1. Welcome 28.09.2011Jan Sammet- 1 - A DC-DC converter based powering scheme for the upgrade of the CMS pixel detector TWEPP-11, Wien 28.09.2011 Arndt Schultz von Dratzig, Lutz Feld, Waclaw Karpinski, Katja Klein, Jennifer Merz, Jan Sammet, Oliver Scheibling, Michael Wlochal

2. Welcome Outline 28.09.2011Jan Sammet- 2 - Introduction Implementation into CMS Challenges… … and how we address them Conclusions & Outlook

3. Welcome Upgrade of the CMS Pixel Detector 28.09.2011Jan Sammet- 3 - CMS pixel detector is going to be replaced during shut down of LHC in ~2016  Additional layer of pixel modules  Number of readout chips (ROCs) increases by factor of 1.9 (16k  30k)  Present powering scheme cannot supply sufficient power through existing cables  Use DC-DC converters to reduce power losses  Aachen develops, tests and produces converters, to be used by CMS 93 cm 53 cm The CMS pixel detector

4. Welcome Strategy 28.09.2011Jan Sammet- 4 - Increase efficiency of power transmission P loss =R cable ·I 2  supply power at higher voltage r·U i.e. lower current I/r (Conversion ratio r >1)  power losses P loss =R·(I/r) 2 are reduced by factor r 2 !

5. How it works Frequently connect and disconnect source and load Duty cycle D = t on /T Conversion ratio r = V in /V out = 1/D (ideal converter) Welcome DC-DC Buck Converters 28.09.2011Jan Sammet- 5 - ASIC includes transistors and voltage regulation circuit ASIC is being developed within CERN electronics group (F. Faccio et al.) [see talk by S. Michelis http://indico.cern.ch/contributionDisplay.py?contribId=21&confId=120853] Radiation tolerance of many semi-conductor technologies evaluated  AMIS I3T80 0.35µm (ON Semiconductor, US) - functional up to dose of 300Mrad & fluence of 5  10 15 p/cm 2 [F. Faccio, TWEPP-10, Development of custom radiation-tolerant DCDC converter ASICs] Advantages Output regulation by pulse width modulation  V out is regulated Provide high currents with high efficiency

6. Welcome Aachen DC-DC Converter 28.09.2011Jan Sammet- 6 - PCB: 2 copper layers a 35µm 0.3mm thick Large ground area on backside for cooling Toroidal inductor: L = 450nH R DC = 40m  Shield/heat sink A = 28 x 16 mm 2 M  2.5g 3.8% of a radiation length “PIX_V7“: Design guidelines from CERN group have been implemented. Pi-filters at in- and output ASIC prototype: AMIS2 by CERN I out < 3A V in < 12V V out configurable; (here: 2.5V & 3.3V) f s configurable, e.g. 1.3MHz

7. Welcome Challenges 28.09.2011Jan Sammet- 7 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions  (covered by CERN group)

8. Welcome Challenges 28.09.2011Jan Sammet- 8 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions 

9. Welcome Implementation into CMS 28.09.2011Jan Sammet- 9 - 2.2m Mockup Integration for pixel barrel onto supply tube Large distance of converters to pixel modules (  ~ 4) (goal is to be able to power detector, NOT to reduce material) CO 2 cooling available 26 DC-DC converters per channel Power dissipation ~ 40W per channel Full channel (PCB, cooling pipes & bridges, 26 converters) weighs 200g and corresponds to 7% of a radiation length 2 000 DC-DC converters required in total Pixel detector

10. Welcome Implementation into CMS 28.09.2011Jan Sammet- 10 - CAEN A4603 PSU Vana Vdig 6 - 7 converters 1 - 4 pixel modules per converter DC-DC dig DC-DC ana 6 - 7 converters Two CAEN power supply units (PSU) per supply tube channel Power supplies need modification I < 2.8A per converter (for L = 2 x 10 34 cm -2 s -1 ) V out = 2.5V V out = 3.3V V in  12V 1 - 4 pixel modules per converter  50m

11. Welcome Challenges 28.09.2011Jan Sammet- 11 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions  

12. Welcome Challenges 28.09.2011Jan Sammet- 12 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions  

13. Electrical shielding: Acts also as cooling contact for coil Shape optimized to fit into edge channels of supply tube Several technologies are under investigation: Hydroformed/deep drawn aluminium Ruled out; required shape not feasible in mass production Plastic shields coated with a metal layer Different materials, thicknesses and depositing techniques have been tested Milled Aluminium shields 120µm of Aluminium etched down to 90µm+ 15 µm of Nickel + 15 µm of Copper Welcome 28.09.2011Jan Sammet- 13 - R&D on Shielding

14. Welcome 28.09.2011Jan Sammet- 14 - Effectiveness of Shielding Measure z-component of magnetic field of DC-DC converter Automated x-y table allows high spatial resolution Many different materials and thickness have been tested  focus on thinnest and lightest functional shieldings here Scanning table BZBZ

