LHC Studies Working Group Notes from the meeting held on 13th September 2011

LHC Studies Working Group Notes from the meeting held on 13th September 2011 The meeting was dedicated to the results of LHC MD#3, which took place from August 24th to 29th. The slides can be found at the following link: https://indico.cern.ch/conferencedisplay.py?confid=153652 1. 1m β* (J. Wenninger) The commissioning of 1m β* was summarized. During MD time, the 1-m optics was tested with tight collimator settings and a half crossing angle of 100µrad. The β-beating at 1m was within 10%, the β* was measured with the K-modulation and was found to be consistent with 1m (within errors), so that no corrections had to be made. The first time two 36-bunch trains colliding in all IPs were squeezed to 1m, some of the bunches lost up to 40-50% of the intensity; observations of the bunch-by-bunch losses pointed to long-range (LR) BB effects (loss shape, for bunches colliding in IP1 and 5 only). In fact, the beam-beam (BB) separation was below 6σ at the triplets. It was noted that with the tight collimator settings, about 0.5% of the beam was scraped away during the ramp. Given that the physical aperture around the triplets in the IPs was found to be larger than initially assumed during another MD, it was decided at the LMC to run with the same collimator settings as before the TS (TCTs 11.8σ in IR1, 5, 8) and to keep the crossing angle of 120µrad, provided the additional aperture would be confirmed by further measurements and analysis (in fact, ~2σ additional margin is available). The two 36-bunch trains were squeezed and no loss problems were observed anymore (no changes of octupoles or ADT gains with respect to 1.5m). The 1m β* is now operational but the exercise highlighted some important issues: the losses during the ramp, the orbit stability in the squeeze that must be improved to run comfortably with tight collimator settings (loss spikes observed consistent with orbit drifts), the beam stability that needs to be proven ok with tight settings and larger LR BB separation. These issues should be studied in 2011 if the tighter collimator settings are to be used in 2012 to lower the β* further. L. Evans asked whether the coupling knob between 1.5m and 1m needed changes, J. Wenninger and R. Tomas answered no, meaning that either the ITs are very well aligned, or the local corrections are very good. 2. Tight collimator settings with β*=1m (R. Bruce) The tight collimator settings were to be used with β* 1m prior to the finding of the extra available aperture at the triplets. The settings are introduced in the ramp function, with TCPs closing to 4σ, TCS s to 6σ. Two nominal bunches were ramped and squeezed to β* of 1m (half crossing angle of 100μrad) and the TCTs aligned. Then the separation bumps were collapsed, collisions found and the TCTs realigned and then retracted to 9.3σ. About 10 minutes per collimator were needed for the alignment, with b2 slower than b1 due to high noise levels. It is not well understood why the TCTH.4R5.B2 center offset was found shifted by 860μm (corresponding to less than 1σ). Loss maps were performed and the cleaning efficiency from the previous MD reproduced (TCTs retracted by 2 sigma; loss maps still fine; at least 2 sigma margin). The loss maps are ok for operation, although a degradation of the loss pattern over time is observed and more visible due to tighter margins. The leakage to the cold magnets in IR7 is only slightly

worse than in the previous MD and could probably be improved by a new collimation setup in IR7. The results confirm the assumptions used in the calculation of the intensity reach, where tight collimator settings are needed. To be noted is that the leakage between TCT and triplet is sufficiently small and that the tight settings with possibly smaller β* could be useful next year. L. Evans commented that in case of heating problems the collimators should be opened as much as possible (collimator heating goes with inverse cube of the aperture), compatibly with good protection and low backgrounds at the experiments. S. Fartoukh asked whether the tight settings are needed for the IR3/7 dispersion suppressors cleaning efficiency, R. Bruce answered that with relaxed settings we are not limited at 3.5 TeV, D. Wollmann added that the intensity limit from the review was reached for tight settings. 