Posted on 01 June 2019

Cooling Power Conversion Designs

Free Bodo's Power Magazines!




The PCB has become a significant thermal path.

Power density is increasing rapidly, but assiduous thermal management can help designers meet the many system-level demands at the best possible price.

By Nico Bruijnis, European Marketing Manager, The Bergquist Company


With more silicon and more physical channels squeezing onto each circuit board, there is pressure on circuitry such as power supply components to reduce in size while at the same time providing more Watts to drive the additional functions designers want to include.

Power Architectures are evolving

Power architectures are continually evolving to deliver the best achievable characteristics, not only in terms of physical size but also operating efficiency, manufacturability, vendor independence and upgradeability. This process is driven by changes such as the ongoing trend towards low voltage logic in processors, FPGAs and ASSPs on each of the boards.

For instance, the Intermediate Bus Architecture (IBA) emerged to reduce the overall losses incurred as the bulk DC voltage, usually a nominal 48VDC, is stepped down to very low levels such as 2.7VDC or 1.8VDC for low voltage logic. Other groups, such as the Distributed-power Open Standards Alliance (DOSA) and Point of Load Alliance (POLA) prioritise the establishing of standards covering power converter form factors, footprints, feature sets and functionality. The goals include reducing development time and vendor dependence, and enhancing the scalability of power architectures. However, given with the perceived advantages of standardisation, many engineers still prefer to “roll their own” architecture, using off the shelf modules or by bringing up a proprietary power supply design.

Ever smaller form factors

A common trend is toward smaller form factors, particularly for board-mounted circuitry or modules such as non-isolated POLs, which occupy premium real-estate that designers need in order to implement additional system level functions. Although POL efficiency is quite high, typically around 95% or better, the demand for more power from smaller modules means that modern POLs dissipate appreciable quantities of heat. For other modules, such as isolated converters that convert the 48VDC into a primary or intermediate supply – which may be regulated or unregulated - open frame designs now dominate.

Bergquist Sil-Pad Applications

Improved magnetics and power semiconductors enable high power density. For example, where the industry standard bricksize form factor used to call for a half-bricksized isolated converter, quarter-brick, eighth-brick and even sixteenth-brick converters are now capable of supplying upwards of 40A at voltages as low as 1.0V, and at over 90% efficiency.

The drive to miniaturise distributed power modules is actually a mixed bag for the power designer seeking to optimise the thermal performance of the power supply. Even though the available surface area for cooling is reduced, the thermal paths from the devices to the surface of the case are shorter, resulting in lower thermal resistance paths to ambient.

Whether encapsulated in a module, contained in an open-frame converter, or mounted discreetly on the board, it is the semiconductor and magnetic components that produce the most heat. Overheating an electromagnetic device can lead to insulation damage, greater losses, and a reduction in magnetic energy storage capability by hastening the onset of saturation. Consequences include the potential for severe damage to DC-DC circuits. Operating a semiconductor component at an elevated temperature accelerates failure mechanisms that can lead to parametric failure, where the device fails to perform in accordance with the manufacturer’s specified parameters. Exceeding the maximum temperature recommended by the manufacturer will likely lead to catastrophic failure of the device.

Maximising heat dissipation

The heat generated within power semiconductors or magnetic components can be conducted efficiently away to the surface of the device package or, for a module, to the housing. The heat can then be removed by convection and radiation, aided by forced air cooling, if necessary. A heatsink may also be fitted, enabling higher power output for a given ambient temperature or airflow rate.

Although an off-the-shelf module may be designed for no-heatsink operation if the rated power is to be below around 30W (subject to conditions including minimum airflow and maximum ambient temperature), vendors of larger modules may offer customdesigned heatsinks that attach conveniently to the module casing or a baseplate. The heatsink may be supplied as part of a complete kit comprising appropriate screws to attach the heatsink, as well as a thermal interface material to ensure efficient coupling of heat from the module case or baseplate into the heatsink. In proprietary power conversion designs, also, a thermal interface material must be inserted to maximise conduction of heat into the heatsink.

