1. Quench levels and transient beam losses at HERA Beam losses in operation (transient, continuous). What beam losses lead to magnet quenches (for top energy AND for injection). How to set BLM to avoid quenches? By Kay Wittenburg, Div. MDI, Deutsches Elektronen Synchrotron DESY, Hamburg, Germany Quench

2. Contents 1.Energy deposition in magnets (quench level) 2.Tolerable loss rates (calculations) 3.Response, calibration and settings of BLMs 4.Quench experiences (comparison with calculations) 5.Avoiding Quenches 6.Conclusions

3. Circumference: 6.3 km Proton Beam: Injection Energy 40 GeV Lumi-Energy 920 GeV Electron Beam: Injection Energy 12 GeV Lumi Energy 27.5 GeV Magnetic field p-ring: 5.1 Tesla at I=5500 A for 920 GeV 422 s.c. dipole magnets 224 s.c. main quads, 400 s.c. correction quads 200 s.c. correction dipoles > 1000 n.c. electron magnets HERA Parameter

4. 1)Energy deposition in magnets a) quench level of a cable (820 GeV/c) Ref 1 Note: All Refs. according to studies around 1984 - 1990! Ref. 1: SUPERCONDUCTING ACCELERATOR MAGNETS. By K.H.Mess(DESY), P.Schmuser (Hamburg U.),.DESY-HERA-89-01, Jan 1989. 62pp. Lectures given at joint CERN-Accelerator School-DESY Course on Superconductivity in Particle Accelerators, Hamburg, West Germany, May 30- Jun 3, 1988. Ref. 2a: M.S. Lubell, Empirical scaling formulas for critical current and critical field for commercial NbTi, IEEE Trans. Mag-19 (1983) 754 Ref 2b: M. A. Green. Calculating the J c, B, T surface for niobium titanium using a reduced-state model.; IEEE Transactions on Magnetics, 25(2), S. 2119–2122, 1989. For NbTi cables (HERA): B=5T (at coil, 4.7 T in gap), T b = He bath temp = 4.4 K critical values: T c (B=0, I=0) = 9.2 K; B c = 14.5 T; T c (B, I=0) = T c (0) · (1- (B/ B c )) 0.59 Ref. 2a current sharing temp.: T cs (B,I) = T b + (T c (B, I=0) – T b ) · (1 – J op /J c ) Ref. 2a critical current: J c = J c (B, T) see Ref. 2b With J op = HERA operating current  0.7 · J c = 5025 A => T cs (B, J op ) = 5.2 K  T c = 0.8 K between He-bath-temp. (T b ) and quench-temp (T cs )!

5. a) quench level of a cable cont. Heat capacity c p of Copper-NbTi composite cable: c p = 10 -3  {(6.8/  + 43.8) · T 3 + (97.4 + 69.8 · B) · T} [mJ/cm 3 · K] Ref 2a  is the superconductor fraction of the cable:  = 0.36 for HERA Type cable => c p = 2.63 mJ/cm 3 · K -1 => E dep = 2.1 mJ/cm 3 is needed for a temperature increase of  T c = 0.8 K (at 820 GeV/c) with minimum propagation zone calculation, still adiabatic: Ref. 1 We took propagation and cooling as a safety margin = 0.15 cm 3 => 6.6 mJ/cm 3

6. a) quench level of a cable cont. Other Refs: (at nom. Energy) (  = 7.9 g/cm 3, area = 0.15 cm 2 ) Tevatron:  T c = 1 K; E dep = 1 mJ/g = 7.9 mJ/cm 3 Ref. 3 SSC Magnets:  T c = 0.6 K ; E dep = 0.2 mJ/g = 1.6 mJ/cm 3 Ref 4 Fermilab Energy doubler:  T c = 1.4 K ; E dep = 9.8 mJ/cm 3 Ref 5 Ref. 3: VanGinneken, A.; FERMILAB-Pub-87/113 Quenching induced by beam loss at the TEVATRON. Ref. 4: Quench analysis of the energydeposition in the SSC magnets and radiation shielding of the low Beta IR quadrupoles. By G. Lopez. In *San Francisco 1991, Proceedings, Accelerator science and technology, vol. 4* 2212-2214 and SSC Dallas - SSCL-426 (91/05,rec.Jul.) 3 Ref. 5: SENSITIVITY OF AN ENERGY DOUBLER DIPOLE TO BEAM INDUCED QUENCHES. By B. Cox,P.O.Mazur, A. Van Ginneken(Fermilab),. FERMILAB- TM-0828, FERMILAB-TM-0828-A, Nov 1978. 19pp. Published in IEEE Trans.Nucl.Sci.26:3885-3887,1979

