Updates from MADX, commissioning and MDs
LNO section meeting on Updates from MADX, commissioning and MDs
News:
Rogelio reported about the LMC decision that the baseline LHC optics for 2024 will be with reversed triplet polarities in IR1&5 without performing a beam test in 2023. Stephane added that an extra week should be foreseen at the start of 2024 commissioning to test and commission this. In case of problems it will be possible to rollback to the present polarity, within a time estimate of two days. Stephane said that the optics are mainly ready. Some fine tuning for forward physics and the preparation of a 19 m VdM optics remain to be finalized. In addition, the polarity change will also require work on beam-beam and collimation.
MAD-X status, Riccardo, slides
MAD-X 5.09.00 was released on the GitHub on May 5 implementing many significant modifications as listed in the a newly created DOI with changes on the TWISS, Track and PTC level. This release is available through CpyMad. It is not yet deployed on lxplus due to failing tests that require better understanding and fixes. The next release will include an important thick dipole tracking fix already merged into the master by pull request PR1182. A priority will be improving the TWISS accuracy for nonzero pt and improving TWISS, EMIT and TRACK with radiation as required for FCC-ee as further discussed in the next presentation. One concern is that some tests (in particular PTC) are expensive to maintain and sensitive to minor numerical differences similar to changes observed when changing compiler versions. There are still many uninitialized variables (checked by memory initialization to NaN) and memory related errors. The list of MAD-X open issues has currently 50 entries. While some of these like the RBEND thick slicing issue have already been addressed with a pull request and merged into the master, others will require significant further efforts and help.
Review of MAD-X for FCC-ee studies, Guillaume Simon, slides
Guillaume reminds that energy losses by synchrotron radiation in FCC-ee will be significant and require special attention to accurately deal with local energy changes (sawtooth, tapering) and
accuracy for non-zero momentum offsets. Tapering for multipoles and octupoles has been implemented for MAD-X 5.09. He shows a comparison for the predicted tunes without radiation,
and with radiation and tapering with and without EXACT option (evaluate on the actual orbit), for the TWISS, EMIT and TRACK modules for several recent MAD-X releases. A scaling of strength parameters with momentum offsets (1/1-δ(pt)) as implemented in MAD-X 5.07 for TWISS is problematic for symplectic tracking. A new implementation of the scaling is in progress to obtain consistent and sufficiently accurate results in TWISS, EMIT and TRACK including tapering and EXACT evaluation.
In the following discussion, Stephane suggested to see if (1/1-δ(pt)) can be replaced with an exponential without spoiling symplecticity.
Tobias, Rogelio and Riccardo commented on assuming small px, py and/or using symbolic algebra tools (like sympy, already used in MADX), and/or applying differential algebra and/or symplectic integration approaches as used in PTC/MAD-NG.
High (Medium) beta* (120 m) commissioning, Tobias slides
Tobias reports about the recent commissioning and optics measurements and corrections to a beta* of 120 m in IR1&5. Transverse coupling was corrected at 75, 100, 110 and 120 m. Even before correction, the beta-beat was found to be already quite good with a peak value of 15%, that got reduced to below 5% after correction. K-modulation was used to measure beta* for both beams in IP1&5 and found to be within the nominal value within 1-2% except for beam 1 horizontal in IP5 which is larger by 4%. Tobias also shows that the measured normalized dispersion around the ring looks good for this optics.
Ilias will write a report including further details in the IP->Roman pots transfer optics from the available measurement data.
3Qy correction in LHC injection, Ewen slides
Ewen describes that the nonlinear optics at injection MD had several ambitious goals.
Four of these, virgin vs nominal b5 correction tests, various 3Qy studies, baseline f2002/f0040 resonance driving terms (RDT) and detuning measurements could be addressed in the available time.
RDT response depending on MO powering and initial checks of skew octupole RDT response and lifetime could not yet be studied.
