1. 1 S. Michelis Inductor-based switching converter for low voltage power distribution in LHC upgrades S. Michelis, F. Faccio, A. Marchioro – PH/ESE/ME Twepp07 – 6 th September 2007

2. 2 S. Michelis Working environment ► High magnetic field (up to 4T in CMS, 2T in ATLAS) ► High level of radiation will be reached inside the detectors: LHC doses probably increased x 5 – 10 so we can extrapolate to hundreds of Mrad in several Tracker location, decreasing to ten(s) in the outer Trackers Twepp07 – 6 th September 2007

3. 3 S. Michelis Present power distribution schemes in trackers AC/DCDC/DC PP PP (with “LDO” in some cases) PP Module 0.5-2m 13-15m 100m 1A/channel (analog or digital), round-trip cables, and sense to nearest regulation AC/DC PP Module 0.5-2m100m PP Module 0.5-2m PP Module 0.5-2m DC/DC Twepp07 – 6 th September 2007

4. 4 S. Michelis SuperLHC requirements AC/DCDC/DC PP PP (with “LDO” in some cases) PP Module 0.5-2m 13-15m 100m Either >>1A/channel, or 1A/channel and more modules More channels required, at lower Vdd, hence overall more current! More or thicker cables: Material budget ! Cooling ! Cable crowding! Twepp07 – 6 th September 2007

5. 5 S. Michelis 4 - 8 V Possible solutions AC/DCDC/DC PP (with DC/DC) PP Module (with DC/DC) 0.5-2m100m Same cables as today can bring more power, It requires efficient DC/DC on module for cooling 12 - 24 V AC/DCDC/DC PP Module (with DC/DC) 0.5-2m100m Small cables can bring all the power It requires efficient DC/DC 12 - 24 V PP Important considerations: Magnetic field, Radiation and Material Budget, plus EMI if inductor-based DC/DC Twepp07 – 6 th September 2007

6. 6 S. MichelisTwepp07 – 6 th September 2007 Aiming at demonstrating the feasibility of a fully integrated (except L and passive components) DC-DC buck converter DC-DC: Work in progress Vin=12-24 V Vout=1.5-3V Iout=1-2A Rad-hard technology Inductor Power dissipation

7. 7 S. Michelis Selection of a technology Design and test of transistors of a 0.35 μm technology usually employed in automotive application Several different transistor topologies available for high-V applications (lateral, vertical) It needs to be radiation tolerant Twepp07 – 6 th September 2007

8. 8 S. MichelisTwepp07 – 6 th September 2007 Radiation Tolerant NMOS The main radiation-induced problem for the NMOS is the source-drain leakage current due to the radiation-induced trapping of positive charges in the thick lateral oxide Normal LayoutModified Layout It is necessary to modify the layout of the NMOS in order to make them radiation tolerant.

9. 9 S. MichelisTwepp07 – 6 th September 2007 Aiming at demonstrating the feasibility of a fully integrated (except L and passive components) DC-DC buck converter DC-DC: Work in progress Vin=12-24 V Vout=1.5-3V Iout=1-2A Rad-hard technology Inductor Power dissipation

10. 10 S. Michelis Inductors Inductor: need of air core inductor because of high magnetic field for CMS (up to 4 T) and ATLAS (up to 2 T) Stefano Michelis MaterialInitial Perm. µoB max (T)Operating Frequencies Fe2502,260-1000 Hz Si-Fe (unoriented)4002,060-1000 Hz Si-Fe (oriented)15002,060-1000Hz 50-50 Ni Fe (grain-oriented)20001,660-1000Hz AMORPHOUS Alloy B30001,5-1,6to 250 kHz High Flux powder14 to 1601,510 kHz to 1 MHz Kool Mu ® powder26 to 1251,0to 10 MHz Iron powder5 to 801,0100 kHz-100 MHz 79 Permalloy12,000 to 100,0008 to 1,11 kHz-75kHz AMORPHOUS Alloy E20,0000.5-0.65to 250 kHz Ferrite-MnZn750 To 15,0000.3 to 0.510 kHz-2 MHz Ferrite-NiZn10 to 15000.3 to 0.5200 kHz-100MHz Permalloypowder14 to 5500.310 kHz-1 MHz A Critical Comparison of Ferrites with other Magnetic Materials, 2000 Twepp07 – 6 th September 2007

11. 11 S. Michelis Inductors Inductance (nH) DCR max (mOhm) Irms (A) 380502,5 538902 Inductance (nH) DCR max (mOhm) Irms (A) 56068 1200108 Different commercial choices Coilcraft Stefano Michelis ABCD 16mm30mm14mm21mm ABCDE 10mm6mm 5mm8mm Renco electronics Twepp07 – 6 th September 2007

12. 12 Twepp07 – 6 th September 2007 11 Aiming at demonstrating the feasibility of a fully integrated (except L and passive components) DC-DC buck converter DC-DC: Work in progress Vin=12-24 V Vout=1.5-3V Iout=1-2A Rad-hard technology Inductor Power dissipation

13. 13 S. Michelis Power losses There are power losses in the circuit due to the non-ideal component. Mainly there are three losses that it is necessary to consider: 1. on R on and R ind, parasitic resistances of the switch and inductor, respectively; 2. on charging the gate capacitance; 3. on the control circuit. Twepp07 – 6 th September 2007

14. 14 S. Michelis Efficiency vs I out and W with V in =24V, V out =2.5V, R ind =6mΩ and L=700nH Twepp07 – 6 th September 2007

15. 15 S. Michelis NMOS design W=0.1m S1D1 G1 G2 SW 2 SW 1 SW2 S2 D2 4 mm 2 Twepp07 – 6 th September 2007 250 cells Each cell has W=400um divided in 8 fingers

16. 16 S. Michelis NMOS irradiation tests – Threshold Votage SD G 3.3V MOS1 0.5A MOS2 SD G 3.3V 20V MOS3 SD G 3.3V Pre-radPost-rad Twepp07 – 6 th September 2007

17. 17 S. Michelis NMOS irradiation tests – Leakage Current SD G 3.3V MOS1 0.5A MOS2 SD G 3.3V 20V MOS3 SD G 3.3V Pre-rad Post-rad Twepp07 – 6 th September 2007

18. 18 S. Michelis NMOS irradiation tests - R on SD G 3.3V MOS1 0.5A MOS2 SD G 3.3V 20V MOS3 SD G 3.3V Pre-radPost-rad Twepp07 – 6 th September 2007

19. 19 S. Michelis Conclusions We have demonstrated that: 1.the technology can be tolerant to TID up to at least 180Mrd and switch transistors of the required size can be integrated 2.Air core inductor can be used in such high DC magnetic field 3.efficiency above 80% is achievable Evaluation of EMI is in progress choosing dedicate architecture and inductor shape making measures and trying to standardize the measurement process. Twepp07 – 6 th September 2007

20. 20 S. MichelisTwepp07 – 6 th September 2007 THANK YOU