Thermal studies at Valencia and Karlsruhe C. Lacasta, C. Mariñas, M. Vos IFIC-Valencia O. Brovchenko, Th. Muller, H.J. Simonis, T. Weiler IEKP-Karlsruhe Transparencies extracted partly from presentations at 2nd Int. Workshop on DEPFET detectors and Applications
Steady state simulation • DCDs always active: The hottest points • 2 Switchers active • 2 pixel stripes active • DCDs active: Solved with cooling blocks • 2 Switchers + 2 pixel stripes active: Very hot! Weneedtocooldownthe DCD Now, theproblemisthe SW • Thisis a personal estimationonpowerconsumption! Justqualitativelyresults! Weneedtoagreeonsomenumbersto do thisstudies! • Next more accuratesimulationis in progress, including real dimensions, materials.... • Preliminaryparameters are usedbetweengroupsinvolved in thermalissues (Karlsruhe and Valencia, regular meetings). Cross-checkbetweenmeasurements and simulationisontheway.
Set up • Measurementsmadeon a smallmicrostrip detector. • Theheateris placed in themiddle of the sensor. • Pt100 resistancefortemperaturemeasurement • Dimensions 34x14 mm2 • Thickness 300 mm • Coolantcomingfrom a chiller • Desired T over a widerange
Influence of conduction • Evolution of theswitcher, fordifferenttemperatures of thecooling blocks, as a function of powerdissipatedbythe chip. • Theslopeisalwaysthesame. Thedifferenceisthe offset. • Theinfluence of thecooling blocks in the center of the module isnot so big.
Influence of convection We really need to blow the air!! • Evolution of thechip’stemperature, as a function of speed of air for 0,4 Watt of power. • The air flowingisaneffectivemechanismforcoolingthe center of the module. Forhigherpower, theeffectisevenbigger • Once the air isblowing, the T variesslow, independently of thespeed (at thisrange).
The heat removal using the support bars • From C. Kiesling talk’s in Valencia, option number 1: • A bigthicknessisneededwiththis material (Al or Cu). • Gluingthesupportstructureunderneaththe module isnotpossibleduetothelack of space! But…
Thermal studies of new materials • Replace the Al for synthetic diamond (H.J. Simonis) or TPG: could Very small transverse area
First DEPFET thermal mock-up Switcher DCD 5 14 In mm 4 17 0.45 Module • Switchers • 6 switchers • Switched on/off sequentially • (0,1s ON/0,2s OFF) • 0.4 Watts each 110 • DCD’s • 1 heater on each end • Always powered on • 1.6 Watts each TPG 4 10
Estimation of power consumption • SWITCHERS • Active: • Idle: • DCD • Always active: • Pixels • Active: • Comparison between real values and what I have used: It is not easy to achieve such a high power with the heaters!
A more realistic approach… • Now the power disipated by the Switchers and DCD’s is bigger than before but half the expected value in the final module. • The contact with the cooling blocks is made by a couple of sheets of TPG. 500mm thick, 20 mm long and 17 mm wide. Overlap of 85mm2 underneath the balcony . Natural convection A bit more power Natural convection Lowpower • Switchers: 0,4 W • DCD: 1,6 W • The sensor is around 39ºC • Switchers: 0,8 W • DCD: 2 W • The sensor is around 50ºC Without air, thetemperature of the sensor increases a lot, evenwithpowershalfthe final ones.
Not everything is lost… Blowing the air: • Conduction through the TPG to the cooling blocks. 500mm thick, 20 mm long and 17 mm wide. • Forced convection. Natural convection Forcedconvection • Switchers: 0,8 W • DCD: 2 W • Tair=20ºC; vair=5m/s • Temperature sensor=41ºC • Switchers: 0,8 W • DCD: 2 W • No air • Temperature sensor=50ºC Nowthesituationis more favorable
Wehavetopumpout of the sensor a bigamount of power… butthere are severalthingsthatwe can do: • Decreasethetemperature of thecooling blocks tothelowervaluepossible. Is 7ºC or 8ºC achievable? Condensation? • Decreasethelengththattheheatmustcover. Isthere a possibilitytomovethesupportstructureclosertothe modules? Longer modules? • Decreasethetemperature of the air • Increasethe air flow • Introduce materialswithbetterthermalcoefficient First rough studies
A couple of options • TPG • In-planethermalconductivity: 1550 W/mK (at 20ºC) • Out of planethermalconductivity: 20 W/mK (no matter! smallthickness) • Density: 2.15 g/cm3 • Wellstudied material: Alreadytested in ATLAS SCT • Verysoft material • CVD (Chemical Vapor Deposition)-Diamond • In-plane and out of planethermalconductivity: 1800 W/mK (at 20ºC) • Density: 3.515 g/cm3 • Thethinnerthecheaper • Goodrigidity • “Cleaner” • Betterformechanicalstability?