Summary notes of the fortieth meeting of the LHC
Commissioning Working Group
Tuesday February 12th,
14:00
CCC conference room 874/1-011
Persons
present
Minutes of the Previous Meeting and Matters Arising
There were no comments on the
minutes of the 39th meeting.
Helmut recalled that a Workshop on
Experimental Conditions and Beam Induced Detector Backgrounds, will take place at CERN from 3 to 4 April 2008. He
invited all LHCCWG members to actively take part in this event, where in
particular the experience at Tevatron, RHIC and HERA
will be discussed in view of the LHC, following a recommendation from the LHC
MAC.
Roger quickly reviewed the agenda
of the upcoming LHCCWG meetings. He announced that, since on 26 February Steve
Meyers will make a general presentation to the AB Department, the LHCCWG
meeting on that day will be held short, and will only look at the results from
the ongoing dry runs. At the following meeting, on 11 March, Werner will
discuss the maximum tolerable strength of the LHCb
spectrometer at injection, Philippe will review the bunch numbering, and Mike
will report on the distribution of bunch information to the experiments. Roger
also announced the Extended LTC days scheduled in the first week of March.
Bunch Filling
Schemes: Experiments Desiderata (Massimiliano
)
Massimiliano presented the luminosity and bunch-filling
demands from the experiments according to his present understanding. He started
with a reminder of LEP, which accommodated 4 symmetrically placed experiments.
At LHC, not only is the fourfold symmetry broken, with experiments in points 1,
2, 5 and 8, but in addition LHCb is longitudinally
displaced by 1.5 bunch spacings.
ATLAS
and CMS want to receive maximum integrated luminosity, with perhaps some
nuances. Several questions are open, but initially ATLAS and CMS prefer a
reduced pile up for equal luminosity and smooth stable running conditions.
The
LHC beam commissioning roadmap towards the nominal LHC had been determined in
view of ATLAS and CMS. Also, since these two experiments are located
symmetrically, ATLAS and CMS will get exactly the same the same bunch pairs
colliding in their respective IPs. The complexity arises from adding LHCb and ALICE.
ALICE
specialties include a Time Projection Chamber (TPC) with 90 microsecond drift
time. This TPC limits the useful luminosity to less than 3e30 cm-2s-1, while
the optimum luminosity is only 1e29 cm-2s-1. For the 25-ns nominal LHC scheme
providing the target luminosity in ALICE requires parallel separation of the
two beams, or defocusing, or both. The beam-gas background also imposes
stringent vacuum constraints. The multiplicity trigger of ALICE is based on a
silicon pixel detector, with signal integration over 100 ns (10 MHz clock). If
trying to reduce pile-up by multiplication of bunches, then the bunch
separations during commissioning should be larger than 100 ns for ALICE.
The
ALICE physics optimally done with early filling schemes includes the study of
high-multiplicity events which require low pile up. ALICE would like to discuss
with the accelerator side the flexibility in the individual bunch populations,
as well as all ways of getting low pp luminosity simultaneously with high
luminosity in ATLAS and CMS.
Massimiliano stressed that the optimum ALICE luminosity is
already reached with a single bunch of 7.5e10 protons at beta*=10 m. Poisson
statistics tells us the distribution of events. It might be better to
distribute the luminosity over more bunches, and keep their spacing larger than
100 ns.
For
LHCb not only the luminosity but also the Poisson
number matters. The latter is an important criterion for the separation of
different events and for suppressing events with nearby vertices. But LHCb can handle higher luminosities of up to 2-5e32
cm-2s-1, and if available it could make use of such luminosity immediately
(from day 1).
From
these considerations, Massimiliano now transited to
the early filling schemes. He started by considering 1 or 2 bunches per ring,
and afterwards addressed the optimized 43x43 scheme, the optimized 50-ns case,
and briefly the LHC upgrade.
The
minimum number of bunches per ring should not be one but rather two to have
collisions in LHCb and/or ALICE in addition to CMS
and ATLAS. One of the two bunches should be displaced, at least in one ring, to
ensure collision at the LHCb IP. Massimiliano
sketched different possibilities with or without parasitic collisions in IP1 or
IP5 (the parasitic collisions could be good or bad). The 43x43 bunch scheme is
one of the LHC running scenarios presented in LHC-Project-Note-323_Revised (“Standard
Filling Schemes for Various LHC Operation Modes,” by P. Collier and R. Bailey).