3. Long bunches and lifetimes (J. F. Esteban Mueller) Longer bunches help reducing heating (at MKIs, beam screens and collimators), cause less e-cloud effect. Additionally, for the same total voltage the emittance is larger, and this is beneficial for stability and gives lower IBS growth rate. Longer bunches though give potentially a degraded beam lifetime (more particle losses). The goal of the MD was to relate single beam lifetime and bunch length. 8 nominal bunches per ring were ramped with blow-up target at 1ns (to be noted, the large spread in intensity), and then blown-up at the flat top with phase noise that was modulated in amplitude along the ring, so to achieve a spread in bunch length. For b1 the bunch lengths were between 1.15 and 1.50ns and for b2 between 1.25 and 1.65ns. It was observed, as expected, that longer bunch lengths correspond to higher losses (factor 2 between 1.2ns and 1.4ns, from 0.5%/h to 1%/h). The voltage was reduced at the end of the study to further increase the bunch length. As a complement to the study, it is suggested to try and change the operational setting from 1.25ns to 1.3ns and observe the lifetime during a few physics fills. P. Baudrenghien added that the operational blow up during the ramp is tailored to excite the core of the bunch only, while during the MD broadband noise was applied and this is likely to have enhanced the tail population. F. Zimmermann suggested redoing the study with the same spectrum as operationally used during the ramp. R. Bruce suggested comparing the measured loss rates with the predictions from different codes. 4. Beam Instrumentation (F. Roncarolo) It was noted how the BI MD was less efficient than the previous ones, resulting in part of the program not being carried out. The B1 direct dump Beam Loss Monitors were calibrated: the signal was normalized to the intensity to get a calibration factor, a 10-30% difference with respect to the b2 calibration was found and is due to the slightly different detector positions. Small residual variations are due to space charge effects in the ionization chambers. The fbct with new logic in the acquisition system was calibrated against the DC-BCT by means of scraping high intensity bunches. A very good agreement was found. Concerning the Beam Position Monitor System, the new IIR filter was tested (it allows much longer filter length, corresponding to lower rms noise on the orbit but slower convergence) and the response studied by changing the orbit by an RF trim. An automatic best IIR filter selection is under discussion (dependent on the filling pattern).

Also, the synchronous and asynchronous orbit modes were compared (the asynchronous mode is auto-triggered, while the synchronous mode allows the gating on a few selected bunches, which could be used for the strip-line detectors to avoid directivity problems); the two modes showed similar results in term of absolute and rms orbit. J. Wenninger added it was recently proven that the bunch-by-bunch orbit is not precise enough to measure differences due to BB effects. S. Fartoukh and/or F. Zimmermann noted that the synchronous measurement is noisier than the asynchronous, F. Roncarolo replied that anyway the rms noise is for both well in specification. Little time was spent on the Beam Gas Ionization, where the new control software for camera gain and gate was tested and a saturation level was found (useful towards the optimization of the working point). Also the Synchrotron Light Telescope suffered from lack of time and only one out of four orbit bumps was performed to study the magnification and the focusing while moving the CCD camera. Measurements were performed for the Wire Scanners during the scraping to test an algorithm to automatically subtract the 200Hz noise (background acquired for a slot in the abort gap, to be used until a solution is found). The Matching Monitor had been installed for b1 shortly before MD#3, Unfortunately the alignment was not well optimized and nothing could be done during the MD; the alignment was then optimized during the TS and the matching monitor is now ready for tests with (Inj&Dump required). J. Wenninger noted that b2 would have been more interesting; F. Roncarolo explained that b1 was easier to install. 