Diodes qualification: Results and plans for LHC and spares MSC Technical Meeting on 2013-12-12 F. Savary, on behalf of the diode working group, with special

Diodes qualification:Results and plans for LHC and spares

MSC Technical Meeting on 2013-12-12

F. Savary, on behalf of the diode working group, with special contributions from M. Bajko, M. Bednarek, Z. Charifoulline,K. Dahlerup-Petersen, G. D’Angelo, G. Dib, C. Giloux, L. Grand Clement, R. Moron Ballester, V. Roger, A. Verweij, G. Willering

Diodes qualification: results and plans for LHC and spares 2

Outlook

• Reminder• Why worry about the protection diode?• Description of the diode stacks

• Quadrupole diode stacks• Dipole diode stacks• COMSOL analyses + Measurements in SM18• Measurements in the LHC machine• Plan for spares – New production and tests


Why worry about the protection diode?

Quadrupole diode lead resistance during a quench @ 5 kA in the LHC machine (first half of 2011): there is a sudden jump of resistance, peculiar and permanent, which could not be explained with the expected behavior of the diode stack

Step in Dt<50 ms

D16R5 a

Courtesy A. Verweij.


Quad. Diode: key parts, and contacts• Diode bus bar• Nickel plated everywhere

• SSS bus bar• Silver plated over 40 mm

• Connection plate• Nickel plated

everywhere

RBB-CP < 2 µΩ

RHS-BB < 2 µΩ

RDiode-HS < 5 µΩ


Dip. Diode: key parts, and contacts

Lower diode bus bar

RHalf-moon

< 2 µΩ

RHS-BB

< 2 µΩ

Diode box, He content: 5 liter

RDiode-HS

< 5 µΩ

Voltage taps on the diode ‘press pack’

Upper Heat Sink

Lower Heat Sink

• Bolted contacts with 4 M6 + helicoil, 4 spring washers, 10 Nm• Diode compressed by a stack of spring washers to 40 kN

All copper partsNi-plated, 2-3 µm


Diode to heat sink behavior• RDiode-HS : has typical behavior,

confirmed by production data, tests in Bloc 4, and more recent tests @ 4.2 K in SM18

Diodes from S34All resistances go up at 5 or 10 kA

Most of the resistances go down at 13 kAExcept for all 16 quad diode resistance (4 stacks measured at the same time).

Rmax up to 43 µΩ at 5 and 10 kA

0

5

10

15

20

25

30

35

40

45

50

1300A

5 kA 10 kA 13 kArun1

13 kArun2

13 kArun 3

13 kArun 4

13 kArun 5

13 kArun 6

13 kArun 7

13 kArun 8

13 kArun 9

13 kArun 10

Max

imum

resi

stan

ce (µ

Ω)

Measurement run

R_diode-heatsinkBloc 4 measurements

S34 diodes

Data recovery and analysis: G.Willering, TE-MSC

MQB0112

Courtesy V. Roger, G. Willering


Quad. connection plates

• Connection plate• Made of Cu ETP• Ni-plated

• Two screws per contact – M5 • Three spring washers – CuBe• Screws made of St. Steel A4-70 (Rp0.2

= 700 N/mm2)• Thread lock (Loctite)• Bus bars

• Magnet side: Ag-plated• Diode side: Ni-plated• Tightening torque 8 Nm

(break @ 12 Nm)

Original design


Quad. connection plates – BehaviorOriginal design

Courtesy V. Roger et al.


Quad. connection plates

• Connection plate• Made of Cu ETP• Ni-plated

• Two screws per contact – M5 • Three spring washers – CuBe• Screws made of St. Steel A4-70

(Rp0.2 = 700 N/mm2)

• Thread lock (Loctite)• Bus bars

• Magnet side: Ag-plated

• Diode side: Ni-plated• Tightening torque 8

Nm (break @ 12 Nm)

• EDMS: 1281434• Connection plate

• Made of Cu OFE (higher RRR)• Ag-plated

• Two studs per contact• M5, screwed in a back plate (made of Inconel

718)• Made of Inconel 718, of much higher yield limit

(1100 N/mm2), preload increased with same M5

• Tightening torque 12 Nm (break @ 20 Nm)• Add washer plate to distribute better contact

pressure• Two spring washers – Inconel 718• Thread lock (Loctite orapi 303)• Bus bars

• Magnet side: Ag-plated• Diode side: Ni-plated

New designvs.Original design


Quad. connection plates – BehaviorOriginal design

New design - Stable Due to higher resistance of Cu at RT


Thermal cycle


In addition …

• Shims• To (re)center the diode

stack ... needed for the installation of the back plates

• New parts (cross)• For electrical insulation of

each pair of bus bars one w.r.t. each other, especially at the bottom

Courtesy L. Grand Clement


Procurement challenges• Material (un)availability for the fasteners made of Inconel 718,

even though not exotic!• Contractor

• Youparts-DE, …and Fdb-Mechanik-CH• Issues during manufacture due to the hardness of the material• The contractor is not the manufacturer, and the manufacturer has

subcontracted to its daughter company in SL,… + lack of QC

• Mitigation actions were launched: other contracts were placed with two other firms• Bold King (UK), delivery expected next week, 540 threaded plates

