Wednesday
November 29th, 14:30
CCC
conference room 874/1-011
Persons
present
There were no comments on the minutes of the 17th LHCCWG meeting.
Roger announced the launch of a sub-working group on the commissioning of the machine protection system, led by Jan, which will report to the LHCCWG. The LHCCWG web site already contains a link to the new subgroup.
Rudiger and Roger reported results from the pressure test performed during last weekend, which was fully successful for the arcs, but revealed a leak in the inner triplet. The problem is suspected to be localized in the interconnect, and in-situ repair appears possible.
Massimo presented the squeeze optics and pertinent power converter settings. Four questions were addressed by him in particular, namely (1) can we ramp with a constant optics and crossing scheme?, (2) the behavior of the optical parameters during the squeeze of IR1 and IR5, (3) the number of matched optics, and (4) the status of the squeezed optics for IR2 and IR8. Massimo recalled that Stephane had presented the squeeze to the LTC already on 31/03/04 (not available on the web), albeit without crossing scheme at the time. The crossing scheme was discussed, also by Stephane, at the LOC meeting of 11.10.2005.
In IR1 and 5, the injection optics can be ramped to 7 TeV without any limitation. For the present optics with beta*=11 m and 200 microrad crossing angle with 2 mm separation, the MCBY correctors at Q4 used in the separation scheme reach 107% of their nominal strength at 7 TeV, which is a value that these magnets can achieve. The conclusion is that the injection crossing scheme can be ramped. Interestingly, for the previous injection optics with 17 m beta*, the ramp at a constant optics was not possible, but a change at about 5 TeV was required.
Oliver commented on the impact of the crossing scheme. He suggested that the separation scheme could be relaxed on the ramp, in order to avoid exceeding the nominal magnet current levels. Jorg replied that in this case the reference orbit would change on the ramp, complicating the orbit feedback. Walter mentioned that the correctors in question may go to up to 120% of the nominal value. Everybody agreed that there was no problem. Jean-Pierre pointed out that for beam dynamics purposes the crossing scheme needs to be held constant only in units of beam size and not in absolute units of mm.
=> ACTION:
Orbit feedback behavior during squeeze?
(Ralph S., Jorg, Massimo)
For IR2 and IR8, with the injection optics ramped to 7 TeV the strength of the triplets reaches 220 T/m, in excess of the nominal 205 T/m. Massimo reassured the team that also this is not a problem, quoting Ranko Ostojic, since the nominal limit contains a margin for high luminosity operation. The IR2 crossing scheme cannot be ramped to 7 TeV due to strength limits of the orbit correctors. There are enough degrees of freedom left, however, which can be used to overcome this difficulty.
Massimo now addressed the status of squeeze for IR1 and 5. A set of 12 intermediate matched optics with different values of beta* are available (the optics solutions with crossing scheme during squeeze were discussed by S. Fartoukh, LOC meeting 11.10.2005). Three criteria are imposed on the crossing scheme during the squeeze (constant parallel separation, scaling of crossing angle, scaling of IP position shift). Massimo next presented some results from S. Fartoukh illustrating the variation of the quadrupole strengths with beta*. All the quadrupole excitations vary smoothly with beta*, except for a jump for Q7 of beam 2. The trim quadrupole curves are a little more erratic for beam 2, and for them zero crossings cannot be avoided. A further plot presented by Massimo shows the beam-beam separation in units of sigma for various values of beta*.
New simulations reveal how the optical functions vary during the squeeze. No errors are included in these studies. The collision tunes are used. During a squeeze of IR1 only, for beam 1 the maximum value of the parabolic tune excursions between two matched optics points is about 1.5e-3. The tune split hardly changes during the squeeze, which is good. The chromaticity does not appreciably vary either. The tune excursions for beam 2 are almost two times larger than for beam 1, up to about 2.5e-3. One remedy reducing the residual tune excursion would be adding further intermediate points.
Oliver remarked that, since the squeeze is performed without head-on collisions, tune excursions of a few 1e-3 should not be harmful.