The
“perfect” 43x43 bunch scheme is based on a longitudinal shift of selected
bunches first described in LHC Project Note 401 (“Alternative Bunch
Filling Schemes for the LHC,” by G. Arduini, W. Herr,
E. Metral, and T. Pieloni).
The underlying idea is to give more collisions to LHCb than to ALICE by a symmetric shift of bunches in every
second quarter of the ring while keeping the maximum number of colliding pairs
in ATLAS an CMS.
Massimiliano highlighted that contrary to the common belief
it is possible to design a 50 ns scheme which makes everybody happy. The trick
is to start from the ideal 25-ns scheme and then remove all “even” bunches from
both beams. The 50-ns scheme offers several important advantages: (1) Only 3 IPs give full head-on collisions, instead of 4. (2)
It requires no separation or beta* increase in ALICE. (3) With known current
limitations, it may give a higher luminosity than either the 75-ns scheme or
the 25-ns scheme.
Finally,
looking at the upgrade, Massimiliano commented on the
“small-large” 25-ns scheme for LHCb. This is not a
problem for the proposed 50-ns filling scheme, since one can always fulfill the
requirement by selecting the right parity of the filling.
Massimiliano concluded that the great flexibility foreseen
for the LHC filling should allow many ways to accommodate the various
experimental desiderata and to satisfy all experiments simultaneously. If the
proposed ideas are correct, they should be approved by the LTC, documented and
advertised at the LPC.
Werner
presented the details of the various bunch filling schemes. Issues include
luminosity, experimental collisions, and beam-beam effects, other collective
effects, e.g. electron cloud, diagnostics, etc. He focused his presentation on
the first three of these issues. However, his presentation had been checked
with Gianluigi and Elias, who confirmed that the
assumptions made on the injector chain were reasonable.
As
already indicated by Massimiliano, an important input
is that ALICE requires 5 orders of magnitude less luminosity than the nominal
luminosity for ATLAS and CMS.
Filling
schemes must also be optimized from the point of view of beam-beam effects.
Here, the aim is to minimize the bunch-to-bunch variation of orbit, tunes, etc.
In the LHC trains of 2, 3, or 4 PS batches are transferred from the SPS to the
LHC. The LHC nominal filling scheme is achieved with 12 injections from the SPS
per beam.
Werner
stressed that the LHC is the first hadron collider
operating with many bunches. The large bunch number dominates all the beam-beam
effects. As a consequence, the exact collision schedule is needed for all
beam-beam related studies. Self-consistent beam-beam and luminosity
computations are employed to predict the orbits, tunes, and luminosity for each
individual bunch or bunch-pair, as well as the measurement response to an
excitation. Missing bunches, bunch-to-bunch fluctuations in intensity or emittance etc are also important elements to be considered.
An appropriate flexible description of the filling scheme is therefore needed,
which fulfills all requirements. The description chosen is based on the 3564
slot numbers. Bunches are numbered according to their slot positions, providing
an easy way to define the two beams. In particular, Werner emphasized that in
the adopted description the two beams can be completely different, e.g. it is
possible to consider 1 bunch in one ring and 2808 bunches in the other.
Werner
warned that, concerning the collision schedules, one should watch out for any
bunch spacing other than 25 ns spacing. In the old days the specifications for
IP2 were different. With 50-ns spacing one must verify good collisions in IP2
and IP8. With 156 bunches the two beams are well separated without a crossing
angle. The nominal filling scheme gives 2808 collisions in IP1 and 5, 2736 in
IP2, and 2622 in IP8 (where bunches at the start and end of each train do not
collide due to the shift of the IP).
With
43 (44) equidistant bunches, there would be no collisions in LHCb. A certain number of bunches can be displaced,
however. By virtue of this displacement it is possible to achieve e.g. 11-19
collisions in LHCb while retaining 21-4 collisions in
ALICE.
Bunches
can be displaced either in one beam only or in both beams symmetrically. The
basic assumptions for the injectors are that we can shift the PS to SPS
injection (1 batch), and/or the SPS to LHC injection for 2, 3 or 4 batches, and
that we can replace the SPS to LHC injection by a single bunch. In the
following, Werner only invoked the last two assumptions, because he received
the information that these two components are reasonably easy to implement.
