Transcript SnubberStatusReportMidApril2011x

Status Report related to Testing, Installation, Commissioning and Operation of the
Capacitive Snubber Circuits in the LHC Main Circuits – April 2011
Reminder:
-Purpose of inserting capacitive snubbers across the LHC extraction switches:
• to obtain a fast transient suppression by eliminating the erratic (and unpredictable) behavior of the heavily
oscillating voltage from the electric arc across the switches during the commutation process at switch opening
• to slow down the front of the voltage waves which penetrate into the magnet strings so to reduce the effects
on the quench signals along the chain (U_qso and U_mag) .
-or, with other words: to mitigate the effects of the difficult coexistence between heavy electrical machines and
the ultra-sensitive quench protection electronics. This is a further measure, adding to the introduction of electronic
delays between the switch openings and with respect to the cutting-off of the power converter.
• Other benefits: Reduce emi emitted noise at its origin, decrease the erosion of the arcing contacts and, herewith,
improve the life expectancy of these elements.
Example of final installation
inside DQS cluster (UA67).
High-current test bench in P-Hall
(Bldg 377) for type- and routine
testing of all LHC snubber systems
Status report on behalf of the Snubber Project Team:
1
Bozhidar Panev, Fabio Formenti, Mathieu Favre, Noel Fournier, Mikhail
Ovsienko, Sergey Avramenko, Pawel Dubert, Jozef Setkowicz, Solomon Goldstein and Knud Dahlerup-Petersen - with assistance by the EPC, P-Hall
team: Hugues Thiesen, Xavier Genillon, Gerard Calegari, Jacky Spiller and Bernard Dubois.
Knud 21.04.2011
Some construction details – a reminder:
Capacitors are of dry, thin-film type, metalized polypropylene dielectric (MKP), self-healing design. Impregnation by
solidifying PUR resin. With over-pressure gauge against inflation rupture. Sealed aluminum housing.
One capacitor per DQS branch, 4 x 13 mF, 1500 V Dipoles, 4 x 40 mF, 350 V Quadrupoles per switch assembly.
Each capacitor will be charged through a dedicated protection circuit - with power fuse, anti-return diodes and their
commutation assistance circuits and two discharge resistor circuits. The diodes prevent ringing between the snubber
capacitors and the main circuit inductance.
One such circuit per DQS branch
During LHC powering the switches are closed and the capacitors are short-circuited (30 µΩ). Capacitor charging
occurs during the switch opening process and takes typically a few ms. Discharge takes place through the resistors of
the protection circuits, the fuse and the dump resistor (τ = 140 s) and in the dedicated discharge resistors (τ = 435 s).
2
Some construction details (contd):
Fuse selection determined through
- PSpice calculation of charge-current profile
-Conversion of the pulse into rms current
- Superposition of the rms pulse onto the
fuse characteristic as given by the producer
- Confirmed by measurements using Pearson
current pulse transformer
- A blown fuse gives warning through μswitch and ACQ system
3
Some construction details (contd):
The 5 stages of breaker commutation: 1) Tmc: the electrical and mechanical reaction time
2) Tac: separation of the main contacts, current flows through the arcing contacts 3) Tbr: release of
the pressure between the arcing contacts and creation of melted metal bridges 4) Tarc: start of the
plasma phase with arc 5) Tgd: the glow-discharge with de-ionization of the exhaust gasses, -the role
of the arc chamber and the muffler.
Schematic
VAB49 LHC dipole
extraction switch
4
Model for Commutation
Calculated impedance
The Snubber Principle:
The best way to achieve arc elimination is to prevent that it occurs
This can be achieved, at least partially, by “starving” the breaker for current
Itotal = 4 Iswitch + 4 C dV/dt + Udump/Rdump
However, a high dV/dt during commutation is contradictory to the wish to slow down the voltage rise. This dV/dt is
given by the switch/dump resistor properties and depends on the current level.
5
The Test Program:
Already Completed:
• Type tests performed at the dedicated Test Bench in P-Hall on the prototype dipole system at
currents up to 13 kA (with 4 mH test load)
• Type tests performed at the dedicated Test Bench in P-Hall on the prototype quadrupole system at
currents up to 13 kA
• Endurance Tests - 50 cycles at Inom (12 kA), Rdump = 70 mΩ applied to the prototype dipole system
• Special fuse blowing tests carried out at 2, 3, 9 and 13 kA on the prototype dipole system, including
verification of continued operation with only 3 capacitors (simulating one fuse blown)
• Verification of the gain with 5, instead of 4, capacitors per switch system
• Routine tests applied to the complete series of 15 further dipole systems prior to their installation in
the LHC.
• Commissioning up to 6 kA of all 16 dipole systems installed in the LHC, with S56 as pilot systems
The purpose of the commissioning in the machine was:
- to verify the correct operation of the systems incl. their associated auxiliary circuits once mounted in the machine
- to checking the capability of the snubber capacitors to assure arc-free opening of the extraction switches up to 3.5 TeV
- to get a quantitative measure for the benefice obtained for operation of the LHC quench protection systems
(iQPS & nQPS)
- to observe any difference between the sectors
The commissioning tests were performed with
- FPA from current plateau at 760 A
- FPA from current plateau at 2’000 A
- FPA during ramping at  2’000 A
- FPA from current plateau at 4’000 A - skipped after successful testing in first two sectors.
- FPA from current plateau at 6’000 A
The Remaining Test Program (see also last slide – future program):
6
• Endurance tests to be applied to the prototype quadrupole system
• Special fuse blowing tests to be applied to the quadrupole system
• Routine testing of the 15 further quadrupole systems – foreseen for completion before end of May
From Type Testing in Power Lab
Dipoles
7
Results 1/13:
From Type Testing in Power Lab
Dipoles
Results 2/13:
Example of the voltage rise across a dipole extraction switch during commutation at 6’000 A in case of NO
Snubber Capacitors. Heavy arcing during  4 ms. From Sector 67.
