Summary notes of the seventeenth meeting of the LHC Commissioning Working Group

 

Wednesday November 15th, 14:30

CCC conference room 874/1-011

Persons present

Minutes of the Previous Meeting and Matters Arising

There were no comments on the minutes of the 16th LHCCWG meeting.

Energy Tracking Between Sectors (Freddy Bordry)

This presentation was an update on Freddy’s talk on power converters at Chamonix 2001. The LHC powering is divided into 8 completely independent sectors: 8 converters power the main dipoles, while 8 each feed the QF and QD quadrupole circuits, giving a total of 24 independent families.

 

The power-converter tracking errors comprise both static part and dynamic components. Dynamic errors arise from timing errors and from lagging of the regulation loops.  The LHC timing is synchronized to within 1 ms, and the LHC regulation loops are designed without any lagging error. The I-to-B transfer function, in particular the relevant time constants, must be known from measurements and can then be corrected for.

 

A measurement campaign of the LHC power converters was performed at String 2. Without correction the measured B2/B1 tracking error was large (about 0.5%). With feed-forward the error was reduced, but it did still not meet the specified tolerance. With feed-back the curve fell within tolerance, and it translates into a maximum tune variation of about 3e-3, which could be further reduced by beam-based corrections, e.g., by a tune feed-back.

 

Concerning the tracking between the 8 main dipole converters, Freddy explained that, if we only look at the effect of converters, i.e. assuming a perfect response of the magnets, the initial global accuracy is 50 ppm. Calibration brings the accuracy down to 20 ppm. All ppm values quoted refer to errors which are normalized to the ultimate current. The 20-ppm accuracy corresponds to a magnetic field error of Delta/B~3.3e-4, or to an estimated peak orbit excursion of 0.7 mm.

 

After some discussion, the 0.7 mm was considered to be the worst case, irrespective of the number of sectors affected.The effect of the residual errors can be corrected with the pilot bunch. The remaining orbit error arises from the converter reproducibility. Following a past recommendation by Oliver, the converter reproducibility had been pushed down to 5 ppm, which corresponds to a 180-micron peak orbit shift.

 

Oliver commented that the effect of power-converter drift can be corrected during the run. One additional source of drift is the persistent-current decay. Stephane pointed out that the main tracking error may be corrected by feed-forward. The non-reproducibility is of order 1e-4, and, therefore, it is comparable to the reproducibility of the magnet itself (see the presentation by L. Bottura in the the 14th LHCCWG meeting).

 

Freddy presented a table summarizing the power converter tolerances for the LHC. Thijs asked whether the tolerances listed assume thermal compensation. Freddy answered that the tolerances given assume that the temperature variation stays within a specified range (+/-1.5 degrees; see Engineering Specifications “General parameters for equipment installed in the LHC”, LHC-PM-ES-0002.00).

 

Freddy next discussed that the tune change for a 20-ppm dipole-quadrupole tracking error is about 0.03. This can be corrected with the matching quadrupoles. The resulting beta beating is small. Stephane commented that the specified converter tolerances appear fairly large. He remarked that the tune error after 30 minutes could be 0.006 at injection. Ralph A. asked how these tracking errors would appear, and if, for example, they would occur during the ramp. Mike also inquired about the ramp. Freddy replied that the tracking error for the ramp is 5 ppm, i.e. there is no additional error contribution.

 

Jean-Pierre asked about the accuracy and tracking of other power converters. Freddy answered that converters for the inner triplets are similar, but that all the others are worse. Ralph A. next asked for the accuracy and reproducibility of the warm magnets. Freddy responded that the warm magnets belong to the “second class”. He elaborated that the converters are grouped into 4 classes. The main dipoles belong to the first class, for which the converter specifications are the most demanding.

 

=> Effect of tracking errors in warm magnets (Ralph A?)

 

Tracking Error Measurement and Correction (Jorg)

Jorg focused his presentation on the absolute calibration of tracking errors using beam measurements. He started with his conclusion that the only good answer to changes is feedback systems.

 

Magnet tracking errors can be determined form the associated orbit changes. The LSA suite is able to correct such errors. An ultimate limit to the correction is imposed by the differences between the two apertures. Jorg commented that we need to distinguish the magnet transfer functions for different sectors.

