1. The HiLumi LHC Design Study is included in the High Luminosity LHC project and is partly funded by the European Commission within the Framework Programme 7 Capacities Specific Programme, Grant Agreement 284404. Ramp down of the inner triplet magnets Power converters for HL-LHC WP 6b : Warm powering Jean-Paul Burnet, CERN TE-EPC

2. 2 Content Introduction of 1-quadrant converter Principle of the ramp down Present limitations Cases of new Inner Triplet magnets Perspectives HL-LHC TC – 30 April 2015

3. 3 Circuit layout with 1-quadrant converter HL-LHC TC – 30 April 2015 1-quadrant converter has a free-wheeling diode at the output. Current and voltage applied to the magnets is always positive.

4. 4 Principle of the ramp down The power converter needs to stay under control to operate it correctly (no voltage transient). Two regulation loops are operating: -High precision current loop -Voltage loop During run1, the ramp down was always done with current control. To improve the ramp down, a new principle was introduced: The new ramp down is executed in two phases: -First phase, current control with constant di/dt at -10A/s -Second phase, voltage control regulated at 0V This is needed to keep the power converter under control. HL-LHC TC – 30 April 2015

5. 5 RQD/F as reference RQD circuit has these characteristics: -I nom = 11870A -L magnet = 263mH -R circuit = 1.1mΩ -Τ circuit = 240s -Di/dt max = 10A/s Phase1 = 950s Phase2 = 550s Ramp-down = 1500s HL-LHC TC – 30 April 2015 Current control only Phase1, current control Phase2, voltage control New principle Phase 1 Phase 2

6. 6 Ramp down present limitation HL-LHC TC – 30 April 2015 Ramp-downPrecycleSolution RB1257 sec (21 min)3110 (52 min) RQXs3323 sec (56 min)4368 sec (1h 13 min) RQSs2195 sec (37 min)3294 sec (55 min) Current reduced (250A) + settings optimization MQYs2051 sec (34 min) RQD/F~2000 sec (33 min)3500 sec (59 min) RQ4/5.LR3/7 (MQWBs) 1935 sec (32 min) To be defined: 1 less cycles? Reduced current? ROs1920 sec (32 min)1 cycle less RQD can be taken as a reference From Matteo Solfaroli, Hugues Thiesen

7. 7 Present baseline for the Inner triplet The present baseline includes two 16.5kA power converters. One power converter for Q1-Q3 and one for Q2a-Q2b. HL-LHC TC – 30 April 2015

8. 8 Proposed layout of the UR gallery HL-LHC TC – 30 April 2015 Distance between DFHX and converter type1 (16.5kA) :15m and29m

9. 9 Resistance of the cables The water cooled cables will have a cross section of 2600mm 2 of cooper. HL-LHC TC – 30 April 2015

10. 10 Ramp down for Q1-Q3 short cables Calculation of phase 1: Limitation at -14A/s I end of phase 1 = 8050A Phase 1 duration = 604s Calculation of phase 2: Free-wheeling process until I standby (490A): Phase 2 duration = 1609s Ramp-down = 2213s HL-LHC TC – 30 April 2015 Rcables is only 0.24mΩ for 34m τ circuit = 497s

11. 11 Ramp down for Q1-Q3 long cables Calculation of phase 1: Limitation at -14A/s I end of phase 1 = 4493A Phase 1 duration = 857s Calculation of phase 2: Free-wheeling process until I standby (490A): Phase 2 duration = 711s Ramp-down = 1569s HL-LHC TC – 30 April 2015 Rcables is only 0.43mΩ for 62m τ circuit = 320s

12. 12 Alternative layout for the Inner triplet HL-LHC C&S Review #1 – March 2015 Requested for beam optics to reduce the tune shift. More complex due to nested circuits, however still feasible. Cheaper due to less SClink, and power converters. Higher inductance (255mH) to reduce current ripple (tune shift). Q1 Q2a Q2b Q3 PC1 17.4 kA PC2 ±2.0 kAPC3 ±0.2 kAPC4 ±2.0 kA Q1-Q2-Q3: -slightly smaller tune shift than Q1-Q2a Q2b-Q3 for current control regime -best compensation voltage control regime => best scheme (for beam dynamics) HL-LHC TC – 30 April 2015

13. 13 Q1-Q2-Q3 circuit The water cooled cables will have a cross section of 2600mm 2 of cooper. HL-LHC TC – 30 April 2015

14. 14 Ramp down for Q1-Q2-Q3 with long cables Calculation of phase 1: Limitation at -14A/s I end of phase 1 = 8302A Phase 1 duration = 585s Calculation of phase 2: Free-wheeling process: Phase 2 duration = 1678 s Ramp-down = 2264s HL-LHC TC – 30 April 2015 Rcables is 0.43mΩ for 62m τ circuit = 594s

15. 15 Can we go faster ? HL-LHC TC – 30 April 2015 Yes but not in controlled way ! When the converter is regulated, the current is going through the output rectifier diode, Dr1, Dr2. In case of fault, the current goes in the wheeling diodes first path. In case of diode breakdown, the current goes in the second path (rectifier diodes). In case of full schottky diodes, the current goes in the third path (press pack diode).

16. 16 Can we go faster ? HL-LHC TC – 30 April 2015 If we stop the control of the converter, the current will go through the first secure path The current will decay in exponential decay.

17. 17 Ramp down without converter control HL-LHC TC – 30 April 2015 The electrical circuit is described by the equation: Vout = R cables. I magnet + L magnet. dI magnet /dt During the free-wheeling process, the current is flowing only through the output diode. Vout = -Vdiode= -0.3V which is the diode voltage when diode is conducting. I magnet (t) = (I 7TeV + V diode / R cables ) * e -t/ τ circuit - V diode / R cables Exponential decay Where τ circuit = L magnet / R cables

18. 18 Case of Q1-Q3 with short cables Time to reach 0A with only the free-wheeling process. T 0A = - L magnet / R cables * ln (V diode / (V diode + R cables * I 7TeV )) HL-LHC TC – 30 April 2015 Need 3.5 τ circuit to reach 0A. Case of Q1-Q3 with short cables: Tcircuit = 497s Time to reach Istandby = 1335s Max didt = -30A/s Is this High didt a problem for the protection? How to take back the current control at the end of the decay?

19. 19 Summary Current ramp down is an issue with the new inner triplet Studies will be done by WP6B to find solutions 2-quadrant converter solve the issue HL-LHC TC – 30 April 2015