Wednesday
February 14th, 14:30
CCC
conference room 874/1-011
Mike, who chaired, called the meeting to order. There were no comments on the minutes of the 19th LHCCWG meeting. However, Mike briefly recalled a few of the lively discussions and actions which were triggered by that meeting, in particular: (1) Jean-Jacques is setting up a BI-LSA study group which will specify the information to be provided by each instrument in different stages of LHC commissioning; (2) The transient orbit tolerances proposed by Stephane were finally agreed upon; (3) Rhodri confirmed that if survey offsets were taken into account a 4 mm peak closed orbit tolerance should be OK.
Massimo reported a problem which came up at the MEB, concerning quadrupole Q11R2 in SSS523 of Sector 2-3. As later clarified by Luca, this magnet showed anomalies in the voltage traces during fast discharge following a quench, which was later traced back to a short. As a consequence the corresponding magnet circuit cannot be excited above 2000 A (translating to a beam energy of roughly 1 TeV). The Q11R2 is a quadrupole in the dispersion suppressor, powered in series with the arc quadrupoles. No spare for this magnet is available. The installation of Q11R2 is foreseen for May this year. In the most optimistic scenario, a spare could be ready in May, but it would still need to be tested and a beam screen to be installed inside this magnet. Karl-Hubert Mess stated at the MEB that this magnet should not be subjected to any quench, so as to avoid a catastrophic damage not only to this magnet but to all magnets powered in series on the same circuit. In view of the critical non-conformity the magnet is not suitable for the ring, and, therefore, the MEB will not approve this special SSS. It is expected that the Q11R2 will be installed despite of the MEB rejection, since there is no alternative. The hardware commissioning of this sector would then proceed with special precautions, and the magnet would be replaced at a later time when the spare quadrupole has become available for installation. The swap of the quadrupole against the newly produced spare could happen before the 2007 450-GeV run. A final decision is still pending. For hardware commissioning and the evaluation of the risk associated with a 450 GeV run, the MEB proposes a close follow-up by the Machine Protection Project. Luca is preparing an email to Lyn, Philippe, and Roberto. Walter and Massimo commented that the cold mass of the spare magnet MQ003 is presently being tested in Block4 in view of being installed in the spare special SSS. Answering to a question, Massimo iterated that this magnet must not be quenched with beam under any circumstances.
Werner first illustrated the present nominal LHC filling scheme both as a formula and in picture form. This nominal filling scheme comprises 39 batches of 72 bunches each. It contains gaps of various size and in different numbers, which are needed to accommodate the rise times of the SPS injection kicker (8 missing bunches), the LHC injection kicker (38 or 39 bunches missing), and the LHC abort kicker (119 bunches missing, which amounts to a 3-microsecond gap).
He then explained why we look for alternatives, the main reason being that in 2006 with 72 bunches an instability was observed in the PS. The source of this instability is probably understood, but the problem is not yet completely solved.
A number of requirements must be fulfilled for any bunch filling scheme, such as leaving space for the various injection and abort gaps, approximating a four-fold symmetry, staying below the SPS intensity limit of 4e13 protons for all LHC filling cycles, maintaining a final injection of longest duration (Brennan and Jan explained that this condition is related to the synchronization of the beam dump and represents a safety feature), and maximizing the total number of bunches in order to optimize the luminosity.
Massimiliano asked for the reason of the SPS intensity limit. After the meeting, Elias explained that this limit is stated in the LHC Design Report Volume 3, Chapter 2, and that it seems to be either due to SPS beam loading and RF power constraints, or due to capture losses in the LHC and SPS electron cloud, according to different sources, namely Gianluigi and Elena.
Werner presented an alternative filling scheme with 48 bunches per batch. He pointed out that transferring 48 bunches from the PS to the SPS was never a problem. In the proposed scheme, the total number of bunches decreases to 2592 bunches, which is about 8% less than nominal. The abort gap increases to almost 4 microseconds. The same number of SPS supercycles (12) is required to fill one LHC ring, but now for accelerating 54 batches of 48 bunches. The maximum intensity per LHC injection is about 17% lower than for the nominal scheme. The abort gap in the LHC is significantly larger. Werner commented on the implications of this filling scheme for the LHC performance. At IP1 and 5 there will be 8% fewer collisions, in IP2 9%, and in IP8 11%. The larger reduction on IP2 and 8 is an effect of the increased abort gap). He asked whether this reduction is relevant, given that we expect bunch-to-bunch intensity fluctuation of +/-10%, pointing out that the luminosity can be recovered by a 4% rise in the bunch intensity. Concerning the beam-beam and PACMAN numerology, the number of bunches missing one or two head-on collisions increases by about 50%. The quality of the regular PACMAN bunches (bunches missing one or several long-range collisions) stays the same.
