid cooling temperature of 0 degrees and the goal is to limit the temperature rise to less than 35 degrees. This is just an overview of our comparative analysis. We used equivalently sized bolt-on plates for the comparison of aluminum versus HIK solution.

In the HIK solution the heat pipe integration is fairly simple from a design and manufacturing standpoint. It basically uses straight heat pipes and moves the heat from the electronic slots to the base very effectively. And this is the result of the thermal simulations we ran. Looking at the hot spot temperature in each case shows a 61 percent improvement from the base line aluminum to a HIK solution. The result in temperature rise decreased by over 50 degrees C, from 82 degrees C to 32 degrees C. For a retrofit design that keeps design costs low the result in thermal gain is pretty massive.

Now here similar to a bolt-on HIK design, this is an integrating heat pipe into the card guide itself. With the heat pipes accounting for the heat transfer often times you can reduce the weight by decreasing the thickness of the card guide. As I mentioned earlier, the solder integration and heat pipes combine for approximately the same weight as aluminum. Therefore any bulk metal that you remove is approximately the amount of weight you would save.

So going back to the analysis we ran in the figure before, if you were to achieve the results on the right with bulk aluminum you would add two inches to the thickness of the plate and that would account for over five pounds of weight to your system, and a lot of times these systems are set up with multiple card guides, chassis, and slots so you can really reduce your overall weight by considering this type of approach.

Now, I’m going to quickly wrap up and we’ll take some questions. As an overview some things we hope you learned a lot, benefits from HIK solutions, but to summarize first off military electronics are increasing in power. There’s little way around it. Even as electronics get more efficient, the demand for increased capability continues to drive higher powers on electronic boards. Thermal management is often at times a limiting factor with the ambient temperature and environmental concerns associated with military designs.

With a small thermal budget designs must be as efficient as possible. So to achieve those efficient thermal designs, high conductivity or HIK plates should be considered. They can benefit at all locations along thermal path including increasing performance in the conduction cooled card frame, the card frame and chassis interface by removing resistances associated with the wedge lock, and also at the card guide or card chassis itself.

So these solutions also provide customers with size, weight, and power advantages over a lot of similar options. Again we thank you for joining us and now we will be taking any questions that may have come in.

John: Okay. Thank you Brian and as he said now it’s time for questions and answers and comments from our audience. Scott Garner from Advanced Cooling Technologies will join us to answer some of the questions. So we’ve got quite a list of questions here, and the first one is how much power was on the module presented in slide 15? I realize you’ll probably have to go back to slide 15 for that.

Scott: Give us just a second. We’ll pull that slide up and see which one you’re discussing. Yeah, okay.

This was a large board so this wasn’t a standard 3 or 6 shoot board. This was about 18 by 12 inches roughly. A very long wedge lock joint to liquid cooled chassis, and I think this was dissipating somewhere in the park of about 400 watts total power into this. Again, double sided heat input into that HIK plate. Electronics were mounted on both sides of that conduction card.

John: Okay. Another question we’ve got is does ACT work with DARPA TGP heat pipes?

Scott: Yes, we were very active on that program. We were voted through DARPA to investigate CPE match and the whole goal of that program was to, TGP stands for thermal ground plane, so integrate almost within the card itself a plane just like an electrical ground plane, a plane that can be tied to a constant temperature and just cool electronics regardless of where they’re placed.

We designed a ceramic vapor chamber assembly that handled very high heat flexes on that program. We were partnered with UCLA and University of Michigan, and there’s information on our website regarding that program for anyone that wants to dig a little further. Or they can call and we can talk about it.

John: Okay. All right, another question. Can heat pipes still be effective when designed into a 3U or 6U conduction frame that gives access for one or more mezzanine cards?

Scott: Yes. Absolutely. It’s just a matter of how you couple, if you’re trying to pull the mezzanine itself you need to somehow couple that to the conduction frame which is pretty typically done anyway. If you’re just cutting out the conduction cord for access to that mezzanine board, as you can see from some of the hardware in the presentation we route heat pipes around cut outs, pedestals, pads, mounting holes, etc. in the standard conduction card. So it’s very flexible in the overall design.

John: Okay. Now you can please re-explain or perhaps cast some more light on how to eliminate R3 and R4 with dual conducting HIK plates?

