LHC Splice Repair

Let's see if I can summarize the situation with the LHC splices.

The LHC has three main power circuits: one for the dipole magnets ("RB" = bending) and two for the quadrupole magnets ("RQF" = focusing and "RQD" = defocusing) These are known as the 13 kA circuits, although this is somewhat of a misnomer. With the LHC at full power (7 TeV) they carry 11.8 kA and produce an 8.33 T field. However the design limit of the magnets is higher: 9.0 T, corresponding to a current of 12.8 kA.

The superconducting cables are Nb-Ti Rutherford cables, the joints between them soldered with high-temperature silver solder (96 percent tin, 4 percent silver). Normal joint resistance varies considerably with temperature, being about 10 μΩ warm, 300 pΩ cold. The cables are supported by a copper bus stabilizer with a cross-section of 280 mm2. Each joint is compressed between a 12 cm copper wedge and a 15.5 cm copper U-profile, together referred to as a joint stabilizer.

In the LHC there are a total of 24,000 splices - 10,000 interconnect splices and 14,000 magnet splices. If a quench happens (sudden loss of superconductivity) the current must temporarily pass through the surrounding copper. Magnet splices are protected by diodes, but interconnect splices are not, and the copper stabilizer must be designed to handle the full current for up to 100 sec.

The magnets have been commissioned to 5.5 TeV. The Sept 19, 2008 incident occurred during the last phase of the 5.5 TeV commissioning, and is believed to have been caused by a 200 nΩ interconnect, which is 1000 times too large.

Several types of problem have been identified:
- Anomalously high contact resistance between the superconducting cables (tens of nΩ at 1.9K).
- Lack of continuity between the superconductor and the stabilizer, or between parts of the stabilizer. This can be due to a joint which was poorly soldered or even missing solder.
The most dangerous situation is when a cable-stabilizer defect and a stabilizer-stablizer defect both occur at the same joint.

Cable-to-cable resistances at cold have been measured with excellent accuracy, and are not considered to be a problem. The worst joints are 2.7 nΩ for RB and 3.2 nΩ for RQ, and both of these are acceptable. The new Quench Protection System (nQPS) continuously monitors these resistances.

Measurement of resistances involving the copper stabilizer requires a warmup. The four sectors (S12, S34, S56 and S67) that were warm during 2009 have been thoroughly surveyed, segment by segment, using a "Biddle" hand-held nanovoltmeter (named for the manufacturer). One sector (S45) has been done at both 80 K and 300 K for comparison's sake, while the remaining three (S23, S78 and S81) have not been measured at all. Such measurements are sufficiently accurate for the RB circuit but not the RQ.

The term R16 refers to an individual joint resistance measurement at warm of the 16-cm long stabilizer. Only a sampling of these have been done. Typical R16 values are 12 μΩ for RB and 19 μΩ for RQ, but others have been found that range up to 60 μΩ. Also, some splices had acceptable R16 values but nevertheless upon visual inspection were found to be physically defective.

Complete repair during the "long shutdown" will involve a full warmup, then location and resoldering of each defective joint. It is expected that 15 to 20 percent of the joints will need repair. In view of the fact that even good joints can degrade over time due to electromagnetic and thermal cycling, the installation of an additional copper shunt on *every one* of the 10,000 interconnect splices will be required. This shunt will be designed to withstand the full current for 100 secs. So as not to affect the existing silver solder, it will be soldered in place with eutectic 60-40 tin-lead solder, and will include mechanical clamps that will hold the splice together in the event of a failure.

In the meantime, what can be done to further understand the problem? Without opening the individual joints, measurements can only be made at certain externally accessible voltage taps. Thus, resistance can only be measured for busbar segments spannning several joints: 2 or 3 joints for RB, and 8 joints for RQ. The nQPS system monitors these taps, and can be used to measure segment resistances at cold to within 1 nΩ.

There are several possibilities:

1) Warm up the 3 remaining sectors, measure the busbar resistances, open and repair the bad ones, and cool down again. (No comprehensive visual inspection or shunt installation.) This would serve to reduce the highest excess resistance from 90 μΩ to 60 μΩ, but only for the RB bus. (Knowledge of the RQ bus is not good enough.) Time required: more than 3 months, which is too long.

2) Measure the RRR at low current and partial warmup (15 - 35 K). Time required: 2 weeks. The residual resistivity ratio (RRR) is the ratio of the resistivity of a segment at warm vs just above the superconducting transition (9.6 K) The RRR of a factory-fresh cable is 70-100, but it increases when subjected to the soldering process. A well-soldered (and therefore well-heated) joint will have RRR > 150, but for a defective one the RRR may be less. Present operation assumes an RRR of 100. Certification of a higher RRR for the entire machine would mean an increased safety margin, or possibly allow operation at higher energy.

3) The "thermal amplifier" approach. Apply a 3 kA current pulse for 10 secs. This will selectively warm up the bad joints. Then use a low current in conjunction with the nQPS to identify the areas of increased resistance. This would find all splices > 50 μΩ. Time required: 1 month, short enough that this test could be scheduled during the 2011/2012 year-end stop.