Superconducting Circuits, a generic view What is special with superconducting circuits? What are the specifically dangerous issues? What can be tested at all and is it worth the effort? What are the required initial conditions? What has been tested and does not need a repetition of the test? Worst case scenarios. Can we completely rely on the tests? Reiner: What remains to be done during hardware commissioning?

The basic components: Consider a superconductor, already immersed in LHe:

Consider a superconductor, already immersed in LHe: The basic components: Consider a superconductor, already immersed in LHe: As such pretty useless, but the picture is incomplete, anyhow:

Consider a superconductor, already immersed in LHe: The basic components: Consider a superconductor, already immersed in LHe: We need: Current leads and all the warm parts We will have in addition: Inductance, resistance and capacitance

A single wire in details C R L C C R

A single wire in detail R C L C C Frequency dependence Stored magnetic energy R C R L C C R Stored electrical energy

Two Magnets

Critical Elements, other than the Superconductor

Critical Elements, other than the Superconductor Diodes to bypass the energy But the energy must be dumped! Damping resistors to deal with voltage transients

Symbolic Circuit

Symbolic Circuit

Current Lead

Breakdown at points with high voltage

Breakdown at points with high voltage Typically at the circuits extremities and at the voltage taps or feedthroughs, wherever the gas can have low density

Symbolic Circuit Can quench, Has energy stored

Quench - What Went Wrong? Abnormal voltage signals recorded during the provoked quench Courtesy: A. Siemko

What can happen here? In case of a quench: The energy in the magnet is high enough to destroy it. The energy must be spread quickly -> Heater The energy of the other magnets must be guided around -> Diode

Symbolic Circuit

What can happen here? In case of a quench: The energy in the magnet is high enough to destroy it. The energy must be spread quickly Heater The energy of the other magnets must be guided around Diode The time constant is very, very long We have to dump the energy To be done with great care, because we have to open a switch! Time constant is still large (in particular for the dipoles). Be aware of the transmission line effects during switch opening.

Inventory Current Leads 13 kA 6 kA 600 A 120 A in DFB 120 A in magnet Busbars Big busbars Small busbars Difficult, because CL need a working cooling environment to run current. To establish this the load parameters have to varied, which in turn requires various currents through a working magnet circuit. To be discussed. Form part of the circuit, but tested only globally.

Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits Inventory Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits 60 A circuits

“Easy”, Freddy takes care. Inventory Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits 60 A circuits “Easy”, Freddy takes care. The 60 A circuits and most 120 A circuits ( including the current leads and bus bars) are protected by the overvoltage detection of the powerconverter. Its AB-PO.

The 120 A MO and the 600 A circuits have a “global quench protection” Inventory Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits 60 A circuits The 120 A MO and the 600 A circuits have a “global quench protection”

Global Quench Protection Δ V Δ V L dI/dt 24 bit ADC DSP Fieldbus Interlock

Done at individual system test Done at the time of HC: Electronics installed Electronics tested Electrical connection tested Fieldbus tested Generation of interlock signal tested LdI/dt generation simulated Heating of the CL installed and tested To be done Establish cooling conditions Establish interlock Test with small current: -energy extraction, -current lead cooling, dI/dt compensation Interlock reaction Increase current and repeat In case of fast ramp down or quench: Study voltages carefully

Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits Inventory Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits 60 A circuits

Problems to be expected 6 kA quadrupoles ΔU ΔU Long voltage tap, Problems to be expected

Done at individual system test Done at the time of HC: Electronics installed Electronics tested Electrical connection tested Fieldbus tested Generation of interlock signal tested Heater measured Heating of the CL installed and tested To be done Establish cooling conditions Establish interlock Test with small current: -energy extraction, -current lead cooling, Voltage measurement Heater firing Interlock reaction Increase current and repeat In case of fast ramp down or quench: Study voltages carefully

Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits Inventory Magnets 13 kA circuits 6 kA circuits 600 A circuits 120 A circuits 60 A circuits

13 kA busbar protection Courtesy R. Denz

Local quench detector for main magnets Courtesy R. Denz

Done at individual system test To be done Establish cooling conditions Establish interlock Test with small current: -energy extraction, -current lead cooling, Voltage measurements Heater test Interlock reaction Increase current and repeat In case of fast ramp down or quench: Study voltages carefully Selective heater test, very touchy!! Done at the time of HC: Electronics installed Electronics tested Electrical connection tested Fieldbus tested Generation of interlock signal tested Heater installed, measured Heating of the CL installed and tested

Worst case scenarios 1 What can go wrong? Missed quench can result in: Overheating Melting Pollution Overvoltage Can destroy large fractions of a sector Quench avalanche Backward Voltage We are told that the theoretical probability for a missed quench is small (~once per lifetime of the LHC?)

Worst case scenarios 2 The system is designed failsafe. A single fault should not be dangerous. <= to be seen. Double (maybe correlated) failures: Assume: The UPS fails During the fast ramp down, a quench happens. No detection, no heating. Maybe, a switch can not open, but can not tell it. No Post Mortem. The result will be severe damage and confusion.

Worst case scenario 3 Assume a splice (interconnection) breaks. Has been tested successfully but the repetitive forces lead to fatigue. The fluctuation in the arc-voltage and arc-current lead to overvoltage at vulnerable positions. May lead to loss of voltage taps May lead to destruction of a diode Maybe even winding short with quench and destruction of the coil.

Worst case scenario 4 Assume a high contact resistance in a diode connection. Together with a switch open failure and a break of the direct heater signal. Bypass busbar will overheat. Resistance will even grow, the diode may be overheated, ….. …see above The very worst: Negligence and Sabotage To put up an exhaustive list is inconceivable (for me). We have to keep our eyes wide open. And we have to train/tell people in the field continuously. This requires a certain information bandwidth. Given the limited amount of experienced people, which would act as information node, the number of simultaneous fronts is limited.

Summary I What is special with superconducting circuits? Large inductance, large stored energy, low resistance, long time constants, extremely high current density What are the specifically dangerous issues? Shorts, opening connections, high voltage, high energy density, hydraulic problems What can be tested at all and is it worth the effort? Functioning of the safety systems (at that time and against simple failure scenarios) It is worth it, because the existence of the Lab can be at stake.

Summary II What are the required initial conditions? Finished installation, cold machine, electrically OK Electronics tested as far as possible. What has been tested and does not need a repetition of the test? Except for trivial things, which will be tested in the shadow anyhow, everything is new or could have been altered since the last test.

Summary III Worst case scenarios Single faults are supposed to do no harm Combined faults can happen due to noise, interference, network overload or failure, power supplies… you name it. Whenever a quench is not detected: we have a problem Whenever the switches do not work properly: we have a problem Whenever the signal distribution does not work: we have a problem Whenever the readout does not work: we are blind There are certainly more reasons, why it may not work properly. Not in all, but in many cases, a careful test will tell us in due time about problems. To make use of this: we need a careful analysis of each test, “successful” or faulty, to exclude mistakes and faults. This requires patience, experience, communication and above all: time.

Summary IV Can we completely rely on the tests? No, never. All systems could fail at any time. Also, the beam can produce a quench with a voltage distribution in the coil, which can not be tested without beam. We need a good hardware, a good software and experience to make the failure probability small. Experience can only be gained, slowly. The reliability of the tests depends on the quality of the tests. Quality does not come for free.