The
luminosity can be flexibly shared between IP2 and IP8 by adjusting the filling
scheme. For example the luminosity with 156 bunches can be increased in IP8 and
the one in IP2 reduced, again by shifting some of the SPS injections. Two
options were presented.
A
50-ns spacing scheme had already been looked at 20 years ago, but it had been
given up later in favor of 25 ns. An advantage of the 50 ns spacing is that it
promises a high luminosity with much fewer long-range collisions. Selected
trains (SPS/LHC transfers) can be shifted by 1 slot to achieve the desired
luminosity sharing between IP2 and IP8. Werner now presented 5 LHCb collision options based on 50-ns spacing. As his
favorite he recommended scheme “e”, where only 2 bunches collide in ALICE with
almost full nominal luminosity in the other three IPs, in view of the very low
luminosity at ALICE. The scheme requires replacing the first transfer to LHC by
a single bunch. Werner qualified this “modified 50 ns” scheme as a good
alternative to the old 75 ns scenario.
Jean-Jacques
asked if the ‘small-large’ scheme, mentioned by Masimiliano
for sLHC, could become an early scenario because beam
instrumentation is not adapted for this. Werner and Paul replied no.
Roger
asked for Werner’s preference among the early scenarios. Werner’s preference
was 43 bunches to start with, of which 19 bunches would be shifted for LHCb. Massimiliano inquired
whether the 4 bunches colliding in IP2 would be consecutive bunches or not.
Werner answered that they would be at a large enough separation, but not
equidistant. Roger remarked that bunches for IP 1 and 5 would also not be
perfectly equidistant. Werner agreed saying that this would never be the case.
Thomas commented that a large gap would be fully sufficient, and it did not need
to be equal.
Roger
asked for any other comment or any objection. Massimiliano
brought up a question from Alice, namely by how much one can vary the intensity
of a single bunch, e.g. whether one could have one or
two bunches with low intensity. Werner referred this question to the injectors.
Massimiliano added a second related question about the
smallest number of bunches which could have a different bunch charge. Gianluigi responded that one could set up two different
users with different intensities to accommodate any wish, and that for example
3 out of 4 bunches in one injection could have the
same charge. The details would depend on how this is programmed. Massimiliano rephrased his question as “what is the minimum
charge in a bunch that can be used together with a high charge of 5e10 or
1e11?” Gianluigi confirmed that 1e10 for a simultaneous
weak bunch should be fine. However, he cautioned that it had never been tried
to operate with 1e11 and 5e9 bunch intensities on the same cycle. In
particular, Gianluigi observed that low intensity
bunches might have a different RF phase with respect to higher
intensity bunch and this could lead to longitudinal blow-up and/or poor
capture.
Next
Massimiliano commented that it is not quite clear to
what extent ALICE will prefer pile up spread out over many collisions rather
than pile up in a single collision. This depends on the multiplicity spectrum
that they will find.
Paul
remarked that the operational complexity is increased by the special filling
schemes. Massimiliano pointed out that on the
positive side these filling schemes may save time on other efforts, e.g. by
avoiding the need of defocusing or of implementing halo collisions in IP2.
Gianluigi noted that nothing has been done to prevent 50-ns
spacing. He had checked this with the rf
group. Roland’s presentation at Chamonix 2003 described the implementation and
experience with 50 ns operation in 2002. Elias, Gianluigi,
and Paul discussed that the 75-ns schemes misses the triple splitting in the
PS, while for the 50-ns scheme one of the double splittings
is left out. The reported first reaction from Steve Hancock was that the 50-ns
spacing can be produced again. Gianluigi observed
that the mechanics of the injectors must still be checked for the more
complicated filling patterns.
Jean-Jacques
and Massimiliano discussed the issue of the two
dynamic BPM ranges, which could complicate the simultaneous operation with two different
bunch intensities. Rhodri confirmed that BPM would not see pilot bunches mixed
within nominal ones. No other specific issue for BI could be identified.
