LAV HV repair plans Matthew Moulson NA62 Photon Veto Working Group Meeting CERN, 13 October 2015

LAV HV situation at end of run Discharge problems observed for all flanges for LAVs in vacuum: Approximately half of connectors affected Problem concentrated on connectors with HV > 1400V on multple pins Symptoms of discharges: Individual tripped channels Clusters of tripped channels Unstable HV boards (a significant contributor to downtime): Many channels affected Generally, access required to reseat board Possible to try rebooting HV crate remotely, but not always effective and generally difficult to restart OPC server

LAV HV situation at end of run By end of run, all flanges had been cleaned at least once A few several times (LAV4, 7, 8, 10) Plug connectors for affected feedthroughs contaminated with gold flakes and debris A12 not affected – all connectors in perfect shape Near end of run, HV lowered for LAV01-11 channels with highest voltage 13/10: HV capped at 1400V (167/2240 ch affected) 20/10: LAV04, 05 capped at 1350V (23/295 ch additionally affected)

Long-term solutions Conclusions from inspection & discussions of 23 Sep: R. Cometti (Vaqtec), A. Fuierer (SST), D. Tagnani (Frascati) Current flanges are not recoverable and will have to be replaced Unlikely that a high pin-density vacuum feedthrough fully certified for long term use at 1.5 kV exists on the market Need to invent a new solution Focused on observation that discharges occur from male socket pin (or female plug pin) to metallic base of feedthrough Continue to use 32-pin MIL-DTL-26482 format Cover or eliminate metal between pins at base of feedthrough Solutions: 1. Base entirely made of PEEK (Vaqtec) 2. Metal base with glass-ceramic inserts covered by insulator (SST) Ceramic inserts enlarged to ~1.3 mm in radius (instead of 1 mm) Metal base covered by silicone potting compound Silicone sleeves covering bases of male pins

Long-term solutions Conclusions from inspection & discussions of 23 Sep: R. Cometti (Vaqtec), A. Fuierer (SST), D. Tagnani (Frascati) Current flanges are not recoverable and will have to be replaced Unlikely that a high pin-density vacuum feedthrough fully certified for long term use at 1.5 kV exists on the market Need to invent a new solution Focused on observation that discharges occur from male socket pin (or female plug pin) to metallic base of feedthrough Continue to use 32-pin MIL-DTL-26482 format Cover or eliminate metal between pins at base of feedthrough Solutions: 1. Base entirely made of PEEK (Vaqtec) 2. Metal base with glass-ceramic inserts covered by insulator (SST) Ceramic inserts enlarged to ~1.3 mm in radius (instead of 1 mm) Metal base covered by silicone potting compound Silicone sleeves covering bases of male pins

Vaqtec feedthoughs with PEEK base Cannot show construction drawings by request from Vaqtec Dimensions are approximate

Vaqtec feedthoughs with PEEK base 3 prototypes mounted on DN40 CF flanges delivered to Frascati 04/12 Vacuum side Air side

Vaqtec feedthoughs with PEEK base 3 prototypes mounted on DN40 CF flanges delivered to Frascati 04/12 Vacuum side Air side

HV test program for new feedthroughs Not much time to validate solutions: Test as best we can in available time Measurement of leakage current at 3 kV, pin-to-pin and pin-to-ground Endurance testing of 2 feedthroughs: all pins at 2kV for ~1 month Additional considerations: Accelerate aging by application of thermal cycles? Concerns about hygroscopicity? PEEK hygroscopicity very low but not zero (0.1% water absorption) Difficult to measure – need control over relative humidity

Vacuum test program for new feedthroughs Prototype feedthroughs being leak tested at Frascati (2 of 3 so far) 1 feedthrough supplied with defective weld (visible to eye) 1 feedthrough satisfactory so far: First leak test: Q = 3 × 10−9 mbar⋅l/s 1 thermal cycle to 70°C for 4 hours Second leak test: 3 × 10−9 mbar⋅l/s 2 additional cycles to 80°C currently in progress Results consistent with Vaqtec specifications and very satisfactory For 88 connectors Qtot = 2.6 × 10−7 mbar⋅l/s 5 orders of magnitude less than Straws, Blue Tube, or LAV vessels Real challenges remain: Leak test of all welds before installation Vaqtec responsibility, but leak found in first prototype! Installation of flanges must be done very carefully – may be complicated to identify leaky gaskets

Replacement of plug connectors Plug connectors damaged by discharges and will need to be replaced Flaking of gold plating; gold flakes lodged in holes Oxidation of pins Surface damage to insulator body (insert) In some cases, micro discharges have been observed with cables not connected to flanges Experience with A12 suggests that plug connectors are not the source of the problem Can use the same commercial connectors as before Connectors will be replaced by SPIE (€4000) Replacement work has already started: Emilio and Sauro at CERN week of 30/11—04/12 to remove all HV cables Cables delivered to SPIE in Bldg 286 SPIE currently setting up test equipment to validate cables Connectors ordered from RS (€4000), current delivery date is 05/01/16

Schedule This schedule is extremely tight, in particular: 04/10 (done) Uncable LAV HV, bring cables to SPIE 05/01 Plug connectors delivered. SPIE starts work on cables. 08/01 Last date to order flanges for feedthrough delivery 01/03 01/02 (approx) New HV cables ready 2 weeks needed to cable up to HV flanges 01/03 Feedthroughs ready at Vaqtec, installation in flanges 15/03 Flanges ready for installation and testing 01/04 Begin to evacuate blue tube. All work completed This schedule is extremely tight, in particular: Flanges must be ordered by 08/01 – not enough time to fully validate new feedthroughs Only 2 weeks for installation and test of 11 HV flanges

Addendum: A4531 HV PS failures “Ruggedized” PS received from CAEN 21/10 MOSFETs in sub-PSs replaced with components from new manufacturer Installed in LAV07 on delivery; moved to KTAG on 22/10 PS on LAV02 failed 3 times between 24/10 and 01/11 Ruggedized PS moved to LAV02 on 01/11 No further failures on LAV02 until end of run Meeting with G. Di Maio tomorrow: decide how many PSs to “ruggedize”