When the interface material is supplied as part of a kit, it is often provided as a precut sheet of composite material constructed to have high thermal properties. This may be a mechanically tough fibreglass or kapton carrier, for example, coated on both sides with a high thermal conductivity film. When pre-cut to match the exact shape of the heatsink’s mating surface, this is very convenient for the assembler. Such materials can also be packed, stored and transported easily. The Bergquist Sil-Pad range provides an example of these types of material. They are popular with power module OEMs as well as with engineers bringing up proprietary designs, and can be delivered in a number of convenient forms by arrangement. The thermal conductivity of such materials ranges from 0.9W/m-K to 3.0W/m-K.

Bergquist Sil-Pad pre-cut on tape

Phase change materials are the perfect alternative to thermal grease

Another easy-to-use technique for thermally linking a power module to a heatsink is to use a phase change material. This is an easier-to-use alternative to thermal grease, which is not convenient to apply and can result in variability since it is difficult to control the thickness of the applied layer. A phase change material is non-tacky at normal temperatures, and is normally coated onto both sides of a stiffening material such as a polyimide sheet.

This allows the material to be readily diecut as well and also facilitates handling in manual or automated assembly. When heated by the normal operation of the power device the phase change material then begins to flow, and wets-out the junction between the casing of the module or device and the heatsink. This creates a low thermal resistance path from the device to the heatsink. The phase-change temperature is typically around 55°C. Other available options include reinforcing and electrical insulation, if required.

Guidelines published by Bergquist for using phase change materials such as its Hi-Flow series recommend to physically attach the heatsink to the device by using a fastener such as a clip, to ensure a constant mating force over time. Screws may also be used, but an alternative is to use a thermally efficient adhesive tape such as Bond-Ply. This is a pressure-sensitive tape that also provides for the decoupling of bonded materials with mismatched thermal coefficients of expansion.

The PCB has become a significant thermal path

Dissipating heat from power modules or components into the host PCB is also becoming an important cooling opportunity for designers seeking increased power density and outright power handling capability. In fact, some vendors of board-mounted power modules are now positioning low thermal impedance terminations as a significant extra benefit for designers who are configuring power architectures. The host PCB is becoming a significant thermal path for board mounted power modules.

Some module vendors are paying close attention to thermal design inside the module, for example to minimise hot spots and achieve a more homogeneous module temperature. For example, thick copper layers in the module PCB, or an insulated metal substrate (IMS), can efficiently distribute the heat generated by semiconductor and magnetic components to make best use of the module packaging to couple heat into the heatsink and/or surrounding atmosphere, and the host PCB.

As an alternative circuit board technology, an IMS such as Bergquist T-Clad features an aluminium substrate that is insulated from the circuit layer by a thermally enhanced dielectric. The substrate provides a heatsink of high thermal capacity to absorb the heat generated by board-mounted components. IMS can eliminate the costs of additional heatsinks and associated mounting hardware, and also enables surface mount assembly.

Bergquist T-Clad

Best Practice for Cost-Effective Thermal Performance improvement

It is true that modern power architectures, which aid the distribution of low voltages around the system, have also allowed dissipated heat to be distributed to several physical locations within the system. In addition, improved power semiconductors and magnetic components, as well as enhancements to power design such as increasing switching frequencies, have reduced conduction and switching losses. These factors, too have contributed to progressively higher power outputs in exchange for only incremental increases in cooling airflow, for example.

Bergquist T-Clad Applications

However, designers are constantly under pressure to increase power density. Highperformance components or modules, or complex high-frequency designs, can add to the cost and design time associated with a power conversion solution. Increasing the specification of the cooling fan, to provide greater airflow, also adds to the cost of the delivered product.

When seeking enhanced power output, the first port of call should be to ensure optimum thermal coupling between major heat sources such as converter modules or discrete power semiconductor and magnetic components and the surrounding atmosphere. The use of thermal interface materials to fill air gaps and create low thermal resistance paths away from the active components is the most cost-effective starting point.



VN:F [1.9.17_1161]
Rating: 0.0/6 (0 votes cast)

This post was written by:

- who has written 791 posts on PowerGuru - Power Electronics Information Portal.

Contact the author

Leave a Response

You must be logged in to post a comment.