 

The energy error is estimated from the dispersion-weighted average beam position. If we correct the orbit to 1 mm, we can achieve a measurement resolution of 6 ppm.

 

Stephane remarked that in the LHC we may disentangle the energy error from systematic BPM errors more easily than in LEP, by using the combined readings of the QF and QD BPMs, which are located at positions with different values of dispersion. The systematic BPM offset in LEP was not remembered. Jorg recalled that in the SPS the systematic BPM offset is about 5 mm. Jean-Pierre suggested that a constant offset could be included in the BPM fit. Ralph S. added that the method for detecting energy errors proposed for LHC indeed uses all ring BPMs, including the QF and QD BPMs, and that an extended method also takes into account the systematic betatron oscillations superimposed on the dispersion orbit, as implemented for the orbit/energy feedback system.

 

Jorg cautioned that orbit correctors could cause a spurious bias. Possible cures are avoiding massive correction, comparing results for different corrector seeds, and checking the integrated corrector fields.  At LEP the bias per sector could be controlled at the level of a few 1e-5. 

 

Thijs commented that we may have 8 degrees of freedom. Therefore, we might limit the number of corrections to 8 or to multiples of 8. Jorg replied that, unfortunately, the misalignments will not have the 8-fold symmetry. For this reason, one cannot enforce the 8-fold symmetry in the correction.

 

Jorg now looked at different stages of the measurement. If the rms 1st turn orbit can be brought down to 2-3 mm, we could get a 20-ppm estimate of the MB tracking error on the 1st turn. Averaging of pilot beam data could improve this number, and may reach a few 1e-4 accuracy. With the 1st closed orbit, and assuming the rms orbit can be reduced to 1 mm, the tracking error may be determined with an accuracy of about 10 ppm.  The accuracy of orbit-based measurements at injection is comparable to the magnitude of the expected tracking error. Jorg pointed out that at 7 TeV we can only reach ~100 ppm with beam-based measurements.

 

The absolute beam energy in the LHC at injection can be determined from the orbit of the SPS beam at injection, and from the known SPS energy.

 

The MB-MQ tracking error is best measured with the tune. One can easily achieve an error of 1e-5 with a tune measurement of 0.001 resolution (e.g., PLL). The optimistically achievable measurement accuracy is 1 ppm. Jorg cautioned that other errors can contribute to shifts in the tune value.

 

Small quadrupole tracking errors between sectors cause only minor beta beating, but the phase advance between IPs can be important with regard to beam-beam effects. Determining the tracking errors between different MQ converters requires the measurement of the phase advance across each sector, either using orbit response matrices or turn-by-turn BPM readings processed by a harmonic analysis.

 

Jorg summarized the key points: (1) the relative energy errors between sectors can be determined at the 10 ppm level, with a 1 mm rms orbit error at injection; (2) the tracking errors between MQ and MB can be measured with extreme accuracy using PLL signals, but a limit is set by quadrupole calibration errors; (3) all effects can be corrected by LSA, and no problem is expected.

 

Stephane commented that the SSS correctors may have 10 ppm tracking errors. He recommended that the resulting error in terms of chromaticity be estimated.

 

=> ACTION: Effect of SSS tracking errors on chromaticity (Stephane) - DONE

 

After the meeting, Stephane followed up the effect on Q' due to PC-tracking errors of MCS (b3 spool) and MS (lattice

sextupoles). His conclusions are the following: 1) the MCS can correct about 3000 units of Q' when pushed to nominal field at injection; 2) the SF (SD) can correct about 7500 (4000) units of Q' when pushed to nominal field at injection. As a result, with 100-200 ppm long term stability of the PC's, the effect of Q' ranges between 0.15-0.3 (for the MCS assuming as a worst case a linear addition of errors in the 8 circuits) and 0.75-1.5 units (for the SF and the same worst case). The short term stability or the reproducibility (which is more relevant for us) of Q’ due to the MCS/SF/SD power converter tracking errors should then be in the shadow of other effects (they result in Q' changes of less or much less than 1 units, to be compared, e.g., with the hysteretic behavior of these magnets at low field or with a change of dQ'=100 units due to the decay & snapback of the b3 component in the MB's of which +/-20% may remain uncorrected during commissioning.

