Minutes and Summary

Main purpose of the meeting: Commissioning, VdM, physics MD, HL-LHC heating tests

LPC minutes 19 May 2025

Present (P = in person): Chris Young (P), Chiara Zampolli (P), Robert Münzer (P), Andrej Gorisek (P), Paula Collins (P), Eric Torrence (P), Flavio Pisani (P), Silvia Pisano (P), Klaus Monig, David Stickland (P), Witold Kozanecki (P), Lorenzo Bonechi, Anna Sfyrla, Michi Hostletter (P), Giulia Negro (P), Richard Hawkings, Jaime Boyd, Juan Esteban (P), Rosen Matev, Roderik Bruce (P), Riccardo Longo, Gt (George Trad?), Andres GUillermo Delannoy, Gerardo Vasquez, Joanna Wanczyk (P), Matteo Solfaroli Camillocci (P), Tomasz Bold, Peter Steinberg

Introduction (Chris Young)

Witold Kozanecki: I confirm that the VdM filling schemes for ATLAS and CMS are fine.

Chris Young: I was asked to inform that PPS and Velo should be inserted in the heating tests to check the impedance.

Paula Collins: when would this be discussed?

Chris: I expect at the LBOC. AFP will not be present in HL-LHC so it is not relevant.

Roderik Bruce: [on pO and OO] there are ongoing offline discussions with some people: there will be a joint collimation group and machine protection meeting on 13 june to discuss and make a final decision on the allowed intensity limit for the two runs and the qualification needed. From pre-discussion, it seems reasonable that we can go higher than what was discussed before, with the motivation that we have we have individual bunches which are spaced by at least 1 us, so we’re not inherently more unsafe if we put more bunches. For pO, as you may remember, we need many bunches, we need at least 48b, for OO we had 18. We could simplify the machine settings, using the same configuration for pO and OO in terms of filling scheme with 48 bunches in both cases but with higher intensity. This would mean that you get more luminosity for both pO and OO. The only exception is LHCf which will be leveled all the time anyway. We could imagine having 48 bunches with order of 3e10 charges instead of 1e10 that we had before, both for pO and OO. This should give a significant lumi boost to everyone, but it is not a promise, it should be discussed with rMPP. I mention it here in case some experiment has some constraints, if you need to reconfigure some triggers, or if you there is anything on your side that would prevent this configuration, let me know.

Chris Young: this would trigger the levelling criteria in pO.

Roderik Bruce: it depends. The pileup in pO will be higher because we will have higher intensity. The pileup in OO will be the same because we will have more bunches but with the same intensity. For pO, do we have levelling targets for the other experiments apart from ATLAS?

Paula Collins: not for LHCb.

Chris Young: before we were looking at mu of about 0.1, so you’re going up a factor 3^2.

Roderik Bruce: you can count an order of magnitude higher.

Chris Young: so whether 1 is too high, would trigger levelling.

Paula Collins: where were we?

Chris Young: you were 0.07, now it’d be 0.7.

Roderik Bruce: for OO the pileup stays as we announced before, because we’d have more bunches, but the same intensity. While for pO we’ll have the same number of binches, but higher intensity, so the pileup will go up. The question is whether there is any leveling target or maximum pile up target that you can stand for pO?

Robert Muenzer: we need to check the value for pO but for sure we will not want to exceed a certain pileup. Our limit is 0.05 which we discussed to go higher for OO, but for pO we need to check.

Roderik Bruce: for OO we counted 0.2 for ALICE for the lumi calculation.

Robert Muenzer: yes, but if we have contingency for the lumi, we might even decide to lower it a bit. We need to discuss. For pO we’ll need to check.

Andrej Gorisek: for ATLAS we are driven by LHCf.

Roderik Bruce: for ATLAS, it is LHCf which drives. Another point: we’re also looking at filling schemes. Is everybody happy with 1 us separation for both pp and pO?

Eric Torrence: we’d be happier if it were longer.

Chris Young: but the 48b scheme should give you a pretty good limit, in pO.

Roderik Bruce: with 48b we had 1 us. You were not happy with 1 us if we had less bunches. With 48 it should be ok. 

Andrej Gorisek: there were discussing whether to have 1 bunch per injection, or 2 or 3. 

Chris Young: this was in the 18b scheme. 

Andrej Gorisek: previously we saw that we got very constraining limits when there were more than 1 bunch per injection in the OO scheme as then they were less spread out.

Roderik: wasn’t the issue when we had only a small part of the ring covered? In the 48 bunch scheme we will have more bunches per injection, but we will cover a much larger part of the ring. Instead of 18 where we had holes.

Andrej Gorisek: it would be better to have 2 us or 1.5 us separation from the SPS. 

Chris: we gave you a fill pattern with 48 bunches for pO, which was fine. 

Roderik Bruce: it is the same issue for pO and OO, I guess. Please check again. 

Eric Torrence: in pO we’ll be so far levelled so that the collision rate won’t be very high, so that it is not an issue.

Roderik Bruce: then it is worth checking for OO.

Andrej Gorisek: For the 48b scheme with 32 collisions per IP there was a 53kHz limit. This has 4 bunches per injection and a 1us spacing. A 1.5 us separation would be better.

Roderik: the LHC injection kickers have a certain length, and we cannot have more space with 4 bunches, then we should reduce the number of bunches. I think 4.5 us should be the limit. But then the injection would take much longer for this shorter run. The problem is the LHC injection kicker. The LHC injection kicker pulse is adapted to the current train length of the LHC. You can make it longer but it needs commissioning.  