7. a) quench level of a cable cont. Enthalpy H for all beam energies= energy density: H(T) =  0 T c p dT [mJ/cm 3 ]  H = H(T cs ) – H(T b ) Ref. 5a This figure shows how much deposed energy density is needed to quench the coil depending on the beam energy. T b = 4.4 K T b = 4.0 K Ref. 5a: Fermilab design report, 1979 and The measurement and theoretical calculation of quench velocities within large fully epoxy impregnated superconducting coils / Eberhard, Philippe H; et al, Mar 1981. - 4 p - Published in: IEEE Trans. Magn.: 17 (1981) pp.1803-1806, Lawrence Berkeley Lab. - LBL-12337

8. 2) Tolerable loss rates (calculations) We performed Monte Carlo calculations to simulate the beam loss and the energy deposition in the coils. The critical losses were determined from the critical energy deposition in 1 cm 3 coil volume (hot spot) BLMs cannot protect against instantaneous losses! At Tevatron (Ref. 6) they observe beam loss induced quenches at a continuous loss rate (dose) (/s) 16 times higher than instantaneous losses. Ref. 6: THE TEVATRON BEAM POSITION AND BEAM LOSS MONITORING SYSTEMS. By R.E.Shafer, R.E. Gerig, A.E.Baumbaugh, C.R. Wegner(Fermilab),.FERMILAB-CONF-83-112-E, 1988. Published in Proc. 12th Int. Conf. on High Energy Accelerators, pp. 609-615 [p/s]

9. So far: All calculations valid for dipoles, but losses are expected mainly in quadrupoles Quads: Same cable, same temp., same current, smaller B (increase T cs ) Measurements of quench currents at 4.75 K: Mean (±  ): Dipoles 6458 A (± 114 A); Quads: 7383 A (± 148 A) Weakest: Dipole: 6154 A, Quad: 6518 A Calculated with previous formulas: J c = 6340 A Go on with BLM system conception with previous numbers and use differences as safety margins. Match the BLM response to the cryogenic time constant. Tevatron => 16 ms (Ref. 6 ). Decisions: Measure loss rates in  5 ms intervals = alarm time binning. Definition: critical loss rate/5.2 ms = cont. loss rate · 5.2 · 10 -3 Accepted loss rate  1/10 critical loss rate BLMs on superconducting Quads (+ warm Quads) Tolerable loss rates (calculations) cont.

10. 3) Response, calibration and settings of BLMs

11. Position of BLMs: oCollimators omax. beta function (Quadrupoles near Experiments, normal conducting oEach superconducting Quadrupole (Quench) Response, calibration and settings of BLMs cont. Monte Carlo calculation (Gheisha 8) to determine the response of a BLM during a beam loss in the center of the quadrupol. Bad: Different MC-calculations of deposited energy in coils and in BLMs done by different person!

12. Response, calibration and settings of BLMs cont. Results:  Independent of azimuthally loss position,  Independent of loss angle,  Shower distributed along 1 m, therefore somehow independent of the longitudinal loss position within the quad.  Choose position near end of quad due to mechanical reasons.

13. Response, calibration and settings of BLMs cont.

14. Response of BLM: Fit includes: Sensitivity of the BLM to radiation created by lost protons Amount of radiation at BLM position created by lost protons Ref 7: S. Schlögl, II Institut für Experimentalphysik der Universität Hamburg Einsatz von PIN-Photodioden als Protonen- Strahlverlustmonitore bei HERA Diploma thesis, DESY-HERA-92-03 Ref. 8: F.Ridoutt, II Institut für Experimentalphysik der Universität Hamburg Das Ansprechvermögen des PIN Dioden Strahlverlustmonitors Internal Note: PKTR note No. 91 (1993) Counts/lost proton [·10 -6 ] protons/count [·10 4 ] 1.25 1.67 2.5 5.0 14.2 BLM calibration at the superconducting quadrupole Momentum p [GeV/c] simplified fit: 1 count = 1.5*10 7 /p [lost protons]