 

Remarking that everything looks consistent and the residual effects easily curable, Jean-Pierre asked why in LEP it took a very long time to fix the tracking errors between quadrupoles and dipoles. Paul added that in LEP also a “relaxation effect” had been observed, which was not fully understood. Freddy and Jorg commented that the time constant differences between main quads and dipoles were thought to be responsible for this relaxation. Helmut added that in LEP the unstable magnetization of the concrete used for the ‘cheap’ dipole yokes was a source of non-reproducibility.

 

Jorg observed that the field errors depend on the cycle. Stephane argued that the reproducibility of a s.c. dipole is of order 1e-4, and, possibly, a factor 10 better than that of a normal-conducting one. Replying to a question by Jean-Pierre, Oliver and Freddy stated that we can be confident in the specified performance of the power converters, since all numbers were confirmed by the String-2 measurements. 

 

Paul posed a question in connection with the separation dipoles D1 and D2, namely whether we expect significant c.o. distortions from ramping these dipoles, which form bumps consisting of 2 cold and 2 warm magnets, and the spectrometer system. Differences in reproducibility of warm and cold separation magnets, of order 1e-3 according to Stephane, translates into a non-closure of the bump. Opinions on this point were diverging.

 

=> Possible non-closure of nc/sc separation dipoles during ramp (Stephane) - DONE

 

After the meeting, Stephane clarified that, concerning the reproducibility or short term stability (e.g. with temperature) of the warm magnets, he indeed recalls a value of 1e-3, to be checked with warm magnet experts. If confirmed, at injection the effect is harmless.  In collision this translates into about 3 times the target fixed for the knowledge of the D1 transfer function (a change by 3 units of b1 in one single D1 separates the two beams by about 1 sigma for beta*=0.55m).

 

Ralph A. asked for simulations of the superposition of all these effects on orbit and tunes, and for studies demonstrating that we can correct them as advertised. Ralph S. replied that he had already simulated snapback and ramp (Ralph S.’s talk at Chamonix’06; see reserve slides), but drifts of current in power converters were not yet included. These drifts would add linearly to the results presented at the time. The larger (and fastest) effects are expected during the snap-back and during the squeeze.  He also noticed that the latter may have implications on how we will close the collimators during squeeze. Jean-Pierre informed us that all effects are included in the optical model prepared by Per Hagen. In particular, power converter tracking errors were found to have a negligible effect on the beta beating as compared with magnet errors and other error sources.

 

Paul asked what is known about the stability of the LHC injection septa magnets. Jan and Jorg responded that, at least, the stability of the SPS extraction septa was measured in the TI8 test. The beam trajectory stability was found to be good, that is, the beam moved by less than 200 micron, which is close to the measurement resolution. The residual motion is attributed to field changes of the SPS extraction septa.

 

=> ACTION: Injection septa reproducibility (?)

 

Following the meeting, Jan pointed out that the results of Jorg’s detailed stability analysis for the TI8 line and the MSE can be found on the TI8 pages at 

http://proj-lti.web.cern.ch/proj-lti/LTIcoordination/RelatedMeetings/TI8BeamTest/PublicationsPlanningProcedures/Results%20second%20test.htm

or, more explicitly, on slides 17 and 18 of 

http://proj-lti.web.cern.ch/proj-lti/LTIcoordination/RelatedMeetings/TI8BeamTest/PublicationsPlanningProcedures/results%20second%20test/JW-TI8-no3.pdf

 

Jorg added that indeed the septum stability was 'studied' with TI8 (2004) and written up in AB-Note-2006-021 OP. In the meantime he collected new data from CNGS that is not yet fully analysed, but which reveals a factor ~ 2 improvement w.r.t. 2004 (noise reduction in the BB4 building). For CNGS the RMS ripple of the MSE that dominates the transfer-line stability is ~ 1e-4 which corresponds to a kick of about 1 microrad (trajectory amplitude ~ 0.1 mm). The exact number is still to be confirmed.

 

Next Meeting

Wednesday November 29th, 14:30

CCC conference room 874/1-011

 

Provisional agenda

 

Minutes of previous meeting

Matters arising

Squeeze optics and power converter settings (Massimo/Stephane)
Squeeze strategy for the collimators (Ralph)
Squeeze mechanics and demo (Mike)

AOB

 

 

 Reported by Frank