The simulations indicate for beam 1 an upper bound of 1% beta beating, and a maximum closed orbit change of 0.1 sigma during the squeeze. The values for beam 2 are much larger, with up to 10% maximum intermediate beta beating when squeezing beta* down to values below 1.5 m. Negligible excursions are found for the beam-beam separation and the crossing angle. The dependence on the number of points was studied in complementary simulations, where Massimo removed two intermediate optics solutions, corresponding to a beta* of 4 m and 2.5 m. A direct squeeze down to 1.1 m without any intermediate steps would yield a tune excursion of 2.5e-2, and a chromaticity change of 5 units.
It was remarked that the small orbit motion for beam 1 is obtained only if the initial orbit is perfectly threaded in IR5/IR1. Jean-Pierre pointed out that this assumption might be too strong. Ralph S. agreed, adding that assuming random quadrupole misalignment of only 10 micron would already yield a maximum orbit transient of 300 to 600 um (several sigma).
Ralph A. asked for the motivation of removing points and whether the control system cannot handle as many as 12. Massimo and Mike responded that indeed the control system does not limit the number of points, and Massimo’s study rather addressed the sensitivity and possible simplification. Roger agreed with Ralph that we should take a large number of points and aim for smooth curves.
Oliver tentatively summarized Massimo’s results and the subsequent discussion as follows: for squeezing down to 3 m the situation is not critical, but for smaller beta* the number of steps matters. Ralph A. then raised the question whether or not we need to actually stop at the intermediate points. Mike confirmed that we can ramp without stops, adding that parabolic functions are used for interpolating in and out at each optics step, so as to give us a smooth and consistent stop opportunity when or if desired. John commented that the tune excursions always have the same sign. He suggested that therefore a different type of interpolation between points might reduce these excursions. Rudiger commented that many other effects may alter the tune at the 1e-3 level during the squeeze. Stephane replied that at least we can determine the minimum number of points necessary from the pure optics point of view. Jean-Pierre stressed that the coupling will strongly increase during the squeeze, and therefore a much larger number of points may turn out to be necessary. Responding to a question by Frank, he explained that this expectation is based on LEP experience and that the coupling results from insufficiently known quadrupole tilts.
After remarking that the situation in IR5 is not particularly different from IR1, Massimo addressed the combined squeeze in IR1 and IR5, for which the tune excursions of beam 2 reach peak values of 6e-3 and the maximum beta beating amounts to 10-15%. Frank asked if the optics behaves worse for beam 2 always, i.e. for the squeeze of IR5 as well as for that at IR1. The answer is yes. Oliver explained this as an artifact of the matching procedure. Finally commenting on the other insertions, Massimo described that the IR8 beta* can be reduced not further than to 2 m, due to aperture limitations. Optics files are present but not optimized, and the crossing scheme is not yet included. IR2 optics files for the un-squeeze with protons are still missing.
=> ACTION:
Finalize squeeze/un-squeeze optics for IR2 and IR8 (Massimo)
Concluding his presentation, Massimo underlined that a MAD-X tool for the squeeze has been developed, and that strength limitations will not determine the place where we change the optics or crossing scheme. In particular he made clear that the worse performance of beam 2 is well explained by the jump for Q7, limited to the range 1 m < beta* < 2 m, and easily fixed by adding one additional point in this range of beta*. Transitions for the crossing scheme will still have to be studied. Massimo also recalled that Ralph had proposed to squeeze already during the ramp. This topic came up again during Ralph’s subsequent presentation, summarized below.
Oliver suggested keeping 205 T/m as the maximum value of triplet-quadrupole gradients in IR2 and 8, and continuing the squeeze with other quadrupoles once the nominal triplet strength has been reached. Stephane commented that with this approach the closed orbit would change. Frank pointed out that a simultaneous squeeze in IR1 and IR5 is required in order to restrict the total tune shifts and the bunch-to-bunch tune differences arising from the long-range beam-beam encounters. Jean-Pierre recommended aiming for the best performance with both beams and in particular matching the crossing angle experimentally, including local coupling correction, with imperfections at each point.
Replying to a question by Oliver for the expected time needed for the squeeze, Mike indicated that the squeeze will take of the order of 10 minutes. Roger asked how far IR2 and 8 can go up in beta*. Oliver responded that for beta* values up to 35 or 50 m, there is no issue. For larger values, the orbit correctors separating the beams will run out of steam.
