Commissioning of the LHC Superconducting magnet systems The

Commissioning of the LHC Superconducting magnet systems The challenges of safely powering the super-conducting magnets Quentin King on behalf of Freddy Bordry With thanks to Valerie Montabonnet, Dave Nisbet, Hugues Thiesen, Yves Thurel, Greg Hudson, Andrea Cantone and Jeff Thomsen

• • Definition of LHC Powering Individual Tests Short Circuit Tests Hardware Commissioning Main Dipole Current Regulation Idle Tones Inner Triplet Powering Transfer Function Analysis (demo)

• Definition of LHC Powering • • Individual Tests Short Circuit Tests Hardware Commissioning Main Dipole Current Regulation Idle Tones Inner Triplet Powering Transfer Function Analysis (demo)

Definition of the Powering • For every circuit the accelerator physicists define: – The range of field strengths – Rates of change, accuracy, ripple, etc… • These are translated by the choice of magnets in powering requirements: – – – Current range Ramp rate Ripple Accuracy Resolution Reproducibility = MONEY $$$

Point 4 Point 5 Radio frequency acceleration system SR 4 UA 47 UA 43 RR 53 CMS SR 5 RR 57 SR 6 UA 63 UJ 33 Point 6 Beam dump USC 55 Point 3 UJ 56 UA 67 SR 3 Reference magnets in SM 18. Orbit Corrector Power Converters distributed around arc tunnel. UA 27 SR 7 RR 73 UA 23 SR 1 SPS UA 83 Point 2 SM 18 ALICE RR 17 UA 87 TI 2 UJ 16 Point 1 ATLAS TI 8 UJ 14 Injection Betatron Cleaning RR 77 RR 13 Injection Point 7 Point 8 LHC-B

LHC Power Converters • Underground installation – Low volume, low weight – Only front access possible – no access during operation • • • High efficiency (>80%) EMC Number of Converters: 1720 Water cooling (90% of the losses in water) Total Current : 1860 k. A High reliability (MTBF 100’ 000 h) Steady State Input : 63 MW Repairability and rapid exchange of different parts Peak Input : 86 MW High precision – DC current – low voltage ripple, – perturbation rejection, …

LHC : 1232 SC Main Dipole magnets Magnet inductance : L = 108 m. H Ltotal=1232 * 0. 108 = 133 H Ramp: Ld. I/dt = 1330 V Discharge (quench; 120 A/s): 16 k. V Nominal current 11. 8 k. A Stored Energy = 9. 3 GJ Ultimate current = 13 k. A Stored Energy = 11. 3 GJ L/R 50 hours !!!! One circuit or several circuits ?

Why an Electrical Segmentation of the machine? ü Natural segmentation into 8 units as no cryostat in straight sections. ü Warm cable connections costly in copper, power losses (~30 MW instead of 10 MW) and power converters ü Only 1/8 of the machine needs to be discharged if one magnet quenches ü No risk of total machine avalanche quench, (false quench detection and provocation ü Eight sub-units give easier installation, testing, commissioning and fault finding for many systems ü Allows sector-to-sector correction of magnet errors due to different cable, magnet manufacturers, etc. . Need to track from sector to sector

Tracking between the 8 main dipole converters ppm Accuracy DB/Bnom = DI/Inom = ppm DB = 9 10 -6 = 1. 8 10 -4 T DB/Bo = 15 10 -6 Orbit excursion : d. X = Dx. DB/Bo = ~. 035 mm d. X =. 7 mm => = 20 ppm Could be corrected with a pilot run and new cycle => reproducibility 10 ppm reproducibility Orbit excursion : d. X = Dx. DB/Bo = ~ 0. 35 mm !!! “It would be better with 5 ppm” Oliver Brüning

Power Converter Tolerances for LHC Precision Control

UA 23 (Ex-LEP Klystron gallery) Now used to house the majority of machine equipment including power converters Very Low Radiation Dose

New Enlargement (RR) for Machine Power Converters around ATLAS and CMS Constraints : Volume, back-to-back, losses, weight and radiation!

Main Arc Tunnel Orbit Corrector PCs 4*[60 A, 8 V] 752 converters High reliability : MTBF : 80 ’ 000 h 1 converter breakdown every 4 days One campaign every 2 or 3 months Radiation Dose 1 -2 Gy/year under dipoles

The Challenges : Performance : -High current with high precision (accuracy, reproducibility, stability, resolution) and large dynamics -current range (for 1 -quadrant converter: from 1% to 100%) - a lot of 4 -quadrant converters (energy from magnets) - tracking : Need to track from sector to sector - voltage ripple and perturbation rejection Installation and Operation: - volume (a lot of converter are back-to-back) - weight (difficult access) => modular approach - radiation for [± 60 A, ± 8 V] converters (and others too!) - extraction of losses : high efficiency, water cooling - EMC : very close to the others equipment ; system approach

• Definition of LHC Powering • Individual Tests • • • Short Circuit Tests Hardware Commissioning Main Dipole Current Regulation Idle Tones Inner Triplet Powering Transfer Function Analysis (demo)