(back plate)• Coudray (F), for 1000 studs and 500 threaded plates. This firm may

become a fall back solutionCourtesy R. Moron

Ballester


Consolidation status

Consolidated assembly

Original design

S12 S23 S34 S45 S56 S67 S78 S810

10

20

30

40

50

60

4751 50

47 4751 51

47

0 0 2 0

4751 51

31

TotalTo doDone

180 done out of 391, i.e. 46%

SMACC START Courtesy L. Grand Clement


NCR’s regarding insulation sleevesMissing cutouts (QBQI.30R7)Wrong angular position

Insulation after consolidation

Wrong position - glued



Other NCR’sQBQI.30R7: bare copperin lieu of silver coated bb

(magnet side)

Nickel coated bb in lieu of silver coated bb

(magnet side)

Contact resistance HS to BB > 2 µΩ• QBQI.21R8 (A: 20.77 µΩ / D: 4.75 µΩ)• QBQI.21L8 (A: 3.67 µΩ)

Resolution: extraction of the diode, and contact redone


• Contact resistance diode to HS > 15 µΩ in QBQI.31R6 (A: 19 µΩ)

• The diode stack was replaced


Dipole diode ½-moon – Behavior @ RT

• From coordination of LHC cryodipoles (http://lhc-dipcoor.web.cern.ch/lhc-dipcoor/new_reports/dioderepairs.html)

• Following repair @ CERN, after delivery from Cold Mass Assembler, before installation

• 255 magnets from the 3 CMAs

0 50 100 150 200 2500

0.5

1

1.5

2

2.5

3

3.5

4

4.5AnodeCathodeSpecification

Magnet #

Ha

lf-m

oo

n r

es

ista

nc

e [

µΩ

]

http://lhc-dipcoor.web.cern.ch/lhc-dipcoor/new_reports/dioderepairs.html

http://lhc-dipcoor.web.cern.ch/lhc-dipcoor/new_reports/dioderepairs.html


Dipole diode ½-moon – Behavior @ RT

• CMA series 2000, data to be checked, also for the other 2 CMA’s series 1000 and 3000


Dip. diode ½-moon – Behavior @ cold• Measured for the first time in SM18 in Feb. 2012• Are stable at cold / high current

R < 2 µΩ required R < 1 µΩ measured



In addition …• COMSOL simulations were carried out to

determine hot spot temperatures in the MB and MQ diode leads

• Measurements were done @ cold in SM18• Measurements are being carried out, in the

framework of SMACC, in the LHC machine, to determine the actual resistance of the diode leads in the dipole circuits … @ RT• Done in S67• On-going in S56• The remaining 6 sectors will be done early next year


Rc [mW] Thot [K]

9 350

11 400

13 4500 2 4 6 8 10 12 14 16 180

100

200

300

400

500

600

12kA, 104s, adiabatic

12.5kA, 120s, cooling

12.5kA, 120s, SM18

RHM, µΩ

Tmax

, K

Current 12 kA

Decay time constant 104 s

RRR (all copper parts) 80

Cooling to helium None (adiabatic)

RHM=2,5,10,15mW, no cooling

TMAX over all volume

TMAX of contact area

COMSOL simulations – ½ moon

Courtesy Z. Charifoulline, A. Verweij


Rc [mW] Thot [K]

20 220

50 300

100 525

Current 12 kA

Decay time constant 104 s

RRR (all copper parts) 80

Cooling to helium None (adiabatic)

RHS-BB=2,20,30,50,70,100mW,no cooling

TMAX over all volume

TMAX over Bus Bar

0 20 40 60 80 100 1200

100

200

300

400

500

600

Rc, uOhm

Tmax

, KCOMSOL simulations – BB to HS

Courtesy Z. Charifoulline, A. Verweij


Cold tests in SM18 – Feb. 2012



Measurements in the LHC machine

• Measurement procedure, see EDMS 1334252, without opening the diode container• The measured magnet is powered via the I-taps of the neighboring magnets of the

same line. The test current is limited to 10 A.• Measurements through the voltage taps (D20 and P10 connectors) of the measured

magnet are then carried out, in both conducting (10 A), and reverse (1.5 A) modes• Copper contributions are subtracted

MBB dipole

Courtesy M. Bednarek

MBB dipoleMBB dipole

MBA dipole MBA dipole MBA dipole


Results from S67 ‘as recorded’, …

• Question: where is the high contact resistance?

Courtesy M. Bednarek


… Compared to S23, and Frascati

• Beware that the measuring conditions, and the quantities measured are different• Notwithstanding the above, the ‘signature’ is comparable• S23 is special ‘after CSCM’

Courtesy M. Bednarek, G. Willering,


Another view on these data …

Courtesy M. Bednarek, V. Roger

Let’s look at the production data of the 30 ‘worst’ diode stacks!