5. Long-range beam-beam (W. Herr) The head-on (HO) MD was postponed to allow extra time for the 90m optics, it was agreed that it will be scheduled during the month of October in physics time. The LR MD was carried out, though only one ramp was performed due to lack of time. The crossing angle in IP1/5 was reduced in simultaneous steps and followed by the corresponding TCTs. The filling scheme was designed to have collisions in all 4 IPs, giving a different number of HO collisions for the different 36-bunch trains. Losses were observed mostly for bunches colliding in IP1/5. IP2 was re-separated at the end of the test, giving one less HO for some bunches: these bunches showed fewer losses, even if still dominated by the LR in IP1 and 5. A parasitic observation was done during the 1m β* setting up: a beam-beam separation of less than 6σ (for 2.5µm emittance) resulted in losses only for bunches colliding in IP1 and 5 (separation bumps not collapsed). F. Zimmermann pointed out that the relative losses of b1 and b2 were different in the two experiments. All these observations proved that losses are dominated by LR effects in IP1/5. They show a strong dependence on the separation and the number of encounters; an additional head-on encounter increases the losses slightly according to the first experiment. A bunch spacing of 25ns will make LR effects more severe, and a few trains should be collided to test the minimum separation required (trains of at least 48 bunches, scrubbing required). It was pointed out that no time was available to test the dependence on the tune working point. R. Assmann commented that a BB study with 25ns beams will make an important contribution to the decision on how to run in 2012. 6. IR1 and IR5 aperture at 3.5 TeV (M. Giovannozzi)

These were the first aperture measurements at 3.5TeV and particular care was taken to minimize the losses (MPP approved): only one pilot bunch was used, with blown up transverse emittance, and the local orbit bumps were closely followed by TCT movements to measure the retraction between TCT and triplet aperture. The procedure was the following: TCT opened by 0.5σ, bumps increased by 0.25σ, BLM spikes at TCT vs at MQX/Q2 checked: if the losses were higher at the TCT, then could open more, until the losses are at the triplet. The IR1/5 H/V measurements were completed in ~4 hours. The measured aperture, until the losses were seen at the triplets, is: for IR1H 19.8σ, for IR1V 18.3σ, for IR5H 19.8σ, for IR5V >20.3σ (limited by the strength of the corrector). The findings of this MD were confirmed with a second session of measurements at 1m β* with 120µrad half crossing angle and relaxed collimator settings, and then quickly put to use for physics. An analysis is being prepared to compare these findings to the aperture measurements performed in February 2011 at injection energy (used reference orbit with additional crossing angles, measured absolute aperture by emittance blow-up and local bumps at the triplets). Preliminary results show good agreement between the two measurements (see example for IP5H in the slides, note the 2mm difference between the aperture calculated from the theoretical model and the one derived from the BPM interpolation). An error analysis is being prepared. R. Assmann pointed out that also the other side of the aperture should be measured to exclude possible offsets and find the overall size of the hole, as the results are so far in surprisingly good agreement with the mechanical aperture. M. Giovannozzi agreed, requesting more MD time. Loss maps were performed with the remaining beam (120mrad half crossing, relaxed collimator settings, β*=1m), confirming that the MQX were not exposed to beam halo (some leakage from TCT to MQX for B2 IR1L was noted). Parasitically tune and coupling were measured during the scans and might provide information on the triplet field quality. 7. Blow up with ADT (W. Hoefle) The MD was dedicated to the setup of controlled blow up with the transverse damper (b2 only). Gated noise is used (from white noise generated on the FPGA running at 40 MS/s), similar to what was prepared for abort gap and injection cleaning. Due to firmware issues this was deployed on a test crate in SR4 with gating connected to b2 H/V dampers. The noise is wideband, but can be filtered by IIR low-pass filters to have more power available per betatron band. The noise was typically triggered for a 1s duration and then repeated a number of times to achieve the desired blow up/losses. The noise generator is a 43-bit long linear feedback shift register (repetition period: 61 hours) and was implemented on the FPGA