A Selection of LHC Converters Point 8: Main Dipole (RB) Atlas Torroid Point 4: 120 A Point 4: 600 A

What is a power converter? Cooling System Power Interlock Controller World. FIP fieldbus Power Part (Voltage Source) Vref Function Generator Current loop Current Transducer DCCT Magnets AC Mains Supply

PC Tests Before Installation Individual Tests at the suppliers Individual Tests at CERN Power Part FGC PC Integration and test at CERN Power Part FGC Dcct LHC PC Dcct

FGC Individual Tests Individual cards were tested after production by the manufacturer using automatic testers provided by CERN Completed FGCs run for at least a month in special “Reception” crates in Bat. 866

Class 1 DCCTs (13 k. A) - Highest performance - state of the art - Separate Head and electronics chassis 19” rack mounting - Fitted with Calibration Windings - Temperature-controlled environment in the Accelerator. - Full testing and calibration at CERN on the 20 k. A Test Bed.

Other DCCTs 4 k. A to 8 k. A DCCTs 600 A DCCTs 120 A DCCTs

Reception and Integration

• Definition of LHC Powering • Individual Tests • Short Circuit Tests • • • Hardware Commissioning Main Dipole Current Regulation Idle Tones Inner Triplet Powering Transfer Function Analysis (demo)

Short-circuit tests are not only power converter tests: energy extraction tests, DC cables tests, AC network conditions, cooling and ventilation, interlocks, control, … Short-circuit tests (SCT)

Short-circuit tests

Short-circuit tests From October 2005 to September 2007 All tests were successfully concluded by a 24 h endurance test (16 h at ultimate and 8 h nominal)

Power consumption during 24 Hour Short-circuit tests k. W 1400. 0 Total 1200. 0 1000. 0 Other PC 800. 0 600. 0 400. 0 Dipole 200. 0 Power measured by TS-EL in SE 8

Air Temperature during 24 Hour Short-circuit tests 32 27. 5 ºC 20 ºC 3500 31 m 3/h ? ºC 30 29 33 ºC 30? ºC 30 ºC 20? ºC 20 ºC 5000 m 3/h 28 sat 32 sat 27, 28 & 32 27 27 ºC 18. 5 ºC 3500 m 3/h RB Temp (18 H 00) (19 H 00) 28 27 sat (13 H 00) 29 J. Thomsen

UJ 33 24 Hour Run – Systematic Infra Red (IR) survey Y. Thurel 80. 3°C 90°C

Systematic Infra Red (IR) survey An Infra Red analysis to see what a hand cannot feel at less than 20 cm !!! 275°C due to a loose connection Y. Thurel

Systematic Infra Red (IR) survey 4 racks with 8 * [± 600 A; ± 10 V] in UA 67 An Infra Red analysis shows that everything is OK Y. Thurel

• Definition of LHC Powering • Individual Tests • Short Circuit Tests • Hardware Commissioning • • Main Dipole Current Regulation Idle Tones Inner Triplet Powering Transfer Function Analysis (demo)

Tracking between the main circuits of sector 78 Test Method: Current Channel A swapped Regulation with Channel B RB, RQD, RQF synchronized ramp Dave Nisbet

Tracking between the main circuits of sector 78 RB/QF/QD Tracking – 350 A to 2 k. A Dave Nisbet

Tracking between the main circuits of sector 78 2 ppm Free-wheeling : L/r 23’ 000 s Dave Nisbet

Powering of Q 4 MQM Matching Sections - Current loop robustness: L/2 r to L/r - Always: static and dynamic I 1/2 < I 2 < 2 x. I 1 and I 2/2 < I 1 < 2 x. I 2 Dave Nisbet

Squeeze tests (PSQ) : Q 4 and Q 5 RQ 4. L 8 B 2 is close to limit New optic function much improved (15 min squeeze) All systems performed as calculated With LHC Software Application LSA: generation of table (I, t) MQM control touchy during ramp down with 1 -Quadrant converter Good Performance even if the limits are close RQ 5. L 8 B 2 I_MEAS RQ 5. L 8 B 1 I_MEAS RQ 5. L 8 B 2 V_MEAS RQ 5. L 8 B 1 V_MEAS RQ 4. L 8 B 2 I_MEAS RQ 4. L 8 B 1 I_MEAS RQ 4. L 8 B 2 V_MEAS RQ 4. L 8 B 1 V_MEAS Close to Limits 0 V Dave Nisbet

Converter Operation during a sub-converter failure V. Montabonnet

Converter Operation during a sub-converter failure [13 k. A, 18 V] converter : (4+1) x [3. 25 k. A, 18 V] subconverters Tests during 7 -8 hardware commissioning 3. 25 k. A , 18 V 13 k. A, 18 V 3. 25 k. A , 18 V 1 ms 1625 A 3. 25 k. A , 18 V 1300 A Restart of sub-converter 2 Vout 6500 A At injection current : 860 A I_MEAS = About 1 -2 ppm pk-pk V. Montabonnet

• • Definition of LHC Powering Individual Tests Short Circuit Tests Hardware Commissioning • Main Dipole Current Regulation • Idle Tones • Inner Triplet Powering • Transfer Function Analysis (demo)