15. Welcome 28.09.2011Jan Sammet- 15 - Effectiveness of Shielding 90µm milled Aluminium Magnetic field reduced to 15% Price per piece ~ 25 € Plastic shield coated with 30 µm Cu Magnetic field reduced to 30% Price per piece: ~ 13 € Plastic shield coated with 60 µm Cu Magnetic field reduced to 5% Price per piece: ~ 13 € Roughly 56% higher contribution to MB No shield 90µm AL 60µm Cu 30µm Cu

16. Welcome 28.09.2011Jan Sammet- 16 - Thermal FE-Simulation Cooling bridges clamp around CO 2 pipes Chip cooled through PCB backside Shield (soldered to PCB) acts as cooling contact for inductor Temperature of cooling bridge T = -20°C  Chips are at -7°C (  T = 13K)  Coils are at -6°C (  T = 14K)  For room temperature (20°C) operation, both stay below 40°C -20 -10 -15 -6 °C Shielding simulated, but not displayed

17. Welcome 28.09.2011Jan Sammet- 17 - Thermal Measurements PIX_V7, 450nH, 1.3MHz V in = 10V, V out = 3.3V Output current [A] Coil temperature [°C] Converter on cooling bridge at 20°C  Good agreement with Finite Element simulations  Shielding fulfils task of cooling contact for coil  Cooling of chips via backside of PCB is very effective

18. Welcome Challenges 28.09.2011Jan Sammet- 18 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions    

19. Welcome Challenges 28.09.2011Jan Sammet- 19 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions    

20. Welcome Jan Sammet- 20 - Efficiency 28.09.2011 Phase 1 conditions: V out = 3.3V or 2.5V, I out < 2.8A, conversion ratio of 3-4  75% - 80% efficiency: ok – still expected to increase further with new version of AMIS chip [White regions: regulation not working properly, V out too low] PIX_V7, V out = 2.5V PIX_V7, V out = 3.3V Efficiency [%]

21. CopperAluminium Photo Inductance L [nH]~450 Resistance R DC [m  ] 4055 Height h [mm]66 Diameter d [mm]79 Mass m [mg]660370 Welcome Jan Sammet- 21 - Choice of Inductor 28.09.2011 Optimal coil shape has been calculated for different wire types and thicknesses In the available space, copper offers higher efficiency For pixel upgrade, favour efficiency over material budget (due to high  )

22. Welcome Challenges 28.09.2011Jan Sammet- 22 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions      

23. Welcome Challenges 28.09.2011Jan Sammet- 23 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions      

24. Welcome Conductive Noise 28.09.2011Jan Sammet- 24 - Spectrum Analyzer Load LISN = Line Impedance Stabilization Network GND Differential Mode (DM), “ripple“Common Mode (CM) Noise through cables (conductive noise) was studied with EMC set-up EMC = electromagnetic compatibility

25. Welcome Conductive Noise 28.09.2011Jan Sammet- 25 - Differential Mode, no shieldCommon Mode, no shield Differential Mode, with shieldCommon Mode, with shield  Large reduction of CM above 2 MHz due to shield PIX_V7 output noise V out = 3.3V V in = 10V f s = 1.3MHz L = 450nH

26. Welcome System Tests with CMS Pixel Modules 28.09.2011Jan Sammet- 26 - DC-DC converter on bus board Pixel module System tests with original pixel PSU from CAEN and module qualification set-up One full module (16 ROCs, each containing 4160 pixels) is read out S-curve noise of each pixel is measured Many thanks to PSI for hardware and advice VAVA Connector board Module adapter PC interface „Advanced Test Board“ DAQ PC Module DC-DC converters CAEN Pixel PSU multi-service cables (40m) VDVD EASY 4000 SY1527 + Branch Controller PSM A4603 Original pixel power supply unit HV (150V)

27. Welcome Powering with DC-DC Converters 28.09.2011Jan Sammet- 27 - DC-DC converters cause no significant increase of noise powering with DCDC conventional powering

28. Welcome 28.09.2011Jan Sammet- 28 - Impact of Orbit Gaps Orbit gaps of LHC beam  no collisions for 3µs every 89µs  Pixel digital current drops within 25..100ns (1..4 bunches, depending on the occupancy)  Worst case for 2×10 34 cm -2 s -1 : I D_high = ~ 2.7A (86µs)  I D_low = ~1A (3µs)  Stability Test of the whole power supply chain:  Also check impact of “inverted orbit gaps” I D_high = ~ 2.7A (3µs)  I D_low = ~1A (86µs) (LHC fill with only a few intense bunches) VAVA Connector board Module adapter PC interface „Advanced Test Board“ DAQ PC Module HV (150V) DC-DC converters CAEN Pixel PS multi-service cables (40m) VDVD Dynamic Load Box EASY 4000 EASY 4000 SY1527 + Branch Controller SY1527 + Branch Controller PSM A4603 PSM A4603 IDID t IDID t

29. Welcome Impact of Orbit Gaps 28.09.2011Jan Sammet- 29 - Without DC-DC converters  Load variations cause moderate increase of noise With DC-DC converters  Sensitivity is reduced due to filtering components and regulation of DC-DC converter Powering with DC-DC convertersConventional powering