Ralph A. discussed the collimation during the ramp and squeeze. He started by thanking his collaborators, and introducing several colleagues from the ABP group working on collimation, namely Guillaume Robert-Demolaize who presented an AB seminar on 30 November, two PhD students - Chiara Bracco and Valentina Previtali -, and the fellow Thomas Weiler, who are heavily contributing to the studies presented.
Ralph next reviewed the general idea of the three-stage cleaning system, consisting of primary collimators, secondary collimators, and protection devices. He presented an LHC layout prepared by Chiara, including the location of all collimators and protection devices. According to a simple rule of thumb, for the squeezed optics at 7 TeV the absolute collimator gap must be about 10 times smaller than the available triplet aperture. The collimator settings are usually defined in terms of sigma. Ralph went on to explain why the primary collimators are set at an odd value of 5.7 sigma and not e.g. at 6 sigma. This peculiar setting is determined by the aperture of other devices. At collision, we adopt a 4-stage cleaning system 4-stage system, consisting of TCP's, TCS's and TCLA's in IR7 plus the tertiary collimators TCT’s close to the triplets, which are further complemented by the physics absorbers TLCP’s. The transverse distance between the primary and secondary collimators is about 1 sigma (1.2 mm at 450 GeV, and 0.2 mm at 7 TeV).
The fundamental two-stage cleaning principle says that the secondary collimator should not become the primary collimator. Orbit errors could compromise the cleaning efficiency. Above 20% of “transient” beta beating the cleaning inefficiency increases by two orders or magnitude. If the beta beating is as small as in the optimum system, beam loss up to 1 permille during 1 second can be accepted without quench. The acceptable beam loss decreases by a factor >100, to less than 1e-5 of the nominal beam per second if the beta beating exceeds the 20% tolerance.
Stephane asked how exactly the cleaning inefficiency is related to the n1/n2 ratio. Ralph responded that the dependence is weak.
Ralph presented a formula which gives the minimum beta* arising from collimation constraints, including orbit and optics errors. Large triplet aperture, larger beta function at the collimator, or smaller primary collimator gaps allow for lower values of beta*. In principle, during the ramp the collimators could remain at their injection settings, but a partial closure improves the machine protection. Two scenarios for the collimator settings during ramp and squeeze have been analyzed.
Earlier it was estimated by Ralph and others that the global efficiency would improve during the ramp, since the collimation system is optimized for 7 TeV. Closer inspection has revealed, however, that the local cleaning efficiency degrades. The two main changes during the ramp are that (1) the quench limit goes down, and (2) the local inefficiency gets worse if the collimators are not closed. Losses in the dispersion suppressor dominate the local inefficiency. The quench limit was calculated by considering a 0.2 hour lifetime and nominal beam parameters. Single-diffractive scattering in the collimators lies at the origin of the efficiency limit. This physics process changes with beam energy, while at the same elastic scattering angles decrease, degrading the confinement of losses inside the collimations insertion if the injection apertures are maintained.
Regarding the optimized collimator settings during the ramp, Ralph S. suggested that the collimators be closed not immediately, but only after the snapback.
Oliver asked whether there is an issue in synchronizing all the collimator motions during the ramp. Ralph A. replied that Hermann Schmickler and colleagues are taking care of this aspect, adding that the time scales are long. Stephane suggested varying the n2/n1 ratio during the ramp in order to ease the tolerances. Rhodri commented that the orbit and golden orbit may change during the ramp. In particular, he pointed out that if the collimators are closed on the ramp the absolute orbit tolerances would get tighter. Stephane asked what happens in case one jaw does not move correctly. Ralph replied that this will be detected and the beam will be dumped immediately.
Next Ralph showed that the squeeze reduces the overall machine aperture for beta* values below 6 m. Tune spread created by octupoles is also required to stabilize the beam in the squeeze. A plot of the aperture vs. beta* illustrates that MQXA.1l5 is the limiting aperture. Frank asked whether the protective tertiary collimator should not be the limiting aperture rather than a magnet. Responding to a question by Stephane for the minimum n1 value after the squeeze, Thomas quoted a final n1 value of 4.7 sigma. According to Stephane, this value should be about 7. Ralph replied that the official tools were used to compute n1 but that the source of this discrepancy will be checked.