Main dipole circuit powering by RB converter 1. 9 K 4. 5 K

First powering of a Sector 78 main dipoles 45 V The Inductance changes below 120 A! 350 A 3 A/s 15. 5 H 12. 6 H

First powering of a Sector 78 main dipoles The first ramp from 300 A to 350 A

First powering of a Sector 78 main dipoles 7 ppm (100 m. A) No tracking error

First powering of a Sector 78 main dipoles 2 ppm (20 m. A) No overshoot

Main dipole Power Converter: Start Up • Start up must avoid rapid voltage changes that can trigger the QPS • If current is less than 1% of I_MIN then a blocking voltage must be applied during the pre-mag phase – this winds up the voltage loop integrators • This could result in an aggressive start up that could trip the RB QPS so ~6 s open loop voltage ramp is now included to make the start up smoother: New open loop voltage ramp included until I > 1% of I_MIN The FGC regulates d. I/dt by controlling d. V/dt with a proportional controller RB successfully started at 10 A/s Blocking voltage during Pre-Mag

Main dipole Power Converter: Power Off • RB decay from 350 A takes more than 1 hour with the discharge switch closed • New Switch Off algorithm ramps down the current to 1% of I_MIN (< 4 A) before switching off in about 1 minute • The algorithm is also be used for SLOW ABORT End of a Switch Off ramp on RB d. I/dt of ramp down is regulated by controlling d. V/dt with a proportional controller Voltage is reduced in proportion to the current to smoothly end the ramp to 1% of I_MIN Converter is switched off with V_REF = 2% of V_NEG

RB Failure → Unexpected succesful test “Current capture” Earth Fault Detected → RB Trip @ 9 k. A RB OFF => ON @ 5 k. A - Current captured - I Loop closed I. meas → No Need to open 5 k. A I. ref switch saves time → QPS does not trip PO In Action PO specialists mobilized → Problem diagnosed as a weak component Not a TRUE earth fault 1 h 30 Load Time constant 13 740 sec (4 hours). Time 9 k. A → 0 A is more than 7 hours → Repair made on LIVE circuit (5 k. A circulating)

• • • Definition of LHC Powering Individual Tests Short Circuit Tests Hardware Commissioning Main Dipole Current Regulation • 22 -bit ADC Idle Tones • Inner Triplet Powering • Transfer Function Analysis (demo)

RB, QF & QD use high precision ADCS 900 A Up to 20 ppm 820 A Up to 20 ppm • Spikes on the RB/QF/QD current of 20 ppm peak-peak were seen around 820 A and 900 A • All ADCs tested had the same behavior although the DC levels varied slightly • Measurements in the lab revealed idle tones at the same DC input levels as seen in RB • Problem was traced to a nonoptimal digital filter function. A new function was developed and distributed over the network – no hardware change was required Andrea Cantone

Delta-Sigma ADC Idle Tones DCCT ΔΣ ADC Fibre 500 k. Hz 1 -bit FGC FILTER DSP 1 k. Hz 24 -bit • ΔΣ ADCs are the most linear and precise class of ADCs • However, ΔΣ ADCs can be vulnerable to “IDLE TONES” • The choice of digital filter function is critical!

ADC Modulator frequency = 500 k. Hz Modulator Nyquist = 250 k. Hz

- 40 d. B ! FGC Nyquist = 500 Hz FGC Sampling rate = 1 k. Hz

20 ppm glitch Old Filter No glitch New Filter

~0. 8 ppm noise Old Filter ~0. 45 ppm noise New Filter

Latest News • Tests with a 600 A converter on a warm circuit using the internal ΔΣ ADCs show the noise on the regulated current reduces by more than a factor of 3 when using the new filter compared to the old • The new filter will be put into operation by the end of May

• • • Definition of LHC Powering Individual Tests Short Circuit Tests Hardware Commissioning Main Dipole Current Regulation Idle Tones • Inner Triplet Powering • Transfer Function Analysis (demo)

Inner Triplet Commissioning Complex system with interleaved circuits: – – Crucial for machine operation Inductive coupling requires a de-coupling circuit to meet performance targets First tests have begun in sector 56 but more tests required So far so good!

M. Pojer 25/04/2008 - Powering Sector 5 -6 LHC-ICC Powering of the inner triplet at nominal: [11. 4 k. A (Q 2) AND 6. 8 k. A (Q 1/Q 3)]

LHC-ICC Powering Sector 5 -6 M. Pojer 25/04/2008 Provoked quench at nominal for inner triplet magnets

• • Definition of LHC Powering Individual Tests Short Circuit Tests Hardware Commissioning Main Dipole Current Regulation Idle Tones Inner Triplet Powering • Transfer Function Analysis (Demonstration)

Transfer Function Analysis Demonstration with simulated load • A simulator circuit is connected to an FGC in my development lab • The box has 1 -pole at ~0. 1 Hz – Can we find it with the TFA tool?

Transfer Function Analysis TFA Application FGC gateway Ethernet World. FIP Fieldbus Circuit Simulator ←Vref I meas→ FGC

TFA Interface

FGCrun+ Expert Interface