Scott: Yes. So, if you go, Brian’s going to pull up the slide. If you look at the schematic in the center of this slide you’ll see the conduction card coming in and tying to the chassis and the wedge lock provides the force to mechanically attach that to the chassis and lock it in place. The bulk of the heat comes through the cord and comes through R2.

To get the heat out to the other side of the chassis or that large area up top you need to go through an interface to the wedge lock itself. You need to conduct the heat through the wedge lock and then a second interface between the wedge lock and the chassis. By using heat pipes to pull the heat over, if you click to the next slide, in this one you can see the red is the wedge lock and the heat pipe comes off the conduction card and goes to the top of the wedge lock and is in direct contact to the chassis.

So this splits the heat between conduction at the bottom and the heat pipe pulling its share of heat to the top and the wedge lock is no longer in the thermal path. So those two resistances are eliminated and heat flows equally into both sides of that chassis slot.

John: Okay. What is a cost impact using dual condenser HIK for 6U cards?

Scott: I mean, it’s certainly going to be more than a standard machine plate but for the thermal benefit it’s certainly a cost effective option. It’s going to depend on the number of heat pipes, the design configuration, we also do quite a bit of secondary work plating, heat coil installation, EMI gasketing. So it really depends. To give an exact price is challenging in this format.

John: Okay. I’ve got a question about spray cooling and that is how does your technology compare to direct spray cooling, and what would be some of the design tradeoffs and considerations?

Scott: It’s a very different piece. Spray cooling involves obviously much more complex systems. You need to spray directly onto the devices. You need to collect that. This is a passive system. Heat pipes can be integrated and bent to existing geometries and there’s no moving parts other than the wedge lock which is in there anyway. So it’s an overall, it’s much more passive. Much more integratable into existing configurations and you don’t need to worry about, and it’s also more reliable. You don’t need to worry about pumps and that’s kind of a contrast between the two.

John: Right. Well, and I have a question myself that’s come to mind. It seems to me that electronics engineers when they’re designing typically just think of ‘Okay, I’m just going to be a conduction cooled system or I’m just going to be a forced air system, or I’m just going to be a spray cool, or just a liquid flow through cooling, or even exotic air conditioners’ and things like that, but it seems like it’s putting all eggs into one basket.

If I was looking at your presentation correctly it seems that perhaps sometimes a better solution is to look at how we might blend some of these technologies. In the industry when you’re talking to your customers do you see some of that same mentality of ‘I’m only going to be conduction cooled. I’m only going to be air cooled. I’m only going to be liquid flow through cooling’ without consideration to how you might blend some of these technologies?

Scott: Yeah, sure. I mean, the goal typically is to do it as passively and as mentally invasive as you can. So if your electronics are blowing up power, that you don’t need anything other than conduction. That’s the way to go. If you’re looking at air cooled or liquid cooled chassis and you’re having issues getting the heat from the components on the board out to the edge of the sidewall, the HIK plates are certainly an option. We’re seeing them integrated and used across industry.

What I think it allows you to do is it allows you to go to higher powers, higher heat fluxes devices which everyone is pushing limits on, without integrating either spray cooling or liquid cooling in the board itself, which is always a reliability risk. If you have a leak then it’s in your electronics, where as if you’re liquid cooling your chassis and you have a leak your electronics are protected. What these heat pipes do is allow you to move the heat from your device out to your liquid cooling which is in the chassis and it allows you to run higher power components without converting to that flow through or spray cooling on the board itself.

John: Okay. What is the effect of different surface platings on thermal conductivity?

Scott: Minimal. It would have more of an effect at the interface and that’s almost so small, but conductivity is through the cross section of the metal. So if you have a tenth inch thick aluminum conduction card, whether you put it a few thousandths thick layer of nickel on it, that’s not going to have a big impact in the conductivity. It would have more impact at the interface and that’s usually also very small unless you’re using a coating like anodization which has, even though it’s thin it’s a ceramic and has a slightly higher impact on the thermal resistance.

John: Okay. Does gravity significantly impact performance of heat pipes? Does it have any influence?

Scott: Yeah, it can. You certainly want to design the heat pipe or your end solution to function and have sufficient margin in the worst case orientation. A 4 millimeter heat pipe at 6 inches long may do 50 watts fully gravitated where it will only do 30 if it’s fully against gravity, but if you design it for 25 then you’re safe in any orientation. So it may result in the addition of one additional heat pipe but you always want to make sure you design it for worst case operating condition and then you’re safe in all other orientations.