Responding to a question by Frank, Werner added that the fraction of nominal bunches decreases from about 45% in the nominal scheme to 35% in the proposed alternative.
Werner next explained a new filling scheme without crossing angle which derives from the above proposal by keeping only the first bunch in each PS batch, and which provides a completely safe bunch distance of 600 ns. The total number of bunches is 54 bunches, larger than for the nominal 43-bunch scheme. A difference to the latter is that these 54 bunches are not equidistant. In order to provide collisions in LHCb two options exist, namely either to displace bunches in one beam only, or to shift bunches simultaneously in both beams. In the latter case, the full number of collisions can be kept in IP1 and IP5, but about twice the number of collisions are lost at IP2 (10 instead of 5).
Werner concluded that the proposed 48-bunch scheme constitutes a good alternative to the nominal filling scheme, in case the production of 72 bunches per batch encounters problems in the PS. It provides almost the same luminosity, has several positive implications for the PS and SPS (addressed in the following talk by Elias) and represents a possibly interesting intermediate step towards the nominal scheme.
Massimiliano commented that at the luminosity workshop in January a request has been made by some of the experiments to provide a non-colliding bunch for background studies at least in the beginning of LHC commissioning. After a clarification that “non-colliding” here means not colliding at all, i.e., not colliding at a longitudinally displaced distance either, Werner replied that such bunches can easily be provided in the filling schemes with few bunches, but that a detailed calculation would have to be done to produce a definite example.
Paul commented that he did not see any advantage of the 54 bunch scheme as compared with a symmetric 43 bunch filling. Mike remarked that the design luminosity in the alternative filling scheme might drop by 10%, and would then fluctuate around 9e33 /cm^2/s instead of 1e34 /cm^2/s. Paul recalled that the flatness of the injection kicker limits the maximum number of batches which can be simultaneously injected from the SPS to 4 nominal batches. It was confirmed that the new scheme also complies with this constraint. Frank commented that due to the abort gap none of the filling schemes features a true four-fold symmetry. Werner referred to the symmetry considerations discussed in several of his papers (see, e.g., Consequences of Periodicity and Symmetry for the Beam-Beam Effects in the LHC, CERN LHC Project Report 49 (1996)).
Elias explained the reason for developing the alternative scheme and its implications for the injectors (PS booster, PS, and SPS). He also reviewed another different filling scheme which had been proposed earlier by Michael Benedikt at the LTC 09/03/05 for reaching the ultimate LHC beam intensity. The new alternative filling scheme uses 48-bunch batches which arrive in 2.4-s PS cycles instead of the nominal batches with 72 bunches requiring 3.6-s PS cycles.
Elias first presented some example results with the nominal LHC beam in the SPS from 2004 and November 2006. In the latter case, 3.6e13 protons were accelerated by the SPS, which is in excess of the required 3.3e13 protons for the nominal LHC. He cautioned, however, that the transverse and longitudinal emittances etc. were not completely characterized. These examples nevertheless indicate that the PS and SPS can provide the nominal beam. However, throughout most of 2006 a horizontal instability was observed in the PS when operating with 72 bunches. The instability did not occur with 48 bunches. A new scheme with 6 times 48 (instead of 4 times 72) bunches was adopted for the 2006 collimator tests in the SPS in order to avoid the instability. This scheme was later refined by Werner and Tatiana in regard to LHC beam-beam consideration.
Michael had presented a different scheme at the LTC 09/03/05. Michael’s scheme was developed in order to produce the ultimate beam via batch compression in the PS, using 9 harmonic numbers instead of the present 4. It featured trains of 42 bunches accelerated by 3.6-s PS cycles. In all these aspects, it was rather different from the new proposal.
Implementing the new scheme in the PS booster is easy. No modification is needed. The number of LHC “users” is reduced from 2 to 1, which would facilitate maintenance.