Helmut
commented that the background in ALICE could be a concern, since the signal to
noise ratio will be much worse than for the other experiments. Roger and Massimiliano observed that this fact was almost independent
of the filling scheme. Helmut speculated that there could be some possibilities
to improve the signal-to-noise ratio via the timing. Massimiliano
commented that, yes, he indeed believed that ALICE uses timing correlations to
suppress beam-gas events.
Gianluigi commented that we might expect a different
lifetime for bunches with varying intensity. Paul added that the bunch emittances will also be different, which could have
implications for collimation. Gianluigi warned that
the emittance for 50-ns spacing could be different
from, smaller than, nominal. Also the 75-ns bunches have a much smaller emittance than the 25-ns beam.
Alick asked if there were other motivations for the choice
of bunch spacings. Roger recalled that the two main
motivations had been the electron cloud and the desire to maximize the
luminosity at constant bunch charge or at constant beam current. Ralph agreed
with Massimiliano that we must go to higher bunch
charge and larger spacing than 25 ns to reach maximum luminosity. Massimiliano
re-emphasized that the 50-ns spacing is the spacing for which one might hope to
get the highest luminosity given the maximum permitted total beam current.
In summary, no reason was found why the proposed 50-ns scheme should not work.
Machine development studies should be performed early in 2008 to reproduce the
50 ns beam and to study the possibility of injecting bunches with different
intensity in the SPS for the 43 bunch scheme.
Verena
presented a quick overview of what is planned for the LBDS dry runs. All
relevant information is posted at http://wikis/display/LHCOP/LBDS+Dry+Runs .
The
present week (week no. 7) foresees tests of point 6 BI devices in XPOC
acquisition mode, and the running of the LBDS equipment from CCC. In the next
week (8) the sequence should be (almost) ready for a reliability run from CCC. Further LBDS dry runs are coming up in weeks 11 (programmed dump),
15, and 19.
Verena reviewed the signals and the sequence of events for
a programmed dump. The energy tracking
system BETS normally receives the beam energy from 4 arcs to define kicker
voltage. The sequence of events for a programmed dump consists of 7 steps. Verena explained that the scenario for week 7 is different
from the nominal scheme: The beam dump is triggered by the timing system
directly instead of via the open permit loop. The “beam dumped event” is sent
out by operations and not automatically by timing. The BI acquisition is
triggered by this “beam dumped” event, for the XPOC analysis. And the energy
information comes from a “BETS simulator” in 4 arcs and not through the RB
DCCTs.
Items
which are tested in this dry run comprise: control of LBDS equipment from CCC,
arming of LBDS through the sequencer, BI devices in point 6, and the XPOC
acquisition. The equipment already checked includes most of the BI instruments.
Verena mentioned that the sequencer will be used to
run the LBDS through a cycle. The sequencer will be essential for the
reliability run starting soon. First conclusions will be presented after the
dry run.
Jean-Jacques
asked if the BETS signal is also provided to the BLMs via the timing system. Verena replied that this signal is not yet available, but
it could perhaps be made available for week 11 if BI ask
for this (which according to Jean-Jacques they will to verify the BLM threshold
selection). Brennan commented that BT will check whether it is possible that
the BETS simulator energy be distributed via the SLP system.
Mike
announced that his team got its paws on some of the magnets in Sector 45. The
matching section with magnets D4, Q5, Q6 etc had been done on Monday. The
region left of IP5 was being worked on. A full blown test was planned for
Thursday, including a ramp to maximum energy. The dry run would be terminated
by the scheduled warm up on the following Monday morning. In detail, the ongoing tests address LSA,
timing, pre-cycles, the parabolic ramp, etc. The realistic transfer functions
and coefficients from FiDeL were downloaded into LSA.
As for the LBDS dry run, conclusions would be reported in a couple of weeks.
Brennan
asked whether the overall ramp would be limited by the commissioned current of
the weakest circuit and to what level. Concerning an observed discrepancy
between the target current numbers and the maximum excitation for a limited
number of circuits, Massimo remarked that this problem might be resolved by
updating the transfer functions. Walter commented on the plans for the magnet
powering tests. Paul remarked that the recommendation is not to quench.
Tuesday
February 26th, 14:00
CCC
conference room 874/1-011
Provisional
agenda
Minutes
of previous meeting
Matters
arising
Summary of sector 45 operational tests
(Mike or Reyes)
Summary of LBDS dry run (Verena)
AOB
Reported
by Frank