John: Okay. I’ve got a time to development question here. What is the time frame to develop a custom dual condenser HIK-type system plate.

Scott: It varies obviously with how far along the design is and how much prototyping and touching’s required. Typically we’ll do about a week, work with a customer for about a week on design and analysis, finalizing and requirements and then finalize the mechanics.

So that’s thermal analysis. Orienting the heat pipes, laying them out, matching the components, how many do we need and where they will lay in. Then maybe an additional week to finalize the mechanical drawings of the components. Then it’s typically a six to eight-week development time depending on the number of plating operations or how much secondary mechanical features we add to it.

John: Okay. I’ve got a fairly involved question that involved slides 29 to 31, so as I’m asking the question if you could perhaps bring those slides up.

Scott: Sure.

John: The question is what is required in terms of number of bolts at the bolted interface between the chassis and the bolt-on HIK plate so that the bolted interface does not become a bottle neck. And really the question specifically is what was assumed in your example around slides 29 to 31.

Scott: We built these for customers, and it really depends on the interface materials that you’re using and the heat flux that’s going through the wall. The good news is once you spread it out over that chassis it’s not a very large heat flux. Same amount of power but it’s spread over a much larger area.

So when you drive it into these cards and down it should be somewhat lower heat flux. I’m not sure what we used in the analysis specifically. I’m assuming a standard gap pad material with some bolting force because interface resistances are a function of pressure so it really depends on the interface materials to how many bolts are required obviously, but we certainly account for that in the analysis and the models.

John: Okay. I wanted to remind our online audience that you certainly can ask a question. We’ve got a little bit of time left. If you’d like to ask any questions of Brian or Scott just submit the question by clicking on the ‘Ask a Question’ button.

I’ve got a question here. Did these function over a wide temperature range? I mean, what kind of range are you talking about with your technology?

Scott: Yeah, certainly. We provide these primarily to military markets or working with military temperature ranges from minus 45 to plus 50 degrees C typically. The heat pipes and almost all of these are copper water heat pipes so the fluid will freeze below zero but when it’s that cold out you usually don’t need the enhancement associated with the heat pipes and just conduction to the metal itself is more than sufficient.

The fluid in the heat pipe is contained in the wic structure which has a very low modulist and acts as a sponge so when you do freeze there’s no mechanical damage to either the pipe or the assembly in any way, and as soon as it warms up above zero they’ll start functioning and transferring heat efficiently.

John: Well, that kind of raises a question in my mind and it has to do with the operating conditions and operating environment. Do you ever or often run across customer situations that something could be operating in a relatively cool environment that in some conditions might require more exotic cooling, but since it’s operating in a cool environment, maybe up in Alaska or something, that you can get by with fins and heat pipes rather than something more active. Is that every a consideration?

Scott: Sure. If your solution has a known fix position and it’s somewhere up north or that the temperature never gets up to the high end of these military specs, that’s always an option to design around that. A lot of what we do they don’t know the end use. It’s a vehicle of some sort, ship, airplane, ground vehicle, or even if it’s fixed electronics they’re not sure where it will be deployed in the world. So we always have to design for typically that worst case high-end ambient.

John: Okay. I’ve got a question and a common here. It says ‘This may sound simple but is there a strategy to develop the thickness of a simple conduction cooled plate? What are the considerations there?’

Scott: Yeah, if you’re relying on pure conduction and you have a high gradient from your high powered device that’s maybe located towards the center of that plate, you can do finite element or straight conduction analysis to determine how thick you’d need it to move the heat from that component to the edge where it’s cooled, and that’s using straight conduction equations for either aluminum or, as Brian said, copper has better conductivity but it has the mass penalty.

What you can see from some of the information from the presentation when we integrate heat pipes it allows us to maintain very thin lightweight aluminum structural components and get the benefit of embedding the heat pipe so we get much higher effective thermal conductivity.

And if you’re just playing around in your model and you throw a conductivity in of aluminum which might be just under 200 W/mK and you want to see what embedding heat pipes will do, you can throw a value of 600 in there and that should give you a good ballpark of what we could accomplish by embedding some heat pipes into that system.

John: Okay. So for your HIK plates what are they typically made of? Is there a wide variety of materials to consider or are they typically just made of one type of material?