15. The BLM PIN Photodiodes to satisfy the special conditions in HERA Two PIN-PDs in coincidence to count charged particles Signal (in Si): dE/dx = 3.7 MeV/cm 3.7 eV/e-hole pair => 10 -15 C/MIP => 10 000 e - /MIP Small dimensions: Area: 2.75 · 2.75 mm 2 or 20 · 7.5 mm 2 Sensitive and fast amplifier with low noise and with a fast coincidence following Efficiency to charged particles: 30% TTL output for counting Very low noise: Dark count rate < 0.01 Hz max. count rate > 10.4 MHz Very high dynamic range: >10 9 Insensitive to synchrotron radiation: Efficiency to  : 3.5 · 10 -5 Coincidence + lead: 1 Hz at 1.5 Gy/h (e- ring at max.) Diodes Pre-ampl. Video ampl. Comperator +5V +24 V Bias TTL driver -5V Threshold -5V

16. 4) Quench experiences (comparison with calculations) Two types of thresholds : 1.1/10 of critical loss rate produce an alarm (rates varies with energy) 2.More than 4 BLMs above 1. dump the beam (for all energies above injection) Includes the sensitivity to showers

17. Note: A quench in HERA is not a disaster! It takes typ. 1-2 h to recover from cryogenic  = 200 Quenches HERA experience with Very fast losses (< 5ms) Slow losses More about failures: K. Wittenburg (DESY): Beam loss & machine protection 33rd ICFA ADVANCED BEAM DYNAMICS WORKSHOP on HIGH INTENSITY & HIGH BRIGHTNESS HADRON BEAMS Bensheim, Germany

18. No quench Dump due to losses HERA BLM Alarm System

19. Mean value of 666 ms = 300/5.2ms Max value in 666 ms = 4200/5.2ms

20. Quench at OL 198 HERA ring view Short mode (5.2 ms / bin) Long mode (666 ms / bin) Mean = 500/5.2ms Max = 1300/5.2ms

21. Fast losses, but > 5.2 ms Quench 40 s after losses! (note: here long mode) Quenches due to very fast losses < 5.2 ms were not taken into account due to possible saturation of count rate. Max. count rate = 180 Bunche/turn Other loss types:

22. Quench experiences (comparison with calculations) cont.

25. More statistics 5) Avoiding Quenches, what to do: Very fast losses (< 5ms) Slow losses BLM- AlarmsQuenches 999205 1994-2004 Faster Alarms Software enable BLMs

26. ALIs Alarm loop-Zentrale Alarmloop DUMP BLMs + BPMs +Alarm-modules “Alarm-Loop- Interface” Adding more and faster alarms to avoid 5 ms events New Beam-Loss-Alarm-Topology at DESY Internal Power- Supply- Alarm Galv. Trenn. faster Active New ACCT-Alarm DCCT-Alarm Faster clock rate Magnet current- Alarm More Failure inputs: PS, HF, …

27. More sensitive thresholds on all BLMs Less than 4 BLM Alarms required to dump the beam => might be more sensitive to malfunction of BLMs o add. dump criteria: Long losses at one location near threshold. Increase weight of BLMs at collimators and other aperture limits (e.g. dispersion) More reliable Alarm system Block injection in a not well prepared machine Better educated operation crew => will help in any case Avoiding Quenches, what to do: cont. Some more proposals, not (yet) tested at DESY

28. Conclusions Threshold of Quenches is about factor 5 below calculated critical beam loss (count) rate (response of BLM). All safety margins were eaten up. –Possible Reasons: Simulations (no magnetic field) Loss position inside Quadrupol Inaccurate parameters in calculation … Quench probability depends on beam losses, not on weak magnets. Threshold of 4 BLMs is to rigid, sometimes only 1 BML is affected. Very fast losses (< 5 ms) can occur.

29. The End now open for discussion

32. start after 5 month shutdown

33. start after 5 month shutdown (Lumi upgrade) All by 5 ms PS failure events

34. Many thanks to R. Bacher (DESY) K. H. Mess (DESY/CERN) M. Swars (retired) K. Willmer (->) S. Schlögl (->) F. Ridoutt (DESY) H. Schultz (retired) P. Duval, H. Wu, M. Lomperski (DESY) a lot of DESY colleagues