Michi Hostletter: the problem is an interplay between the injection kicker length and the abort gap keeper: you need to set up the abort gap keeper so that if you inject in the last bucket you are allowed to inject, you don’t leak the tail of a potential train in the abort gap. So by shortening of the injection kicker length to the 144 bunch we need operationally, we can make a bit more space by shortening the abort gap keeper to fit more trains for normal pp operations. To change this for a single run is a few hours of changing and then again back.

Chris: we can try with injections with 3 bunches. This would mean to get the 16 bunches in each part of the ring we would have 6 injections; 5 of 3 bunches and a single bunch. The total for each beam would be 18 injections, rather than 12 in the 48b scheme with 4bpi.

 

ATLAS (Eric Torrence)

Michi Hostletter: what does not help, is that now, during the intensity ramp up we cut short the fill exactly at the end of the xing angle antilevelling. To distinguish this from for instance a potential residual pileup dependency, we’d need to have a fill with a few hours tail behind. The question is how much the two would diverge if you did nothing but just wait. You are out of levelling 45min - 1h earlier than CMS, and we start the xing levelling when also CMS is out of levelling. Maybe there is something with the orbit effect, parasitically one might introduce a bit of separation while doing the xing angle antileveling. There is a feed-forward for that, but it might not be perfectly tuned. To get an handle about how this would evolve with time, it’d be important to have a fill that we don’t dump immediately. This we’ll get for free when we get to the full machine, when we have 14h fills, not just 8-9h. We’d need to see how this evolves without touching any beam parameter. We’ll have the full machine in a few days.

Eric Torrence: for now we’re not concerned about this. 

David Stickland: it would be very useful to continue after the end of the antilevelling to measure linearity. Not always, but doing it from time to time it’d be useful.

Andrej: we make the assumption that if we ask to have the beam spot centering at the beginning of 2026, this will not be an overhead.

Roderik: we need to check the aperture and realign the collimators, but if this is at the beginning, it should be fine. 

Chris: the additional time should be quite minimal.

 

CMS (Giulia Negro)

Michi Hostletter: [about the moving of the beam spot] this is consistent: in the crossing plane we see some separation (3-3.5 sigma) at the end of the b* levelling, we tuned a bit, let’s see in the next fill. 

Chris Young: do you see it in ATLAS?

Michi Hostletter: I saw it in a similar order of magnitude.

Eric Torrence: yes, we see something similar.

Giulia Negro: [about the fills with INDIVs] one every few weeks or every month is fine for CMS.

 

ALICE (Robert Münzer)

Michi Hostletter: [concerning the lumi spikes] I compared 800b and 400b, but if you look at the separation, there is not much difference. The effect on the hight of the steps is a percent, not huge. To be seen how it evolves with full machine. But I don’t think that they will magically disappear, even if they can get a little bit less violent. If you then get back to high lumi, they will be like now. 

Robert Muenzer: it might be that with the same luminosity but more bunches, the effect is lower. In the end this is something we might not expect in PbPb.

Michi Hostletter: in PbPb you’re closer, you go up to head on at the end of fill, and when we go head on, the effect is negligible because the derivative of the lumi with separation is zero. If you look at the DOROS, you see that you go 3 um lower in separation and then 2 um higher and then back to zero, and when you are head on, these 2-3 um do nothing, but if you are in the deep slope of the separation because you have an initial separation, then it is much more significant.

 

FASER (Jaime Boyd)

Chris Young: would you like a small test in ADJUST to see what moving the TCL6 would look like?

Jaime Boyd: it would surely be interesting, to be seen if it is easy to fit it in the schedule.

Chris Young: a short thing at the end of the fill might not be difficult. Having more bunches would make it faster. 

Roderik Bruce: of course we cannot move TCL6 by a lot at the end of the fill, since we have interlocks. I think we can do 400 um. In the past we did some tests with setup beam. 

Chris Young: this is something we can think about.

 

LHCb (Paula Collins)

Matteo Solfaroli Camillocci: how long do you need to be warned before we go in ADJUST from SB? We have some tests planned possibly in the night. 

Flavio Pisani: we do not need much time for the phone call, but if we know in advance that in a certain fill they will move to ADJUST, we can pre-warn the detectors. Then few minutes are enough. If we know that the next test in ADJUST can be dangerous for the detectors, we need to know in advance.

 

SND (Gerardo Vasquez)

Chris Young: is the excess coming from the blob on the right [see s2]? The one that does not come from the IP?

Gerardo Vasquez: we are not sure at the moment. We need to look at the new fills and do lumi scaled plots. 

Roderik Bruce: do you think that the extra 40% is problematic?

Gerardo Vasquez: the lower the better, but it is better than last year. We would not need to do ½ of the program as we did last year.

Chris Young: at some point we need to discuss with you and FASER when to do the exchange. 

 

AOB

Michi Hostletter: think for the strategy for the emittance scan. Now we do 15 points at every fill, but this is probably too much in operation. For ATLAS and CMS, since on the machine side we’d like to have both with the same configuration, we should agree. This should not be done at every fill, so we should converge on a strategy.

Chris Young: do you want to do them also at the end of fill?

Michi Hostletter: 50% of the fills will not be dumped by OP, for the rest it would be nice to have them to have a handle on the emittance evolution. There, they cost not much since the other experiments are head on in the best lumi they could ever get. So the ones at the beginning are a bit more constrained, since we cannot start the ramp up till both experiments are done scanning. 

Chris Young: we agreed to do it every second fill. It could be either every even of odd number, or depending on what was done in the last fill, and we can use 15 points. You can then decide on the machine side if to do one at the end, if there was one done at the beginning.