Scott: Most often because we get the thermal benefit we don’t need to use copper and we’d use aluminum which saves the weight in the end system, and we embed copper water heat pipes either through soldering or epoxy.

John: Okay. Do you see a great deal of difference between the performance of a three-piece wedge retainer versus a five-piece wedge retainer?

Scott: Yes. There is, and it’s also the difference between quarter inch and three-eighths. They have various thermal performances for the variety of different designs.

John: Okay. I have a reliability question here. What is the life cycle of your heat pipes in terms of meantime between failures?

Scott: Yeah, that’s a difficult question. There’s a reliability guide on our website that we have. We don’t have good meantime between failure data because we just don’t have enough failure to put a solid number together. Typically if there are issues in the field it’s mechanical issues from some of the secondary things. Either plating or [helicoil] screws. Things like that. The heat pipes themselves are very reliable. Particularly copper water assembly. It’s the most proven heat pipe material system and there’s life tests and reliability data all over the place on those systems.

John: Okay. I wanted to remind the audience that we still have a little bit of time left if you would like to answer a question. Any kind of question of our two panelists Brian and Scott you can ask any question you like just by clicking on the ‘Ask a question’ button and they would be happy to answer those questions. One here says ‘What’s the nominal heat pipe diameter being used in embedded heat pipes?’

Scott: Mostly what we’re doing in 3U and 6U conduction cards is either going to be a 3 millimeter or 4 millimeter diameter heat pipe. However we typically end up flattening that into the plate itself. The thinnest assembly we have to date is just under 2 millimeters. About .072 inch thick assembly. That’s very tight. Typically they’re more on the order of a tenth inch or above in these aluminum spreader plates and we flatten the 3-millimeter heat pipes to fit within that web thickness.

John: Okay. A question I have is in terms of the technology offerings that ACT has on the drawing boards what can we expect from ACT in terms of any imminent technology announcements and where do you see particularly heat pipe technology going in the next say, two to five years?

I mean, what kind of technologies and capabilities in cooling will we see in the not-too-distant future versus what we have now and, you know, with those new technologies what types of capabilities would you expect particularly in terms of some of the new microprocessors like the fourth generation Intel Core i7? I mean, what’s on the horizon?

Scott: Well, I think one of the things that everyone’s looking at is thinner, higher power capability heat pipes. Custom width designs that have increased pumping capabilities so you can handle higher powers and smaller footprints or higher heat flux into the device itself certainly was touched on with the thermal ground plane concept that DARPA put quite a bit of money into in funding.

I think there’s been some great strides forward in that technology. CTE matching of the heat pipes and vapor chamber assemblies themselves where you have the same coefficient of thermal expansion as the electronics so you can do a direct di attach, increasing that m1 interface. Things like that.

If you go to our webpage we have a very broad and diverse scope of technologies that we’re working on. Way too much for me to go into any kind of detail here but feel free to go to our webpage and click around. If you have any questions on any of the technologies there we’d be glad to discuss them with you in more detail.

John: And the URL to your webpage is what?

Scott: www.1-act.com

John: Okay. Terrific. I see that we don’t have any additional questions. Is there anything that you would like to say to wrap up or is there anything that I or the audiences really forgotten to ask that’s most pertinent to embedded computing cooling?

Scott: No. We appreciate your questions and we appreciate everyone’s attendance. If anybody has any questions later on feel free to shoot Brian an e-mail and we’ll certainly be in touch.

John: Okay, and is Brian’s e-mail address, what is that e-mail?

Brian: My e-mail is, if you go to the website, it’s my name Brian dot Muzyka. So you can get that information there, and I believe that’s also on the military aerospace website, and it’s at one dash ACT dot com, just like our website.

John: Okay. All right, so on behalf of Military and Aerospace Electronics and Pennwell, I would like to thank Brian Muzyka, sales manager at Advanced Cooling Technologies, and Scott Garner, Vice President and manager of electronics products group at Advanced Cooling Technologies.

This presentation will be archived within 24 hours and you can access it from the homepage at www.militaryaerospace.com. I’m sorry. A reminder e-mail will go out to everyone who registered for the webcast which will give you access to an archive that has the presentation so you’ll be able to review these presentations and have those for your files. So we thank you for joining us today. I’m John Keller, chief editor of Military and Aerospace Electronics. Thank you very much.