=> ACTION:
Check n1 for squeezed optics (Ralph, Thomas)
Ralph’s master piece was an outline of the draft squeeze procedure. Critical steps are the verification of the collimator settings (jaw position with respect to orbit), the switching on of the octupoles, the squeeze to beta*=6 m with collimator settings compatible to the arcs (partially closed during the ramp), measuring the tail population down to 6 sigma via controlled scraping, adjusting collimators for n1 of next step, also closing dump protection and triplet collimators, beta* squeeze, and finally an iteration.
Gianluigi asked why the feedback should be turned off before the activation of the octupoles. He recommended keeping the transverse feedback system on until collisions are established. Jean-Pierre agreed. Stephane recalled that the octupoles should indeed be turned on earlier, namely during the ramp, quoting LHC Project Report 91 from J. Gareyte, J.-P. Koutchouk and F. Ruggiero. Frank suggested that the octupoles could perhaps be excited at a low value already at injection and then ramped. Stephane replied that this would compromise the dynamic aperture.
Rudiger expressed concern that the cleaning efficiency may be compromised by the scraping. Oliver commented that the scraper would work just as a primary collimator. Stephane remarked that the material of the scraper is different from that of a primary collimator.
Frank commented that the LHC design had always assumed the octupoles stay on with colliding beams at 7 TeV, and, indeed, are needed to ensure the beam stability. Jean-Pierre concurred that the beam-beam interaction will not have the same effect as the octupoles. Frank also asked how long the collimator squeeze procedure might take. Helmut commented on the tails that indeed these are not unavoidable. However, every active element could produce tails. He emphasized that, therefore, octupoles are preferred over the fast feedback. Ralph commented that more discussion is needed on the octupoles and the fast feedback. Helmut added that learning curves and alternative settings should be considered (e.g., finer and coarser settings of the collimators).
Ralph presented the dependence of the orbit tolerance on beta* and beam intensity. The tightest tolerances are of order 100 micron at nominal intensity. The expected orbit excursion in the collimator system was calculated by Ralph S. to be at the level of 100-200 micron during a squeeze with global orbit correction, but possibilities exist to improve these numbers.
Ralph S. added that, assuming a steady-state machine with non-moving quadrupoles, the required orbit corrections during the first squeeze could be fully incorporated into the next fills and thus minimize the transient "theoretically" to zero. However, based on estimates presented in CERN-AB-2005-087, random ground motion is expected to contribute to quadrupoles shifts in the order of 5-10 micron r.m.s., resulting in a fill-to-fill uncertainty of the orbit during squeeze of about 300 to 600 micron, if not compensated by beam-based feedbacks, though he added the question whether we need to rely on the orbit feedback all the time. Ralph A. and Stephane both agreed that regular orbit corrections by operator or feedback will likely be needed.
Ralph A. pointed out that even with only a few bunches the collimator impedance can lead to instability. This issue will be addressed at the LTC of 6 December 2006. Oliver mentioned that all instabilities can be suppressed with higher chromaticity, but then the question then is how much chromaticity can be accepted with regard to beam lifetime.
Ralph recommended squeezing in steps in parallel for all IRs. He added an interesting consideration that the ratio beam-loss power over quench power scales at least with the second power of the beam energy. We would therefore gain a factor of 2 in acceptable beam loss if the squeeze were done at 5 TeV, i.e., there is a clear preference for squeezing at the lowest possible beam energy, for example beta*=1 m at 5 TeV. Rudiger commented that there are two ways to squeeze during the ramp, either continually or in steps, and he asked which of these two ways would be the one considered. Oliver remarked that the final squeeze must always be done at top energy, due to aperture constraints. Stephane mentioned that we should not change twice the slope of the magnet excitation.
Another question is whether the squeeze and ramp be combined. Rudiger noticed that in this case we might not be able to stop the squeeze. Rhodri and Ralph added that also the reference orbit would change. Oliver expressed the worry that we might need to measure tails at 6 sigma amplitude in routine operation. He asked if there are any monitors which could provide this halo information, such as residual gas monitors. Rhodri replied that the only conceivable detection technique is the corona measurement for the synchrotron light. Jean-Pierre wondered whether this question was not academic, as we are obliged to scrape in any case, if there is significant halo at 6 sigma, while in the absence of halo the scraping is not a problem.