Elias next reviewed the generation of the nominal LHC beam in the PS. First 4 bunches are injected from the PSB, and on the next cycle another 2 bunches (“PS double-batch injection”). The total number of 6 bunches are then processed with triple splitting at 1.4 GeV, which is followed by two quadruple splittings at 25 GeV. Elias showed that the “polished” LHC beam in the PS had the same quality in 2004 and 2006, for a similar intensity of about 1.3e11 protons per bunch. Elias compared the beam losses observed for a polished and for an unpolished LHC beam.
Responding to a question by Frank, Massimiliano and Paul suggested that the LHC beam in the PS will always be polished prior to sending it to the LHC. Elias pointed out that this means specialists will be needed, if for instance in the middle of the night the LHC wants a new fill. In this regard the new scheme may be more robust.
In addition to the need of polishing, other problems inflicting PS operation are the activation of the injection area. Space-charge driven trapping phenomena on the injection flat-bottom contribute to this activation. There also is a need to stabilize the LHC PS beam by linear coupling, which would no longer be necessary without the double batch injection. A related issue is the instability of the PS magnetic field, which can drift by several Gauss before the second batch is injected, resulting in longitudinal injection oscillations. At the APC 22/09/06 Steve Hancock highlighted that the non-reproducibility of the PS magnetic field leads to shot-to-shot and day-to-day variation.
The main motivation for the new filling scheme is the high energy instability in the PS. Beam screens in the TT2 line revealed this instability in July 2006. The source of the instability was ultimately traced back to a problem with one of the 40 MHz cavities. When using this particular cavity, the beam had always been unstable, with another nominally identical cavity always stable. The problem was explained by a wrong voltage calibration. After recalibrating, both cavities now deliver the same voltage. Even so, there still remains an instability threshold for full bunch lengths of 11.5 ns or smaller prior to bunch rotation, most probably due to electron cloud.
Another potential problem concerns the figure of 8 loop, which is used to control the PS tune and chromaticity. The maximum rms current per cycle for this 8 loop is 560 A. For the “TSTLHC” beam cycle alone, the rms current was 680 A, but the rms current was much smaller when computed over the full supercycle. Obeying this limit will not be a problem for any of the LHC filling schemes.
Elias now turned to the SPS. The rise time of the resistive-wall instability with the nominal 72-bunch scheme is 97 turns. With 5 batches of 48 bunches it increases to 110 turns. The small difference will hardly be noticed. Electron cloud build up simulations for both nominal and alternative schemes were performed by G. Rumolo. With a maximum secondary emission yield of delta_max=1.3 the electron cloud saturates at about the same density level in both cases. With a slightly smaller value of the maximum secondary yield of delta_max=1.25, a clear improvement is found for the 48-bunch scheme. Therefore, this alternative scheme will be marginally better in view of electron cloud. Elias presented comments on the longitudinal plane by Elena S. according to which, with the present “long” acceleration ramp, there should be no problem from the RF side. The RF transient will be slightly increased due to the larger number of gaps. The larger spacing between PS batches (9 missing bunches compared with 8 bunches in the nominal scheme) are an advantage, since the rise time of the SPS injection kickers is at the limit. The reduced intensity for each SPS extraction is also beneficial, in regard to machine protection.
Elias now compared the LHC filling times. The nominal scheme requires 8 minutes and 38 seconds. The ultimate beam scheme by Michael increases the filling time by 33%. The new alternative filling scheme on the other hand reduces the filling time by 5.5%. An MD performed on 09/11/2006 established 5 and 6 injections of 48 bunches with about nominal LHC bunch intensity.
Elias concluded that the new alternative scheme is much more robust through the entire injector chain. Only 4% more intensity per bunch would be sufficient to compensate for the loss of luminosity, which is a small change compared with the 10% intensity fluctuations. In addition, if the new scheme proves indeed more robust, even without increasing the intensity per bunch the integrated luminosity at the end of the year could be higher than with the nominal scheme.
