Effect of both PCB thermal conductivity and nature of forced convection environment on component operating temperature: Experimental measurement and numerical prediction

Abstract

The steady state thermal performance of two package types is experimentally characterised on up to four different thermal test Printed Circuit Boards (PCBs) operating within two different forced convection environments. Both standard and non-standard test PCBs and test environments were used so that component thermal performance under standard test conditions could be compared with that measured under non-standard conditions that mimics those found in service. The design of the test boards spanned from the low and high conductivity JEDEC standard, FR4 test PCBs to typical eight-layer application boards, whereas forced convection airflows were generated by both a windtunnel, representing the standard test environment, and upstream axial fans that are typically used in applications. All test components contained a built-in thermal test chip, mounted within an SO-8 and SBGA352 package types, and junction temperatures were measured for power dissipation levels of 0.5 and 6 Watts respectively. Windtunnel generated air velocities ranged from 0.5 m/s to 5 m/s, whereas flow rates corresponding to mean velocities of 0.5 m/s to 2 m/s were delivered by a set of four upstream axial fans. While junction-re-ambient thermal resistance (*ja) data for the SO-8 package showed a sensitivity of up to 48% to PCB construction, differences of up to 11% were also recorded between both test environments. The experimental data not only helps identify the sensitivity of component thermal performance to both PCB construction and operating environment, enables one assess the applicability of standard thermal resistance data for design purposes on non-standard PCBs in forced convection fan flows, but also provides benchmark data to assess numerical prediction accuracy in these non-standard but application-type environments.