MP3 meeting

Wednesday 15 Jul 2009, 09:00 → 10:30 Europe/Zurich

30/7-012 (CERN)

Jean Philipppe Tock (CERN)

Slides

• Discharge_Time-Constants_MP3.pdf
• Discharge_Time-Constants_MP3.pptx
• MP3_Verweij_2009-07-15.pptm

MP3 meeting 15 July 2009

Matters arising

Significant leaks were found in sectors 81 and 23, they must be repaired before the LHC can operate.
The issue will be discussed at today's LMC.
A partial or a global warmup are the possible options.
The mechanism of the leaks is not yet understood.
Answering to a question from Walter, Jean Philippe Tock said that the DFB flexibles were changed in sector 81.

New busbar resistance measurememts at 300K in sector 45 are available. According to Bob, the uncertainty on the quadrupole buses is reduced by properly taking into account the bus bar geometry. Also the temperature stability was improved.
Nuria asked what the uncertainty is now on the quadrupole busbar resistance
Bob answered 30 microOhm (RMS) from measurement at 300K.
Rob Wolf asked how much is the maximum busbar excess resistance left in the main dipole circuits
Bob said 49 microOhm
Three quadrupole outliers in the order of 60 microOhm are as well to be registered
Ruediger asked about the measurements at the extremities of the sector.
It turned out that there are 64 segments in the machine that are not measured
Marta Bajko asked how these results will be documented
Ruediger said that this will be presented at the next meeting of MP3 in one week

Minutes of the last meeting

There was a discussion on Adriaan Rijllart's comment to the last meeting, that firing the quench heaters as a last resort in case of splice runaway should be reconsidered. Adriaan explained his arguments, saying that the reliability of the system could be improved to avoid false triggers.
Ruediger said that Andrzej Siemko had the same opinion limited to the quadrupole busbars. He added that the dynamics of quench propagation to the neighboring dipoles and busbar joints must be taken into account
Karl Hubert Mess said that the reliability of the heaters is an important element, we have already lost 6 heaters in the dipoles (and none in the quadrupoles)
Arjan said that firing all the quads means firing all the main bends some minute later, which matters for the heater wear consideration.
Nuria said that this matter needs to be studied by a small team, she insisted that we must be sure that this is really the last resort
Karl Hubert said this move would only mitigate the damage.
Howard Pfeffer did not agree with that, he said that by virtue of Arjan calculations the faster time constant might actually prevent runaway of a bad joint
Action-> Nuria will study all the aspects of firing all quadrupoles

On the reduction of decay time constants (K. H. Mess)

Karl Hubert presented (see annex slides) the options and constraints to increase the dump resistances of the main bends and of the quads. Several constraints exist, with different levels of rigidity. For example, the present limit for the main bends is a small resistor (voltage divider) that could be changed rather easily, while other limits are given by the voltage ratings of the switches or by energy and temperature considerations and are more difficult to remove.
Presently it would seem that, irrespective of the joint resistaces left in the machine, the time constants at 7 kA could be reduced to 10 s for the quads and 50 s for the main bends. The recommendation of MP3 is not to limit the current in the circuit below 5TeV equivalent. The time constant of the quad circuits could be reduced at 12 seconds without any additional hardware modification. It could also be reduced to 10 seconds if a capacitor battery is installed in the extraction switches to decrease the voltage seen by them. The time constant for the dipole circuit could be further reduced to 55 s by limiting the current at 8kA. Going faster would limit as well the energy of the machine, be it perfect i.e. with only good joints.

