Abstract
Discrepancies between estimated and actual thermal solution performance may negatively bias early system design decisions and ultimately result in costly redesigns. In particular, an overreliance on manufacturer-provided performance data and insufficiently accurate prediction methods are two potential causes of large discrepancies. To demonstrate deviations between predicted and end-thermal-performance, three methods of predicting thermal performance were investigated. The reference design was a forced convection air-cooled computer chassis, consisting of four printed circuit boards (PCBs), each populated with four processors that utilize heat sinks. The three predictive methods used were rudimentary analytical calculations, spreadsheet based modeling, and computational fluid dynamics (CFD) simulation. The results from these methods were compared to measured results taken from the physical hardware test bed. A thermal resistance network model was used to describe the problem, and the comparison between the measured and calculated results for each element of the model was reported as a percent error. Note that in cases in which physical measurements could not be obtained, the comparison was made between the CFD results and the other calculations. All predictive methods using manufacturer-supplied thermal interface material (TIM) data indicated adequate thermal margin. When corrected TIM values were used, both the hand calculation method and the spreadsheet modeling method indicated adequate thermal margin; however, the CFD analysis predicted significant thermal issues, which were observed in the thermal test bed measured results.
| Original language | English (US) |
|---|---|
| Title of host publication | 34th Annual Semiconductor Thermal Measurement and Management Symposium, SEMI-THERM 2018 - Proceedings |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| Pages | 119-126 |
| Number of pages | 8 |
| ISBN (Electronic) | 9781538644027 |
| DOIs | |
| State | Published - May 9 2018 |
| Event | 34th Annual Semiconductor Thermal Measurement and Management Symposium, SEMI-THERM 2018 - San Jose, United States Duration: Mar 19 2018 → Mar 23 2018 |
Other
| Other | 34th Annual Semiconductor Thermal Measurement and Management Symposium, SEMI-THERM 2018 |
|---|---|
| Country | United States |
| City | San Jose |
| Period | 3/19/18 → 3/23/18 |
Fingerprint
Keywords
- error
- thermal interface material
- Thermal resistance
ASJC Scopus subject areas
- Hardware and Architecture
- Electrical and Electronic Engineering
- Electronic, Optical and Magnetic Materials
- Fluid Flow and Transfer Processes
Cite this
Comparison of junction temperature prediction methods through analysis of isolated 1-D thermal resistance model variables of an application utilizing forced convection of heat sinks. / Klitzke, N. A.; Polzer, S. C.; Wilkins, W. L.; Gilbert, Barry Kent; Haider, Clifton R.
34th Annual Semiconductor Thermal Measurement and Management Symposium, SEMI-THERM 2018 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2018. p. 119-126.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Comparison of junction temperature prediction methods through analysis of isolated 1-D thermal resistance model variables of an application utilizing forced convection of heat sinks
AU - Klitzke, N. A.
AU - Polzer, S. C.
AU - Wilkins, W. L.
AU - Gilbert, Barry Kent
AU - Haider, Clifton R
PY - 2018/5/9
Y1 - 2018/5/9
N2 - Discrepancies between estimated and actual thermal solution performance may negatively bias early system design decisions and ultimately result in costly redesigns. In particular, an overreliance on manufacturer-provided performance data and insufficiently accurate prediction methods are two potential causes of large discrepancies. To demonstrate deviations between predicted and end-thermal-performance, three methods of predicting thermal performance were investigated. The reference design was a forced convection air-cooled computer chassis, consisting of four printed circuit boards (PCBs), each populated with four processors that utilize heat sinks. The three predictive methods used were rudimentary analytical calculations, spreadsheet based modeling, and computational fluid dynamics (CFD) simulation. The results from these methods were compared to measured results taken from the physical hardware test bed. A thermal resistance network model was used to describe the problem, and the comparison between the measured and calculated results for each element of the model was reported as a percent error. Note that in cases in which physical measurements could not be obtained, the comparison was made between the CFD results and the other calculations. All predictive methods using manufacturer-supplied thermal interface material (TIM) data indicated adequate thermal margin. When corrected TIM values were used, both the hand calculation method and the spreadsheet modeling method indicated adequate thermal margin; however, the CFD analysis predicted significant thermal issues, which were observed in the thermal test bed measured results.
AB - Discrepancies between estimated and actual thermal solution performance may negatively bias early system design decisions and ultimately result in costly redesigns. In particular, an overreliance on manufacturer-provided performance data and insufficiently accurate prediction methods are two potential causes of large discrepancies. To demonstrate deviations between predicted and end-thermal-performance, three methods of predicting thermal performance were investigated. The reference design was a forced convection air-cooled computer chassis, consisting of four printed circuit boards (PCBs), each populated with four processors that utilize heat sinks. The three predictive methods used were rudimentary analytical calculations, spreadsheet based modeling, and computational fluid dynamics (CFD) simulation. The results from these methods were compared to measured results taken from the physical hardware test bed. A thermal resistance network model was used to describe the problem, and the comparison between the measured and calculated results for each element of the model was reported as a percent error. Note that in cases in which physical measurements could not be obtained, the comparison was made between the CFD results and the other calculations. All predictive methods using manufacturer-supplied thermal interface material (TIM) data indicated adequate thermal margin. When corrected TIM values were used, both the hand calculation method and the spreadsheet modeling method indicated adequate thermal margin; however, the CFD analysis predicted significant thermal issues, which were observed in the thermal test bed measured results.
KW - error
KW - thermal interface material
KW - Thermal resistance
UR - http://www.scopus.com/inward/record.url?scp=85048520277&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85048520277&partnerID=8YFLogxK
U2 - 10.1109/SEMI-THERM.2018.8357362
DO - 10.1109/SEMI-THERM.2018.8357362
M3 - Conference contribution
AN - SCOPUS:85048520277
SP - 119
EP - 126
BT - 34th Annual Semiconductor Thermal Measurement and Management Symposium, SEMI-THERM 2018 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
ER -