by D. Valuch. If faster losses are needed in the future, a band pass filter can be implemented to limit to a particular betatron line. In the first part of the MD, a pilot was scraped away (with two pilots in the machine) as a proof of principle that the blow up is selective. The emittances were measured for both pilots: while the first pilot kept an emittance of ~2µm, the second one was blown up to ~18µm (to be compared to collimator settings). Then, two 6-b batches spaced of 925ns were used to prove that it is possible to touch only one batch (proven by fbct and BSRT measurements). The 925ns spacing proved that the MKI gap is sufficient for the ADT risetime. S. Fartoukh asked if the blow up can be selective to single bunches, W. Hoefle answered that a spacing of 925ns is needed. Finally, loss maps at 450GeV were

performed with one nominal bunch, comparing the 3rd order resonance method and the excitation with the damper (with ADT feedback loop off). The loss maps give very similar results. Small differences are to be commented on by the collimator team. Note that there are differences if the noise is applied with the ADT feedback loop on or off (off: timescale similar to 3rd order resonance; if the loop is on losses are spread out on longer timescale for the same kick strength). This method opens new possibilities for loss maps at 3.5TeV: EoF physics beam can be used and the blow up applied to selected bunches (starting in abort gap and then moved into the beam). Given that the ADT feedback loop should be off for the loss maps, D. Valuch added that switching off the ADT feedback loop for selected bunches should still be tried. R. Assmann noted that the fact that the loss maps by different methods show good agreement proves that the two methods to produce them are both ok. R. Assmann also commented that this new method opens also possibilities for future MDs, e.g. on quench limits. P. Baudrenghien asked if blow up by anti-damping was tried out, D. Valuch answered that the losses would then be very fast (few turns), unless the gain was heavily lowered. D. Valuch also added that there is only 1MB of free memory left in the ADT system and a major upgrade of the system is planned to overcome that (switch to Linux, new drivers, new FESA classes), until when additional features have to wait. 8. 25ns injection (B. Goddard) Also this MD suffered from little availability, resulting in only 2.5h of beam time. The beam consisted of about 1e11ppb, with 2.5-3.0µm and strong SPS scraping (10-15%). No Transfer Line (TL) steering or collimator adjustment was needed between 50ns and 25ns cycle (some issues with settings copy/functions though). Trajectories were good for both transfer lines (300µm rms for all planes); losses also look very good, in fact better than for 50ns despite the larger emittances, maybe due to the large scraping, to be studied further. The injection of 12 25ns spaced bunches as intermediate intensity for TL steering worked well. Injection of 24-b trains gave no issues, a minor vacuum activity was observed. B2 48-b injections were tried twice, first with ADT on then with ADT off. In both cases the beam got dumped right away (after 1000 and 500 turns respectively) due to IR6 BPMs and IR7 BLMs. The aim of the next MD is to reach at least 72-b injected and stored, possibly injecting 36-b trains (before 48-b) and using higher chromaticity. W. Hoefle added that 4 hours/beam are required to setup the ADT for 25ns beams and are a prerequisite before the continuation of these studies. 9. 25ns instabilities (H. Bartosik) An analysis of ADT pickup data for the 2 48-b injections during the LHC 25ns MD was presented. The data consists of the ADT pickup bunch-by-bunch readings for the last 73 turns before the dump as recorded in the post mortem buffer. The first injection was with ADT dampers on, the beam was lost after ~1000 turns due to IR6 BPMs. A growing motion along the train is observed mostly in V, but a coupled bunch motion is not evident (as a very high frequency motion is observed). A small oscillation amplitude for the first 20 bunches is followed by increasing amplitudes for the last 28 bunches (especially in V). From a single bunch FFT analysis it is visible that the V tune appears in the H plane for the last bunches; from an FFT along the bunch train for each turn, it is shown that the low frequency part of the spectrum is suppressed by the ADT dampers.