At the end of this LHCCWG meeting, Mike aimed at providing some light relief. He introduced David Nisbet, who joined the meeting from the power converters group, Marek Strzelczyk from OP, and Vito Baggiolini from CO, all of whom are developing the squeeze mechanics. Recently, David tested the insertion-quadrupole squeeze functions with the power converters connected to short-circuit loads. Reyes, Marek and the operations group are working on the sequencer.
Concerning the
optics. Marek wrote a Perl script looping over
Mike then defined the parameter space. A parabolic roundoff at each intermediate step provides the possibility for stops. Squeeze optics functions and power converter currents can be displayed from LSA.
Stephane
asked for the maximum current ramp rate over all converters. David
replied that according to his memory this should be about 10 A/s. Walter asked
which magnet transfer functions were used. Mike responded that he assumed the
default, namely linear functions. Oliver stressed that the regions with low
excitation deserve attention, as has been highlighted in
=> ACTION:
Include magnet current limits and check current ramp rates (Mike et al)
Mike presented a live demo of ramping and squeezing test. The power converters in point 5 were considered here. In the LHC sequencer prototype he switched on the power suppliers, went to injection settings, armed the converters for the ramp, started the ramp, and squeezed – all successfully. To reduce the waiting time, he had to interrupt and to speed up the ramp rate. Ralph S. asked whether such rate changes would be intercepted in the future. Mike’s answer was yes, they will.
Ralph A. commented that the synchronization of the magnet ramps does not look to be OK. Mike replied that we have seen an asynchronous ramp (not function driven). Gianluigi wondered about the checks that the ramps are correctly loaded. Mike agreed that the sequencer must include lots of checks.
Mike’s “other stuff” included squeeze factor, collimators, bumps & crossing angles, and feedback systems.
Stephane asked whether the LHCCWG could come up with a recommendation regarding the squeeze during the ramp. The obvious advantage would be a gain in turnaround time. Roger replied that for commissioning we want to separate the ramp and squeeze if we can. First high energy physics runs will be with un-squeezed beams (injection optics). Once we have mastered how to do that, we will commission the squeeze at top energy with low intensity (e.g., down to 4m or 2m) and then go for physics runs with a partial squeeze, pushing the intensity. If as the intensity is increased we find that we cannot get to the same point in the squeeze, we would have to decide whether to back off in intensity or in beta*. Only when all these issues have been stabilized would it be possible to consider making a combined ramp and squeeze. In this case we would make the low energy squeeze to a point that we know is OK, say 6m, and almost always make the rest of the squeeze at top energy.
Oliver asked whether we want to stop on the ramp to check the optics. It was noticed that also dump tests will have to be done during the ramp. There may be requests from collimation to stop on the ramp too. Mike answered that we will be able to stop at pre-defined points on the ramp. Oliver pointed out that the D1 transfer function could be sorted out more easily by squeezing after the ramp. Ralph A. commented that there might be a trade off between squeeze on the ramp and maximum intensity. Rudiger agreed with Roger that this issue is not up to discussion now; and that we should start with the least complex optics to simplify the initial commissioning. Ralph S. recalled that the combined ramp and squeeze also puts much tighter real-time constraints for the orbit feedback which would then need to compute and invert the orbit response matrix at a much faster rate. Jean-Pierre drew the tentative conclusion that this question should be decided when the machine runs.
Stephane commented that slope changes may have an effect on transfer function and magnet performance. Taking up this point, Rudiger commented that field measurements for different ramps and stops have been done and that the results could be reported to LHCCWG. Walter remarked that the magnet measurements had probably not yet been done yet with the relevant cycles. Jean-Pierre confirmed that such measurements are foreseen for 2007.
After the meeting, Massimo commented that studying the magnet behaviour during the squeeze could provide guidance on the final choice on whether or not to squeeze during the ramp. He believed that some relevant data could already be available, even with special measurements still pending, or that, at least, preliminary conclusions could be drawn based on first principles of superconducting magnets. This question may be dicsussed at the next meeting.
=> ACTION:
Study of combined ramp and squeeze in regard to magnet properties (Walter,
Massimo, Jean-Pierre?)
Wednesday December 13th, 14:30
CCC conference room 874/1-011
Provisional agenda
Minutes of previous meeting
Matters arising
Summary of
parameter tolerances (Frank)
Proposed beam
measurement programme in 2007 (Frank)
What to do if
we cannot get in tolerance (All)
AOB
Reported by Frank