Mike commented that he could see only advantages in the proposed alternative scheme. Paul pointed out that most of these advantages are only realized as we approach nominal intensity. He stressed that at lower intensity no problem is expected with the nominal scheme. Elias replied that the effects of the magnetic field drift on the PS flat bottom are independent of intensity. Paul recommended that a full analysis and experiments be done for lower intensity. Increasing the individual bunch intensity in the LHC to1.2e11 could be challenging. He recalled that the nominal LHC bunch intensity has already been increased several times over the last years. He stressed that the most robust scheme should be identified for delivering the intensity needed in the first years. Paul plans to produce commissioning beams, rather than pushing the limits. The goal is to render pilot bunches and moderate intensity beams really operational. After the meeting, Elias stressed that the new scheme is (mainly) proposed as an alternative to the NOMINAL scheme, i.e. to accumulate luminosity over the years.
=>
ACTION: Identify most robust filling scheme for first years of operation (Paul
et al)
Mike explained that the CO timing section is in the implementation stage and that concerned parties should make sure that all their requirements are being met properly. He then described the nominal scheme which foresees injecting pilot bunches in either ring, intermediate bunches, dumping and re-injecting pilots, and finally injecting nominal beams interleaved between the two rings. Mike explained the distinction between machine modes and states. Each mode contains a number of states. The mode and state information is broadcast. More specifically, in each mode we execute a sequence of states necessary to start with next mode. Example states in the mode “injecting pilot” are “ready for pilot”, “injecting pilot”, or “circulating pilot”.
Oliver asked which type of BPM data are taken by default. Jean-Jacques replied that the closed orbit data will be sent at 10 Hz. Simultaneously also turn-by-turn data are provided.
For each state, the beam presence flag (BPF) and the safe beam flag (SBF) are set to specific values (true or false). Their setting will be monitored by the sequencer. The mode “injecting nominal” includes injection quality checks (IQC) like abort gap, beam loss, lifetime, energy mismatch, injection oscillations, trajectory in lines, mountain range, beam sizes, as described in Verena’s presentation at the Post-Mortem workshop. Certain events inhibit injection or hold the sequence.
Paul asked whether the first time we get a pilot bunch we will bring it all the way down to check the whole channel or terminate it on an extraction beam dump (TED) in TI2 or TI8. Moving he TED takes about 1.5 minutes.
Mike presented Ralph S.’s list of the triggered data acquisition at injection, comprising beam-based information and hardware status information. Paul asked how long the data persist and whether the timing includes quality checks. Jean-Jacques responded that LSA will do the checks upstairs. A 1-s data persistency in the tunnel as foreseen is sufficient.
Next Mike flashed a list of LHC timing events at injection which are provided by Julian Lewis and friends including Jean-Jacques. He then discussed in greater detail a number of injection events, such as injection forewarning, warning, or start function, a variation of which will be played at each injection. “Telegrams” describing ring, beam type, and the next bucket will also be sent out. The present proposal is that the states inside each mode are part of the telegram. Mike briefly illustrated the role of the Central Beam and Cycle Manager (CBCM) API for injection requests, and commented on timing issues for the RF system.
Responding to a question, Paul explained that the modes “adjust”, “unstable”, and “stable” were introduced in order that the experiments do not dump the beam when the machine performs some tuning.
Oliver urged to save snapshots of all settings. He asked whether such snapshots are stored automatically and who decides which settings to use on the next cycle. Mike replied that the LSA team is implementing this at the moment. The acquisition is event triggered, and LSA is taking care of all settings. He agreed that in addition having an explicit copy is a good idea. The offline run configuration will be determined by the EIC. He stressed that with the present architecture we can go back to any point in time. Ralph S. commented that this statement only applies to systems which are not controlled by feedback. Oliver suggested defining a recipe for the back up configuration, e.g., first set up, second back up etc. Verena clarified that comments can be entered when saving a configuration. In reply to a question by Brennan, the intermediate beam was defined as referring to 12 nominal bunches, which roughly corresponds to the safe beam limit.
=>
ACTION: Define recipe for back-up configuration (Coordinators? All?)
Mike
concluded that the principal LHC timing requirements have been agreed upon, the
interface to the timing system has been defined, its testing should start soon,
a loose FSM (finite state machine) approach is being followed for creating the
LHC sequence, and a first version of LHC “beam modes” has been established.
Magali was talking on behalf of several people: herself, Reyes, Verena, Laurette, Stefano, and Walter. Her talk had two parts. First she issued some general comments, valid for all phases, where she explained the purpose of the procedures, their basis, and the input required from the LHCCWG. The second part addressed the specific procedures developed for phase A1, and it included comments, example web pages, and open questions for this phase. This was a first trial presentation. In the future, one phase will be presented per meeting, or perhaps at every second meeting, explaining the layout of the procedures, main steps and questions, exemplified by the web pages. The presentations cannot go into great detail and they represent work in progress.
