![Page 1: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/1.jpg)
Cooling of GEM detector
CFD-2012-03_GEM2012/03/06
E. Da Riva CFD-2012-3_GEM 1
![Page 2: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/2.jpg)
1. Geometry and
material properties
E. Da Riva CFD-2012-3_GEM 2
![Page 3: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/3.jpg)
Geometry and material properties
Copper, k = 400 W m-1 K-1
PCB, FR-4, k = 0.25 W m-1 K-1
E. Da Riva 3CFD-2012-3_GEM
A single chip is considered in the CFD simulations
![Page 4: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/4.jpg)
Geometry and material properties
Soldered = perfect thermal contact Gap-pad k = 2 W m-1 K-1
Chip, silicon, q = 1 W
Water, T = 18°CAll surfaces are adiabatic but the inner surface of the pipe
E. Da Riva 4CFD-2012-3_GEM
![Page 5: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/5.jpg)
Geometry and material properties
Connector k = 0.27 W m-1 K-1
Volume including the GEM sensors is not taken into account.This surface is assumed as adiabatic.
E. Da Riva 5CFD-2012-3_GEM
![Page 6: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/6.jpg)
2. Schematic and estimations
E. Da Riva CFD-2012-3_GEM 6
![Page 7: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/7.jpg)
E. Da Riva CFD-2012-3_GEM 7
Schematic of the problem
18°CWater heattransfer coefficient
TCu strip
T T
Gap-pad PCB
Tchip
q = 1W
TGEM
PCB, connectors, etc.
q = 0 W
The temperature of the GEM sensors will be the same as the PCB. In order to reduce the sensor temperature, the thermal resistances must be reduced:-) higher water flow rate -> higher heat transfer coefficient-) shorter Cu strip or larger “cross section”-) Thinner gap-pad or higher thermal conductivity-) good thermal contact Cu strip/gap-pad and gap-pad/PCB
![Page 8: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/8.jpg)
Water in the pipe
E. Da Riva CFD-2012-3_GEM 8
Di [mm]
Vel.[ms-1]
Flow rate[kg s-1]
Temp. rise[K]
Re[-]
HTC [Wm-2K-1]
Δp (2 m pipe)[bar]
6 1 0.028 0.29 5700 4700 0.066 1.8* 0.051 0.16 10200 8200 0.174 1 0.013 0.65 3800 4500 0.114 1.8* 0.023 0.36 6800 8400 0.28
• Heat load per VFAT chip = 1 W• # VFAT chips = 30• Total heat load = 1 W * 30 + 4 W = 34W
* Erosion limit for copper pipe
Low water temperature rise is good to achieve a uniform sensor temperature. The diameter of the pipe may be reduced if needed because of geometrical constraints.
![Page 9: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/9.jpg)
3. Simulation results
E. Da Riva CFD-2012-3_GEM 9
![Page 10: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/10.jpg)
E. Da Riva CFD-2012-3_GEM 10
With water at 18°C, the temperature of the GEM sensor is expected to be around 60°C.
![Page 11: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/11.jpg)
E. Da Riva CFD-2012-3_GEM 11
The dominant thermal resistance is due to the low PCB thermal conductivity and the small interface area between chip and PCB
![Page 12: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/12.jpg)
E. Da Riva CFD-2012-3_GEM 12
The resistance due to water heat transfer coefficient is negligible (wall temperature very close to water temperature)
![Page 13: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/13.jpg)
E. Da Riva CFD-2012-3_GEM 13
Schematic of the problem 2
18°CWater heattransfer coefficient
TCu strip
T T
Gap-pad PCB
Tchip
TGEM
PCB, connectors, etc.
Negligible ~ 10 K Dominant thermal resistance
Increase heat contact area
-) Increase Cu thickness-) Increase Cu width
![Page 14: Cooling of GEM detector CFD-2012-03_GEM 2012/03/06 E. Da RivaCFD-2012-3_GEM1](https://reader036.vdocuments.net/reader036/viewer/2022062505/5a4d1bea7f8b9ab0599e39c3/html5/thumbnails/14.jpg)
E. Da Riva CFD-2012-3_GEM 14
Conclusions
• The dominant thermal resistance is between the chip and the PCB
• Direct thermal contact between the copper strip and the chip may be needed (putting the chip above the PCB instead of below?)
• A further step may be increasing the thickness and the width of the copper strip
• The diameter of the pipe may be slightly reduced if needed.