Experimental validation of numerical heat transfer predictions for singleand multi-component printed circuit boards in natural convection environments

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

“…Following a previously described technique, an in-situ calibration [13] was performed using a water-cooled, cold-plate centrally located in the wind-tunnel test section. For the zero flow condition, the IR camera was calibrated over the temperature range 25°C to 95°C using a thermocouple with a calibrated accuracy of ± 0.1°C attached to the cold-plate surface.…”

“…These models were generated using a CFD based tool, Flotherm version 2.1 from Flomerics [28], which is a dedicated software used for thermal analysis of electronic systems. Based on evidence from previous forced convection studies on similar singlecomponent test PCBs [12] which showed less than a 1°C difference between predictions from the laminar and revised k-ε turbulent flow models available with this software, results from just the laminar flow model are presented in this study. The modelling strategy employed to ensure that predictions were both computational domain and grid independent are outlined by Rodgers et al [29].…”

“…Based on evidence from previous forced convection studies on similar singlecomponent test PCBs [12] which showed less than a 1°C difference between predictions from the laminar and revised k-ε turbulent flow models available with this software, results from just the laminar flow model are presented in this study. The modelling strategy employed to ensure that predictions were both computational domain and grid independent are outlined by Rodgers et al [29]. Figure 8 shows the half-geometry model in forced convection test environment presented in Figure 4 and Figure 9 presents a close-up cross sectional view of the component model.…”

“…While these tools have the capability to assess complex thermal designs in a timely manner, previous studies have identified the ongoing need to validate numerical prediction accuracy and applicability of modelling methodologies by comparison with experimental analysis [7,8]. However, where such comparisons have been undertaken [9][10][11][12][13], these have focused mainly on characterising component performance on low conductivity test PCBs based on the recognised SEMI (Semiconductor Equipment and Materials International) or JEDEC (Joint Electronic Design Engineering Council) specifications [14,15]. Both standards specify single copper trace layer PCBs, for which a suitable modelling strategy exists [9][10][11][12][13].…”

“…However, where such comparisons have been undertaken [9][10][11][12][13], these have focused mainly on characterising component performance on low conductivity test PCBs based on the recognised SEMI (Semiconductor Equipment and Materials International) or JEDEC (Joint Electronic Design Engineering Council) specifications [14,15]. Both standards specify single copper trace layer PCBs, for which a suitable modelling strategy exists [9][10][11][12][13]. However, both SEMI and JEDEC recognised that these low conductivity test PCBs no longer represent the multilayer builds currently used in many applications and recently introduced high thermal conductivity standard test PCBs [16,17].…”