8
From Type Testing in Power Lab
Dipoles
Results 3/13:
Example 1: 6 kA
Example 2: 12 kA
9
From Type Testing in Power Lab
Dipoles
Results 4/13:
9 kA the four branches
9 kA the four branches
Capacitors stay charged in
considered time interval
10
From Type Testing in Power Lab
Dipoles
Results 5/13:
Voltage rises across three extraction switches during commutation at 6’000 A with Capacitive Snubber
Circuits connected.
Arc-free commutation is proven by:
1) At any point the total circuit current (pushed by the magnet chain) equals the SUM of the capacitive current
(= 4Csnubber dV(t)/dt) and the resistive current in the dump resistors (= Udump(t)/Rdump).
2) The voltage rise is exponential, with the expected time constant τ = Rdump 4 Csnubber = 8.0 ms (measured 7.9 ms).
3) The voltage curves are exempt from any erratic part. No arc means no current in the extraction switch.
At the very beginning of the voltage rise the current absorbed by the 4 snubber capacitors equals the full system current.
11
From Type Testing in Power Lab
Dipoles
12
Results 6/13:
Results 7/13:
From Type Testing in Power Lab:
Dipoles
5 x 13 mF capacitance
dV/dt peak = 100 V/ms
4 x 13 mF capacitance
dV/dt peak = 100 V/ms
13
Results 8/13:
From Type Testing in Power Lab
Quadrupoles
14
Results 9/13:
Results from Commissioning of the systems in the machine
15
From iQPS U_qso
(Aperture Voltage Difference)
2’000 A Sector 56 Without Snubbers
From 2010 FPA during a Ramp (10A/s)
From ‘snapshots’ – all dipoles of S56
From iQPS U_qso
(Aperture Voltage Difference)
2’000 A Sector 56 With Snubbers
04.02.2011 FPA during Ramp (10A/s)
16
Results 9/13:
From nQPS – DS (SymQ) unfiltered data
Umag - all arc dipoles Sector 67 (R6,L7)
6’000A Without Snubbers
From 05.02.2011 FPA after cutting PC
From nQPS – DS (SymQ) unfiltered data
Umag - all arc dipoles Sector 67 (R6,L7)
6’000A With Snubbers
14.02.2011 FPA after tripping PC
17
Conclusions I:
1. The snubber capacitors do the job they were expected to do:
- For I < 6000A electrical arcs in the breakers are basically eliminated
- For I > 6000A arcing is strongly reduced, with arcing times at 13 kA of  1ms (compared
to 5-6 ms without snubbers)
Up to 9 kA the arc is so small that it remains almost invisible on the voltage waves
- Smoothening of the voltage waves during the commutation is achieved
- The front of the voltage waves is slowed down.
2. The snubber auxiliary circuits performed correctly:
- The protection fuses never ruptured during normal operation as well as under 3-branch
testing
- The fuses blew for I>2 kA when a deliberate short-circuit of the capacitor was introduced
- The selected fuse types and ratings appear the correct choices
- The diode protection prevented oscillations, also in case of a capacitor short-circuit
(simulations have predicted such ringing if no diodes). The fast recovery diodes and their
snubbers seem sufficiently dimensioned.
3. The system can continue to operate correctly with one branch capacitor in open state.
4. The 50 cycle endurance test was completed without any incidents.
5. The results from simulation of the test bench behavior (E. Ravaioli) matches well the measured
voltage and current waveforms.
6. The routine tests on the 16 dipole systems were all successful. No shortcomings were detected.
7. Switch overvoltage tests showed large breaker voltage margin for dipoles (arc at 1350 V successfully
broken). For the quadrupole breakers, however, 220 VDC appears as a hard limit (after which arc
extinction becomes long and problematic).
18
Conclusions II:
1. The snubber systems reduce the U-qso peak values by typically 40%, bringing them well below the
detection threshold level of 100 mV in all sectors. In some cases the aperture voltage differences due
to switch openings are now inferior to those originating from the power converter switch off during
ramping.
Proof of improvement: Dipoles A23.R2 and B23.R2 in S23 do no longer trigger the iQPS system upon
switch opening (such as experienced on 01.09.2010 during provoked FPA).
However, a small number of dipoles with large cable differences between the two apertures will still
require the installed, higher threshold (eg 240 mV).
2. The snubber systems reduce significantly the two disturbing transients of the Umag signals, occurring
during commutation of the DQS switch systems, used by the algorithm of the nQPS – DS (SymQ)
detector.
19
Future Program – Completion of the Task:
Proposed Continuation of the Snubber Program for the Main Quadrupoles:
- The majority of components are already at CERN (all capacitors, diodes and their commutation aid components,
discharge resistors).
- Procurement of remaining components (fuses, micros-switches, raw materials): Ongoing, delivery by 1 June 2011.
- Manufacture of the protection circuits: Ongoing, by AGH/UST team (Cracow), Workshop B117 and CERN Central
Workshop. Completion by 15 June 2011.
- Assembly of the 64 protection circuits: By AGH team + QPS staff, completion by 1 July.
- Endurance tests and fuse rupture testing on one quadrupole system: By QPS staff, early May.
- Routine testing of 15 systems: 15 June – 15 July: By QPS staff.
- All systems ready and packed for dispatch and installation: 1 August.
Storage in B287.
- Installation foreseen during the Long Shutdown in 2013. By teams from IHEP, Protvino and AGH, Cracow, followed
by complete commissioning.
END