30. Welcome Influence of the Switching Frequency 28.09.2011Jan Sammet- 30 -  Slight reduction ( < 1e ) at 2 MHz  Not significant yet  Might become relevant in larger system  Not directly comparable to previous measurements   (slightly different cabling, no copper plate here) 

31. Welcome Different Sensing Points 28.09.2011Jan Sammet- 31 - Dynamic Load DC-DC converter CAEN PS multi-service cables (40m) Module Powering with DC-DC converters Sensing at PS Sensing at DC-DC converter  No increase of noise due to sensing at PS  Sensing at DC-DC converters not necessary DC-DC converter is seen by PS as negative impedance  Sensing at converter could cause instabilities  Modifications of PS would be more complex and costly  Measure influence of sensing point in system tests

32. Welcome Challenges 28.09.2011Jan Sammet- 32 - Ensure radiation tolerance of high voltage (~15V) power transistors Implement DC-DC converters into CMS Provide magnetic field tolerance  air-core inductor  risk of radiated noise Provide good shielding and cooling Provide high conversion efficiency Provide compact device with small impact on material budget Reduce and control noise emissions       

33. Welcome Conclusions & Outlook 28.09.2011Jan Sammet- 33 - Conclusions Good progress has been made towards a DC-DC converter based powering scheme Plans for implementation are far advanced Efficiency and noise performance of our DC-DC converters are satisfactory Outlook Improve system tests with more modules, cold temperatures… Perform thermal cycling to simulate accelerated aging and to ensure reliable operation in the long term Implement new versions of the AMIS chip (AMIS4 expected soon) Implement control scheme for DC-DC converters

34. Welcome Back-Up Slides 28.09.2011Jan Sammet- 34 -

35. Welcome The CMS Pixel Detector 28.09.2011Jan Sammet- 35 - Pixel size: (100 x 150) µm² Present barrel New barrel

36. Welcome Aachen DC-DC Converter 28.09.2011Jan Sammet- 36 - Design guidelines from CERN group have been followed.

37. Welcome 28.09.2011Jan Sammet- 37 - Mechanical & Thermal Integration cooling pipes lower part of cooling bridge upper part of cooling bridge chip area Cooling bridge clamps around pipe Area reduced to reduce material Aluminium (could be Graphite, but gain for material budget is low)

38. Welcome Effectiveness of Shielding 09.09.2011Jan Sammet- 38 -

39. Welcome 28.09.2011Jan Sammet- 39 - Conductive Noise with Shielding

40. Welcome Set-Up for System Tests 28.09.2011Jan Sammet- 40 - Original pixel power supply system Pixel module DC-DC converters

41. Welcome Powering with DC-DC Converters 2 28.09.2011Jan Sammet- 41 - Variations with different but identical converters ~ 2e Reproducibility of measurements ~ 1e VD = 3.2V, VA= 2.5V

42. Welcome Influence of the Switching Frequency 28.09.2011Jan Sammet- 42 - With inverted orbit gapsWith simulated orbit gaps Additional 2A const. LoadNo additional load (VD = 3.2V, VA= 2.5V)

43. Welcome Different sensing points, w/o Add. Load 28.09.2011Jan Sammet- 43 - Powering with DC-DC converters DC-DC converters CAEN PS multi-service cables (40m) Module Sensing at PS Sensing at DC-DC (VD = 3.2V, VA= 2.5V) µ-twisted pair cables DC-DC converter is seen by PS as negative impedance  Sensing at converter could cause instabilities  Modifications of PS would be complex and costly Test influence of sensing point in system tests  No increase of noise due to sensing at PS

44. Welcome Different sensing points, w/o Add. Load 28.09.2011Jan Sammet- 44 - Powering with DC-DC convertersConventional powering  No increase of noise due to sensing at PS DC-DC converters CAEN PS multi-service cables (40m) Module Sensing at PS (with voltage divider) Sensing at DC-DC/load (with and w/o voltage divider) (VD = 3.5V, VA= 2.7V) (VD = 3.2V, VA= 2.5V) µ-twisted pair cables

45. Welcome Different sensing points with Orbit Gaps 28.09.2011Jan Sammet- 45 - Dynamic Load Box DC-DC converters CAEN PS multi-service cables (40m) Module Powering with DC-DC converters (VD = 3.2V, VA= 2.5V) Sensing at PS Sensing at DC-DC/load µ-twisted pair cables Test influence of sensing point in system tests  No increase of noise due to sensing at PS  Simplifies modifications of PS by CAEN

46. Welcome Different sensing points with Orbit Gaps 28.09.2011Jan Sammet- 46 -  Noise increases if voltages are senses at PS  No increase of noise due to sensing at PS Dynamic Load Box DC-DC converters CAEN PS multi-service cables (40m) Module Sensing at PS (with voltage divider) Sensing at DC-DC/load (with and w/o voltage divider) Powering with DC-DC convertersConventional powering (VD = 3.5V, VA= 2.7V) (VD = 3.2V, VA= 2.5V)