The second injection was without the ADT, the beam was lost after ~500 turns (due to fast growing losses in IP7). A small oscillation amplitude is seen for the first 20 bunches, while the amplitude is significantly larger for the last 28 bunches. The single bunch FFTs shows small sidebands in H (maybe linked to large losses), while the FFT along the bunch train for each turn shows that the low frequency part of the spectrum is dominant (indication for coupled bunch motion). It can be concluded that an unstable motion of the second half of 48 bunches was observed on the ADT pickups, the low frequency part can be handled by the dampers, but the high frequency oscillation remains (mainly V). Without the dampers, a coupled bunch motion is observed in both planes. To be noted that the damper delay was not yet setup for the 25ns beam at the time of the MD. G. Arduini underlined that the headtail monitor should be available for the next MD to conclude whether it is single bunch phenomena or not (Action: BI group). E. Metral commented that the observations are consistent with F. Zimmermann s predictions of saturation of electron cloud and fast instabilities after about 20 bunches in the train. F. Zimmermann was puzzled by the complete absence of the high frequency component of the oscillation observed with the damper off, G. Arduini commented that with damper off the instability could have been due to resistive wall in addition to e-cloud. 10. RF observations with 25ns beams (T. Mastoridis) The RF team conducted parasitic measurements during the 25ns MDs. It is noted that no issues were observed on cavity beam loading or klystron power. There was no evidence of growing longitudinal dipole or quadrupole oscillations after the injections of the 25ns batches (while they have been observed with a 50ns spacing). The difference between the phase of the cavity sum and the individual bunch phase was monitored, and a clear correlation was observed between the batch spacing and the mean phase offset per batch, most probably due to electron cloud related energy losses. For bunchby-bunch measurements, the beam phase loop resolution needs to be improved. The injection phase error for the first 24-b train showed a clear correlation with the expected response of the SPS cavities. G. Papotti commented that some additional information on the relative bunch-by-bunch positions can probably be extracted by the LHC BQM. B. Goddard asked about a calibration of the phase error in terms of electron density. E. Shaposhnikova answered that a paper was published at IPAC from GSI that quoted 10-4 deg/m. P. Baudrenghien added that the mean phase over all bunches is logged, while single bunch measurements still need development. G. Arduini underlined that it would still be interesting to compare 75ns, 50ns and 25ns data at least in relative terms, e.g. beginning to end of the scrubbing runs, to compare the results to other measurements (e.g. heat load on beam screen). E. Shaposhnikova had estimated about 1 deg from heat load per 1 deg of phase, a consistent result had been acquired in the beginning of the scrubbing run (average phase data had been presented for b1; for b2 drifts had been observed that were not yet understood). E. Shaposhnikova recalled that the stable phase is a very sophisticated measurement and not fully operational yet. The next meeting is devoted to the preparation of MD#4, meeting to be held in 874-1-011 on October 4th at 15:30. Giulia Papotti

List of participants ARDUINI Gianluigi BE-ABP-LIS ARGYROPOULOS Theodoros BE-RF-BR ASSMANN Ralph Wolfgang BE-ABP-LCU BAER Tobias BE-OP-LHC BARTMANN Wolfgang TE-ABT-BTP BARTOSIK Hannes BE-ABP-LIS BAUDRENGHIEN Philippe BE-RF-FB BHAT Chandrashekhara BE-ABP BRACCO Chiara TE-ABT-BTP BRUCE Roderik BE-ABP-LCU CALAGA Rama BE-ABP-LCU DEHNING Bernd BE-BI-BL DROSDAL Lene Norderhaug BE-OP-LHC ESTEBAN MULLER Juan BE-RF-BR EVANS Lyn PH-UCM FARTOUKH Stephane BE-ABP-LCU FERRO-LUZZI Massimiliano PH-LBD GIOVANNOZZI Massimo BE-ABP-LCU GODDARD Brennan TE-ABT-BTP HERR Werner BE-ABP-CC3 HOFLE Wolfgang BE-RF-FB HOLZER Bernhard BE-ABP-LCU JOWETT John BE-ABP-LCU KAIN Verena BE-OP-LHC MASTORIDIS Themistoklis BE-RF METRAL Elias BE-ABP-ICE MOLENDIJK John BE-RF-CS MUELLER Gabriel Johannes BE-OP-LHC NEBOT DEL BUSTO Eduardo BE-BI-BL PAPOTTI Giulia BE-OP-LHC RONCAROLO Federico BE-BI-PM ROSSI Adriana BE-ABP-LCU SALVANT Benoit BE-ABP-ICE SCHMIDT Frank BE-ABP-ICE SHAPOSHNIKOVA Elena BE-RF-BR TIMKO Helga BE-RF-BR TOMAS GARCIA Rogelio BE-ABP-CC3 VALUCH Daniel BE-RF-FB WENNINGER Jorg BE-OP-LHC WOLLMANN Daniel BE-ABP-LCU ZERLAUTH Markus TE-MPE-MI ZIMMERMANN Frank BE-ABP-LCU