0 500 1000 1500 2000 25000

2

4

6

8

10

12

14

16

Resi

stan

ce (μ

Ω)

Another view on these data …• Generally, the higher the resistance at cold, the higher the R at RT• MDB0273 was excellent ‘as produced’• Either it has degraded, or another contact is responsible for the

higher R

Entire production, Frascati, RDIODE-HS, @ cold,10th 13 kA decay

S67RDIODE-HS + RHS-BB + RHM,

RT, Low current

MDB0273 Anode side

Courtesy V. Roger


ARE WE SAFE?

YES


Why can we feel confident?• If any of the resistances approaches 14 µΩ, it is still ok at

12 kA, 104 s, adiabatic conditions, RRR = 80, worst-case conditions (ref. to COMSOL simulations)

• For the moment, the highest resistance over 3 contacts found in S67 is 14 µΩ, i.e. ok

• All the half-moon resistances were below 2 µΩ at the installation of the diode stacks in the magnets. Measurements data are available. The ½-moon assembly is robust and stable (4 x M6 on thick flange)

• There will be other opportunities with diode leads showing higher R (other than MDB0273 in S67) for which we may decide to open the diode container, and to carry out local measurements … hopefully showing that the higher R is located at the diode to heat sink contact


Other considerations• The CSCM that will be done in 3 sectors before the end

of LS1 will hopefully confirm safe operation for these sectors. The other sectors may be done before LS2

• The two diode leads in S23 that showed resistance higher than 12 µΩ were localized in the same diode stack and on diode to heat sink contacts. This diode stack has been removed and will be retested in SM18 (RDIODE-HS = 22 and 25 µΩ) … we will learn from this test

• S23 is special, as it will show higher resistances due to CSCM already done in Feb. 2013. We know that all diode to HS resistances are changing during high current runs, and expect that the high resistances are localized there


Spare diode stacks• Contract with Prodtek AB (SE), now finished, after a series

of difficulties (due to other commitments of the contractor)• For sets of parts, heat sink, bus bars, flanges, helicoils, tie

rods, … for 30 dipole stacks, and 10 quadrupole stacks• Diodes (‘press packs’) are from DYNEX (UK); 270 units

were bought, and there are about 180 left, all Ni-coated• Assembly of the diode stacks @ CERN – MPE• A new specification will be edited for another IT for 60

sets of parts, to be defined how many dipole stacks and how many quadrupole stacks, and up to MSC to decide whether finished/tested stacks are needed, or whether sets of parts can be stored till the moment they are needed

Courtesy K. Dahlerup-Petersen


New production – What’s ‘new’?• On the 40 sets of parts delivered:

• Silver coating on all contact surfaces (4 µm thick),except that of the diode (press pack) that remainsNi-coated

• No EB welding nearby the half-moon, onlyone remaining

• Varnish applied on the G10 insulation partsto avoid absortion of humidity … and electricalfailure

• What will be changed for the new contract (the additional 60 sets):• Avoid machined insulation parts (G10), consider

instead molded parts (not porous)• Insulation U-shapes shall also be redesigned to

avoid cutting fibers in the angles during machining (otherwise cracking occurs) Courtesy K. Dahlerup-

Petersen


Test results – 4 dip. d. stacks - New

• SES37 / 2013-11-05



Test in SM18 – MDA0001Diode to heat sink:

Max 5 µΩ @ 10th 13 kA decay

Look at the scale!

Heat sink to bus bar:

Max 2 µΩ throughout the test



Test results – 4 quad. d. stacks - New

• SES38 / 2013-11-21



Test in SM18 – MQB2006Diode to heat sink:

Max 5 µΩ @ 10th 13 kA decay

Heat sink to bus bar:

Max 2 µΩ throughout the test

Look at the scale!



Status on warm / cold testing

When Stack ID NoteTotal # ‘spare’

Sept. 2013 MDB 2005, MDB 2006, MDB 2007, et MDB 2008

Certified 4 MDB

Sept. 2013 MQA 0423 Certified, and installed in a magnet -

Oct. 2013 MDA 2001, MDA 2002, and MDA 2003

Certified, and MDA 2001 installed in 3097 at BBI.A20R2

2 MDA

Nov. 2013 MQB 2006, MQB 2007, and MQB 2008

Certified 3 MQB

To come 2 MQA ready for cold testing

To come 4 MQB ready for cold testing

To come 8 MDA ready for cold testing, and 4 ready for warm testing

To come 4 MDB ready for cold testing, and 4 ready for warm testing


Conclusions• Consolidation of the unstable bolted assembly of the

connection plates in the quadrupoles progressing well (close to 50% done)• Difficulties with the procurement of the fasteners made of

Inconel now overcome• A few other minor NCR’s were discovered, and resolved

• Further investigations on-going for the dipole diode leads• COMSOL simulations were done to determine operation limits

(also for the quads)• Measurements @ cold, and @ warm, both in SM18, and in LHC• The dipole diode leads are safe. The new test results coming

soon, should confirm this• New production finished, assembly work finished with

MPE, and testing on-going at both MPE and MSC• New IT will be launched in early 2014 for 60 sets of parts


Courtesy G. Willering

Documents

Diodes qualification: Results and plans for LHC and spares MSC Technical Meeting on 2013-12-12 F. Savary, on behalf of the diode working group, with special