Needs for MKE PFN discharge issues and options

Needs for MKE PFN discharge: issues and options Etienne Carlier & Thomas Kramer ABTEF 24 -10 -2017

MKE(4) Architecture SLOW TRIGGER (-15 ms) FAST TRIGGER (-50µs)

BETS INHIBIT FAST TRIGGER SLOW TRIGGER FAST & SLOW TRIGGERS Extraction BIS INHIBIT MKE(4) – Machine Protection SLOW TRIGGER inhibited only by BIS FAST TRIGGERS inhibited by BIS and BETS

Issue How to avoid to go into a deadlock situation?

Normal Cycle - SLOW - Initial condition OK - Vs = 0 k. V - First BIS window OK - RCPS triggered (SLOW TRIGGER) - PFN charged

Normal Cycle - FAST - Second BIS window OK - BETS window OK (Vs = 27. 790 k. V) - MAIN switch triggered (FAST TRIGGER) - PFN discharged

Normal Cycle - FAST - BIS and BETS window OK - Fast pulses - MAIN switch, Short-circuit and TDR - Kick synchronization with beam

Bad cycle - Missing second BIS window (SLOW) - Initial condition OK (Vs = 0 k. V) - First BIS window OK. - RCPS triggered (SLOW TRIGGER) - PFN charged

Bad cycle - Missing second BIS window (FAST) - Second BIS window NOT OK (machine protection) - BETS window OK (Vs = 27. 790 k. V) - FAST TRIGGER inhibited by BIS - Slow PFN discharge

Bad Cycle – Missing BETS window (SLOW) - Next cycle after missing second BIS window event - Initial condition NOT OK. Vs = ~2 k. V instead of 0 k. V (slow discharge) - First BIS window OK - RCPS triggered (SLOW TRIGGER)

Bad Cycle – Missing BETS window (FAST) - Next cycle after missing second BIS window event - Second BIS windows OK (machine protection) - BETS window NOT OK. Vs = 27. 0 k. V instead of 27. 8 k. V (+1. 5% tolerance) - FAST trigger inhibit by BETS. - Slow PFN discharge

Options How to come out of a deadlock?

Do nothing, just wait… • System put in FAULT after three consecutive occurrences of FAST TRIGGER inhibition • Once in FAULT slow discharge of the PFN through the internal circuit impedance (HV divider) • PROS • Clear indication of the source of the failure • CONS • Put system in FAULT • Action required from CCC to restart the system • Downtime (detection + action + reaction)

Decrease circuit impedance • Add bleeding resistor in parallel with PFN • Today discharge rate is ~93% in 15 sec (3 RC not reached) • 100% in ~4. 8 sec to be achieved if we want to be fully PPM (worst case) • PROS • Keep the installation ON • Simple and passive • CONS • Impact on kick strength stability to be checked (variable RCPS length on LHC cycles induced by re-phasing)

Add a DUMP switch • Replacement of the diode with a DUMP switch in order to control the PFN discharge through the TDR • PROS • Keep the installation ON • Clean PFN discharge mechanism (as implemented for others systems) • CONS • Increase operational voltage (if magnets have to be terminated too) • No protection against MAIN switch erratic

Add CLIPPER switch • Discharge of PFN though a “controlled” short circuit (Thyratron) • Fast reaction (a few hundred of ns) • PROS • Keep the installation ON • Can protect the system against MAIN switch erratic • CONS • Another source of erratic • Can affect the beam in case of erratic during MAIN conduction (reflection between magnet short-circuit and clipper switch in conduction) • Not designed for repetitive use

Add a crowbar circuit • Slow controlled PFN discharged mechanism through a solid state switch crowbar (and a low resistance path). • Reaction time in the range of a few hundred of ms (discharge current limitation). • PROS • Keep the installation ON • Not a source of erratic • Designed for repetitive use • CONS • No protection against MAIN erratic • Additional interlocking logic needed to avoid doing a resonant charging while the crowbar is in conduction

Inhibit • Inhibit RCPS SLOW trigger if PFN voltage not equal to 0 V • PROS • Keep the installation ON • Low-level • CONS • More than one cycle lost (two? )

Others • Mechanical discharge switch (ross relay + discharge resistor) • Can we use the PFN electrical discharge switch for this? • Pulse the system in the BEAM_OUT • … • Machine protection consideration…

Summary • System reacts to machine protection interlock (BIS and BETS) as expected • System falls in a deadlock situation as expected • Current deadlock exit strategy works as expected But • Current approach is annoying for operation once BIS inhibitions start to occur too frequently • Different options to solved or improve the situation identified • PRO and CONS of each option to be analyzed more into details in order to find the best strategy • Final solution is probably a combination of more than one single option