Cooling 400 W boards in AdvancedTCA (and other platforms)

In recent years, AdvancedTCA, along with many other industry sectors, has seen the need for more powerful processors. A growing demand for greater processing power brings with it higher thermal dissipation. In this, the first in a series of tutorial columns, David Wright, AdvancedTCA, ATCA300, and MicroTCA Systems Architect at Advanced Managed Platforms, explores design requirements, qualification testing, and life cycle issues. He will focus on the substantial number of convection-cooled products, both in the field and planned for deployment.

In recent years, AdvancedTCA, along with many other industry sectors, has seen the need for more powerful processors. With this ever-increasing demand for greater processing power comes the inevitable side effect of higher thermal dissipation. AdvancedTCA is by no means alone in allowing boards up to 400 W dissipation and above.

This tutorial series will cover design requirements, qualification testing, and life cycle issues. Although much exciting and chal-lenging work addresses liquid cooling, these tutorials will focus on the substantial number of convection-cooled products, both in the field and planned for deployment. Applications continue to benefit from the simplicity and low cost of convection cooling. The aim of these articles is to formalize much of the wide amount of existing expertise in convection cooling in the belief that even 400 W should not be considered a ceiling.

We will set baselines for system design. In addition, we will look at common terms and definitions and remove variables to make life simpler for board, shelf, and system designers and integrators.

Three heat transfer mechanisms – conduction, convection, and radiation – govern equipment cooling. In practice radiation can be ignored, as it depends on a predictable temperature difference between objects; there are many cases in which that predictable temperature difference cannot be relied upon.

The disciplines and tools for conduction cooling are relatively straightforward when applied to the heat transfer path from heat sources (e.g., silicon) to the dissipative structures (e.g., packaging and heat sinks) that will transfer the heat energy into the wider environment. Conduction is the mechanism that takes the heat energy in solid materials and passes it to the gas molecules that are the main mechanism of convection heat transfer. Convection cooling is the most effective way of moving heat out of a system. Forced convection using air movers (i.e., fans or impellers) is (almost) essential for high-power systems but requires the complexities of fluid dynamics involving both thermal and airflow boundary conditions as well as the solution of the Reynolds number and the dimensionless Prandtl and Nusselt numbers. Many of these complexities are handled by modeling, but the user should be aware of their significance at the higher flow rates needed for higher powers.

One of the significant disadvantages of forced convection cooling is acoustic noise caused by the fans and the airflow across the boards and through the shelf enclosure. Managing noise is a system issue and requires close co-operation between shelf and board design engineers and the system integrators.

For any custom or standards-based project, the boards must be designed to promote adequate cooling within the given shelf environment. Equally important, the shelf needs to be designed to cool the expected boards residing within its card cage. These specifications need to be in place “up-front” and must cover all of the requirements before any design work can be carried out. PICMG 3.0 was released in December 2002, and the current Revision 3 was released in March 2008. The cooling requirements covered in Section 5 and the subsequent additions and reviews throughout this time period demonstrate the considerable effort and expertise invested to ensure a viable cooling specification.

From component to office

Thermal engineers work with several cooling definitions when moving from component cooling to the shelf, rack, and office. Component cooling is molecular in nature, as energy is passed from the device to a fluid, usually air. The cooling air velocity is usually given as Linear Feet per Minute or LFM (the “linear” is superfluous). The heat energy moves to the air, and the cooling capacity of the air is used to determine a volumetric measure of Cubic Feet per Minute (CFM). However, it is the molecules that move the energy, hence it is the fluid mass flow that needs to be determined. Fluid mass is independent of air temperature, pressure, and humidity. Table 1 shows the commonly used measurement units.

We will demonstrate in these tutorials how to work with com-ponent velocity (LFM) cooling requirements, shelf volumetric requirements (CFM), and varying air characteristics.

Providing adequate cooling of a shelf populated with 400 W boards will require us to revisit some fundamental principles. We will look carefully at the fluid dynamics involved at the component, board, and shelf levels. In addition, we will examine the effects of temperature, humidity, altitude, and weather, and how these effects can be covered by more rigorous yet simpler requirement specifications. We shall review the acoustic noise implications, discover what causes airflow to be noisy, and examine how this noise can be reduced at the board and shelf levels.

Thermal management is a key compo-nent of the system integration process. This process can be time-consuming, with many test cases needing to be simulated as part of the system-level design verification test protocol. The specifications previously mentioned, and the subsequent steps of thermal conformance testing, are all necessary to ensure adequate cooling performance and will be discussed in more detail.

Moving the bar above 400 W

Interoperability conformance testing of individual AdvancedTCA elements is now an active development area. This started with the “AdvancedTCA Backplane SI Test Plan” in March 2004 and now covers several interoperability areas. We will look at some of the thermal interoperability issues, particularly where they apply to boards with extended power dissipa- tion levels.

The series of articles will conclude with some solutions that will be compatible with both AdvancedTCA and other standards and specifications, as well as existing and planned convection-cooled products and could “move the bar” considerably above the perceived 400 W ceiling.

David Wright has been involved in AdvancedTCA since 2001, and in practical electronics and mathematical modeling for much longer. After working for multinational companies GEC Ltd., MEL, a division of Philips NV, CTS Corp., and Hybricon, Inc. David operated as a consultant for nine years before co-founding Wickenby Ltd. in Israel. After significant investor involvement the company has been renamed as Advanced Platforms Ltd. David functions as the AdvancedTCA, ATCA300 and MicroTCA Systems Architect. Advanced Platforms Ltd. provides third-generation products and services.