On the snubber capacitor to suppress false triggers of the symmetric quench detection (K. D. Petersen)

Knud explained (see slides) the issue of false triggers of the symmetric quench protection as it was observed in the unit under test at the end of June. The switch opening is source of travelling waves that give voltage differences between the magnets. The system was not triggered during the test just because the current was not high enough. The time duration of the voltage spike was 40 ms.
Ruediger asked whether the discrimination time could be increased to filter out the spike
Knud said that this was a possibility
Rob Wolf asked whether it was established that the source of the problem was a travalling wave
Karl Hubert reminded that these waves were observed during hardware commissioning in 2008
Howard Pfeffer said that we must be careful as the detailed shape of the voltage pulse on the detector depends on details like the slight differences between magnets, and many different combinations are possible. We are basing ourselves on a single measurement. He said a next test will tell us more
Karl Hubert said that HERA had as well a double quench detection (for normal and symmetric quenches), and that the data of the 2-2 magnet detectors were scattered due to the differences in inductance between magnets; whereas the normal quench detectors were more robust, likely due to the effect of mutual inductance.
Nuria reminded that the symmetric quench detection is only needed above 7 kA
Ruediger said he was not sure that this was agreed by everybody
Walter pointed out that the snubber capacitor will not eliminate false triggers in the normal quench detectors coming from different losses in the two apertures
Karl Hubert said that another way to see the problem is to look at the impedance in the frequency domain: the snubber will fix the high frequency part, while the eddy currents at low frequency (10 Hz and less) will still be there.
Karl Hubert recalled as well that electrolytic capacitors must be regularly reformed by applying voltage
Walter ask whether these capacitor would fail in short
Knud answered yes, adding that in this respect two capacitors in series might be a good idea
Rob Wolf reminded that the issue of unbalance between apertures was extensively studied last year. We have a fairly good picture of these differences, which should reflect as well differences between magnets
Knud said that a software solution consisting in increasing the threshold when the switch is opened is going to be implemented anyway as a first stage.
It was clear from the discussion that the implementation of snubber capacitors is not exempt from difficulties and will require time and resources
Walter stressed again that the symmetric quench detection should not be needed at low energy
Rob cautioned that this system is the only one which covers the external splice
Also some issues with the UPS were mentioned
The symmetric quench detectors is the only system built with redundancy in the UPS system.
The original system is based in a single UPS.
Ruediger asked Knud what he needs from MP3
Knud answered he needed a decision on the time constant.
A long discussion followed at the end of which the following points seemed acquired:

1) The time constant, the maximum current, and the maximum joint resistance left in the machine are interrelated.

2) At the end of the process, the energy of the machine should be limited by the maximum joint resistance, not by the rating of the dumping system

3)The modifications to the dumping system should not be more costly in time and resources than warming up to repair the splices

4)The granularity of the building blocks for the new quadrupole dump resistors should be decided soon

5) It would seem that 5 mOhm blocks and time constants between 10 and 12 s are the right choice for the quadrupoles

RRR measurements (Arjan Veweij)

Arjan went through the motivations for mesuring the RRR of the busbars (reducing the uncertainty on the safe current calculations; reducing the uncertainy on the joint resistance measurements at 80 K, and modeling the quench propagation at high currents); he recalled the sources of the present knowledge of RRR in the machine, and made a proposal for the measurements.
Walter asked in what respect these measurements will be different from those performed some time ago in sector 1-2;
Arjan answered those were done at 25 K.
Walter asked what could be inferred on the local value of the RRR at the joints, from the technology applied to form the connections
Arjan said that the thermal cycle undergone during brazing was much shorter than that applied during busbar production, therefore it was unlikely to have changed the RRR.
However he said that nothing is known on the RRR of the copper wedge and of the U profile.
Ruediger asked whether it was not the RRR of the copper in the cable that mattered more
Arjan said they were both needed for the simulation
Mike Koratzinos said that the results in sector 12 ( average 117 , sigma 11) were in contrast with what he had estimated from 300K-80 K data in sector 45 where he found significantly higher values
Walter said that using data at 80 K to infer the RRR should be much less accurate
Mike claimed that the accuracy was sufficient
Mike expressed the wish to have more termometers equipped with the readout electronics, as temperature corrections are so important for the resistance measurements
Finally, the present agree that MP3 should recommand to use the worst RRR for the safe current definition
The RRR measurement campaign in sector 1-2 is considered useful. Based on the results, other measurements could be required.