Magali next addressed the purpose of the procedures, which is to make sure that everything required is available and thought of, to make sure responsible people are informed, to point out problems, missing functionalities etc., and to develop an ordered list of tasks and results. The procedures do not provide a complete manual, since experts are expected to perform many of the tasks. The procedures are based on LHCCWG presentations, contributions at Chamonix workshops, Mike’ web pages, discussions with experts and current understanding. Magali encouraged all LHCCWG members to read and check the procedures, and then to give feedback, concerning procedures, entry conditions, exit conditions, problems and questions.
=> ACTION:
Send comments on procedures to Magali, Verena and friends (All)
She now turned to Phase A.1, which refers to the first turn of a pilot bunch at 450 GeV with the goal to establish injection, to thread around the ring, and to commission vital beam instrumentation. The procedures for this phase can be found at http://lhccwg.web.cern.ch/lhccwg/Procedures/stageA/phaseA1/index.htm. This web site features the sub-sites “overview”, “description”, “entry conditions”, “procedure”, “exit conditions”, “problems” and “questions”. Responding to a question by Oliver, Magali explained the color code employed (blue: key part of procedure, magenta: optional, red: questions, black: the rest)
Magali added that one of the entry conditions contains a link to the LHC machine checkout web page from Mike. She emphasized that these procedures are still incomplete: A table listing all steps of the procedures versus all required applications and instrumentation will be added. The procedures for phase A.1 will be discussed in the Injection Working Group led by Brennan. No time estimate for the individual steps is yet included.
Frank remarked that the web pages contain an impressive amount of detail. Paul and Brennan commented that all screens must be taken out at some point, and that this should be part of the exit condition.
Lastly, Magali presented a list of remaining questions, including the need to recycle, the availability or request of SDA (FNAL acronym meaning either shot data analysis or sequenced data acquisition), interleaving of the two rings, magnet status, display of spare channels for BLM cards (required for mobile BLMs), remote change of BLM thresholds, the coarse calibration of BLMs, and the transition from the first turn (inject & dump mode) to circulating beam in view of the screen positions (BTVs).
Oliver asked whether it has been decided to change the LHC main dipoles for the SPS-LHC energy matching. The answer was no – i.e. no decision was taken. Oliver next commented that we must be in the position to recycle if needed. This should be true for all procedures. Ralph S. suggested that many acquisitions and measurements are much faster done with circulating beam than with single-turn acquisition. Stefano replied that these measurements have been considered only for the case that we are stuck during a night shift and cannot get the beam around. In such a case it might be convenient to have a list of optional steps to take.
Replying
to a question by Ezio, Frank remarked that the energy
matching tolerance is 1e-4, or 1 unit. Ezio asked for
a clarification of the comment about which magnets are required. Magali explained that there are many circuits and that
higher-order multipole correctors, e.g., skew sextupoles, octupoles, or decapole, could be switched off for the first turn
commissioning. Walter explained that all magnets will be commissioned and
operational, but that the states in which the magnets should be need to be
defined. Mike remarked that the circuits required during commissioning were detailed
by Massimo at “
http://indico.cern.ch/contributionDisplay.py?contribId=71&sessionId=0&confId=86
).
Jean-Jacques will organize several meetings on software interfaces for the various instruments. He will send out an email to all parties involved including the list of authors for the procedures. Verena stressed that feedback to the commissioning procedures is needed. Jean-Jacques asked who will provide the details, e.g., on the coarse BPM & BLM calibration, reminding that BI will define the commissioning procedures for the instruments. Verena and Magali replied that the BI plans will of course be included in the web procedures if provided.
=>
ACTION: Provide BI commissioning plans and include them in web procedures
(Jean-Jacques, Verena, Magali
et al)
Due
to lack of time, the presentation by Thijs was
postponed to a later meeting.
Wednesday February 28th, 14:30
CCC conference room 874/1-011
Provisional agenda
Minutes of previous meeting
Matters arising
Magnetic model meeting:
- FiDeL
(Luca)
- WISE
(Ezio)
- LSA
(Mike)
Behaviour of the magnets through the squeeze
(Walter)
Documentation and procedures
AOB