Technology Moves the Line Between Convection and Liquid Cooling
When deploying conduction-cooled 3U or 6U boards (including VME64X, VPS, and VXS) in air-cooled environments, traditional heat sinks and fin structures are limited in the amount of heat they can move through the cold plates. Above approx. 100 W per board, liquid or unreasonably low ambient temperatures are required to maintain board temperatures within device limits, and this is confirmed with CFD models. A new cooling system architecture increases the available heat exchanger capability well beyond traditional designs, to over 150 W per slot. This extra cooling capability extends the power range of air-cooled environments, enables technology refresh in legacy systems, and can be traded for a lower operating temperature in lower power boards, increasing the reliability of COTS devices in military environments.
The past few decades have seen explosive growth in the capabilities of electronic circuits and systems. Even with the proliferation of CMOS technology, the power density of electronic systems continues to increase, particularly in the area of high-performance military electronics such as signal processing. Through it all, there is one constant in electronic systems packaging - every Watt of power that enters an electronic system has to go somewhere, and most of it is converted to heat. Conduction cooling adds a complication to the thermal management system because all of the heat from each board must flow through two relatively small cross sectional areas. This restriction is defined by the mechanical parameters of a card cage for conduction-cooled boards, as spelled out in open standards. With the typical environmental requirements of military-deployed systems, this has limited the allowable power dissipation to approx. 100 W per board in an air-cooled environment. Above that, the usual solution is to use liquid cooling, with all of its attendant complexities. Many times, conduction cooled boards are required because the system is deployed in a harsh environment, where the ambient air is contaminated with salt fog, jet fuel vapors, or some other unfriendly agents. Cooling system designs must maintain the integrity of the protected inner environment while dealing with steadily increasing heat loads.
The problem everyone faces is that the geometry at the board/cage interface, as defined in IEEE 1101.2, VITA 46, or VITA 48, is very restrictive to heat flow. CFD software models confirmed that there is a practical limit to the performance of a conduction-to-air heat exchanger. We modeled a three-slot slice of a typical 6U x 160 mm conduction-cooled VPX card cage (see figure 1 at right) based on common initial conditions: transverse fins, 1.0” board pitch, 1000 lfm air velocity, and 0.5” H2O back pressure. 1.0” board pitch is based on indications from our customers, 0.5” back pressure represents what is achievable without employing exotic fan structures, and 1000 lfm is a typical upper limit of performance gain for forced-air convection cooling. We ran through various combinations of fin height, pitch, and thickness, and plate thickness (between the bottom of the board slot and the surface with the heat exchanger fins), and determined the limits of a "theoretically perfect" heat exchanger. Increasing the parameters beyond these theoretical limits produced very little real improvement in heat exchanger performance, as indicated by a useable increase in cooling capacity or a significant decrease in the temperature of the cold plate at the board/cage interface.
Further analysis indicated that to achieve a new level of cooling performance, such as 150 W per slot, we needed to somehow create 40% more heatsink fin area without increasing back pressure or system size. When the heat exchanger is an integral part of the card cage (as is common for best efficiency), the size of the card cage limits the size of the exchanger. To go beyond these limits, Tracewell Systems developed Remote Heat Sink technology (RHS), and was recently awarded patent number 7,460,367 on the design (http://patft.uspto.gov/).
RHS increases the cooling “range” of a conduction-to-air heat exchanger to as much as 150 to 200 Watts per slot, depending on the specific environment and system parameters involved. It is a novel way to move heat to unused spaces within the system enclosure, such as behind the backplane or beyond the end card slots. Even in a chassis with I/O cables connected directly to the backplane (possible in VME64X and VXS, for example), the area behind the system bus (J1) connectors usually is unused space that RHS can use. With RHS, high-efficiency heat pipes are embedded directly into the conduction-cooling frame of the card cage, and transport system heat to an additional heat exchanger in a remote location. Figure 2 at right shows one example, where the heat pipes wrap around the backplane in a top-load VPX system. Under the backplane is an enormous volume, filled with a heat spreading plate and over 30 square feet of additional fin area for heat exchange. Note that this air space is completely isolated from the electronics area. This maintains the sealed integrity of the system electronics area.
Aluminum is an excellent conductor of heat, and removing material to make room for the heat pipes decreases the conductivity of the normal heat path. However, the thermal conductivity of heat pipes is over 40 times that of aluminum (and over 25 times that of copper), and more than makes up for the conductivity of the removed material. Also, the exact location of the heat pipes within the board/cage interface regions is designed for an optimum balance between side and remote heat exchangers. While dissipating 140 W per slot, the system illustrated in Figure 2 demonstrated a Δ-T of less than 20ºC from the board/cage interface to the external ambient air stream.
Tracewell’s Remote Heat Sink technology brings several advantages to card and system designers. While the most obvious is that new cards with higher power levels can be deployed in air-cooled environments, there are other applications. When upgrading electronic systems in legacy platforms, frequently there is not enough cooling air to support higher power levels, and retrofitting the platform with liquid cooling is not an option. RHS can enable deployment without scaling back performance to reduce power.
RHS can mitigate some of the risks inherent in the rapid deployment of new technology. When dealing with new platforms such as UAVs, the amount of available cooling might not be known when system architecture commitments are made. RHS can preserve schedule by preventing a platform change late in the project cycle from forcing a re-design or “dumbing-down” of a critical sub-system.