The 2023 phase knob was found to ‘accidentally’ also result in a significant reduction of the LHCB1 3Qy resonance, with only a minor degradation of the already smaller LHCB2 resonance (slide 4). Using the common MCSSX correctors in the IRs, it was possible to define and test robust knobs to control and independently correct 3Qy for both beams. Ewen shows that the currents used for the MCSSX were typically few Amps with a maximum of 24 Amps, still well below the maximum available (50 or 100 A). As expected by the model, the feed-down on tune and coupling was measured to be small and the effect on beta-beating found to be negligible. The corrections are made for injection and trimmed out early in the ramp. The last minutes of the MD were used to study effects on lifetime and to make sure the proposed corrections do not degrade the beam lifetimes.
e-cloud effects with 3Qy correction, Konstantinos Paraschou slides
Konstantinos presented a first look on electron cloud effect of the 3Qy resonance. Initially pilots were injected on the nominal Qx, Qy = 0.275, 0.293 tunes, and tune, chromaticity and coupling corrected, followed by injection of 12 bunches. Injecting 3x72 bunches trains resulted in coherent instabilities attributed to ecloud. The main ecloud related measurements were then performed with 3x48 bunches of 1.4e11 protons / bunch. The vertical tune was gradually increased in steps to 0.31, 0.32, 0.33 (crossing the coupling resonance), 0.34, 0.35 waiting 5 minutes between changes. This was done for the nominal configuration and repeated using the best 3Qy correction, see slides 3, 4. It demonstrated that the best correction helps to reduce losses and increases lifetimes.
Loss maps were also recorded during the MD, finding no issues and not revealing any dependence pf TDIS losses on resonance driving terms. A first look at the bunch intensities showed that the first bunches in the train loose less intensity. The interpretation of the observations is not straightforward and will require more analysis and dedicated simulation of the MD conditions.
In the following discussion of both 3Qy at injection presentations, Rogelio comments that the results of the 3Qy corrections look promising and proposes to use them after the current short stop (TS1). Ecloud effects depend on both tunes and intensity. The observations presented here show that the 3Qy corrections increase the available tune range.
Joerg says the corrections should be trimmed out in the first minutes of the ramp to disappear around at 1 TeV. The ramp down speed of the RCSS is limited to roughly half an Ampere per second. A proposal should be sent to Matteo to check and put this in the ramp.
b5 correction at LHC injection, Mael Le Garrec slides
Mael reminds that correcting b5 (decapole) and demonstrating beneficial effects is difficult. b5 generates a third order Q’’’ term. The currently used Q’’’ correction is based on field measurements (FiDeL model) and appears to overcorrect, as already observed in Run 2 and the Run 3 commissioning last year. A more linear chromaticity (as function the momentum offset δ) was obtained using a smaller beam-based correction. Improving Q’’’ however appeared to increase the f1004 1Qx - 4Qy resonance driving term. This was further studied in the recent MD. Using a fraction of the FiDeL correction, a uniform or more local correction, Q’’’ could be improved in this MD without significant degradation of RDTs and showing a good lifetime.
Outcome: After the meeting the 3Qy and MCD corrections were sent to Jorg and Matteo and they were incorporated in the physics cycle (operational from the end of the TS).
RDT based calibration of the skew sextupoles in PS and machine learning pole face winding studies, Wietse Van Goethem slides
Wietse reminds that pole face windings (PFWs) were already part of the original PS design and produce quadrupolar, sextupolar and octupolar field components. In the 1960s, low energy quadrupoles were installed in the PS to increase the flexibility in tune control. The current PFW settings and control were established many years ago. They were refitted in 2009. One challenge is that the PS cycle crosses transition and another that tune and chromaticity are not measured independently.
The aim of the study reported here was to improve the PFWs control and corrections using more recent tune and chromaticity date recorded during a PS cycle with radial steering and by applying an improved analysis technique using ANN (Artificial Neural Network). Promising results were obtained both for low and higher energies. It is planned to take more data this year and to apply this technique to obtain improved corrections for the LHC cycle in the PS.