DRAFT R. Tom as
Transcript of DRAFT R. Tom as
Optics correction strategy, cycle optimization
and implications for power converter and
magnetic measurements performance
DRAFT R. Tomas ...
January 16, 2018
The source HiLumi reports
CER
N-A
CC
-201
7-01
0112
/12/
2017
CERN-ACC-2017-0101December 12, 2017
Beam dynamics requirements for HL–LHC electrical circuits
D. Gamba, G. Arduini, M. Cerqueira Bastos, J. Coello De Portugal, R. De Maria, M. Giovannozzi, M.Martino, R. Tomas GarciaCERN, CH-1211 Geneva 23, Switzerland
AbstractA certain number of LHC magnets and relative electrical circuits will be re-placed for the HL-LHC upgrade. The performance of the new circuits willneed to be compatible with the current installation, and to provide the neces-sary improvements to meet the tight requirements of the new operational sce-nario. This document summarises the present knowledge of the performanceand use of the LHC circuits and, based on this and on the new optics require-ments, provides the necessary specifications for the new HL-LHC electricalcircuits.
KeywordsLHC, HL–LHC, circuit specifications, power converters
CERN-ACC-2017-0088. [email protected]
Optics Measurement and CorrectionChallenges for the HL-LHC
F. Carlier, J. Coello, S. Fartoukh, E. Fol, D. Gamba,A. García-Tabares, M. Giovannozzi, M. Hofer, A. Langner,
E.H. Maclean, L. Malina, L. Medina, T.H.B. Persson,P. Skowronski, R. Tomás, F. Van der Veken
and A. Wegscheider.CERN, Geneva, Switzerland.
Abstract
Optics control in the HL-LHC will be challenged by a very small β ∗ of 15 cmin the two main experiments. HL-LHC physics fills will keep a constant luminosityduring several hours via β ∗ leveling. This will require the commissioning of a largenumber of optical configurations, further challenging the efficiency of the opticsmeasurement and correction tools. We report on the achieved level of optics controlin the LHC with simulations and extrapolations for the HL-LHC.
Geneva, SwitzerlandNovember 10, 2017
Triplet trim circuits news
CSchem
eofthe
newH
L–L
HC
circuits
Q1MQXFA
Q3MQXFA
Q2aMQXFB
Q2bMQXFB
OC (Q1)MCBXFB
OC (Q2)MCBXFB
OC (Q3)MCBXFA
SF HOMQSXF (200A)
MCSXF/MCSSXF (2x120A)MCOXF/MCOSXF (2x120A)MCDXF/MCDSXF (2x120A)MCTXF/MCTSXF (2x120A)
D1MBXF
D2MBRD
OC (D2)MCBRD
Q4MQY
Q5MQY
Q6MQML
OC (Q4)MCBY
OC (Q5)MCBY
OC (Q6)MCBC
QP: OL QHs + IL QHs + CLIQ
± 2 kA
QP
: QH
s + EE (Op
tion
)
QP: QHs
± 600 A
x4
± 120 A
x8(x6 via DFBL – x2 Local)
x6 Local x2 Local
QP
: PC
Cro
wb
ar
QP
: PC
Cro
wb
ar + EE (O
ptio
n)
± 200 A
± 120 A
x8
QP
: PC
Cro
wb
ar
QP
: PC
Cro
wb
ar + EE (Op
tion
)
6 kA
Cold powering via DFHX2 leads 18 kA2 leads 13 kA3 leads 7 kA 5 MIITS12 leads 2 kA2 leads 0.2 kA
16 leads 0.12 kA
Cold powering of Q4/Q5/Q6 via DFBL+DSLCold powering for correctors: Q4 MCBYs powered via DFBL+DSL (x6) and
locally (x2) Q5 and Q6 MCBYs powered locally9 leads 6 kA (+2 spares)
32 leads 0.12 kA
QP: QHs
6 kA
Cold Powering via DFHM2 leads 13 kA
8 leads 0.6 kA
x2
x2
x2
13 kA
DFHX DFHX
DFHX DFHX
DFHX DFHX
DFHX DFHX
QP: QHs
13 kA
DFHM DFHM
DFHM DFHM
DFBL DFBL DFBL
DFBL/Local
DFBL/Local
QP: Quench ProtectionQHs: Quench HeatersEES: Energy Extraction SystemPC: Power ConverterOC: Orbit CorrectorsxN: Number of Circuits per IP Side
6 kA 6 kA
DFBL DFBL DFBL
x1
x1x1 x1
x1
x1 x1
QP: QHs
Magnet Layout ( Crab Cavities Not Shown) Right of Point 5
CP
± 250 A
Local Trim
Local Trim
MBH (11T)
In series with the RB circuit
11 T Dipoles – Right and Left
of Point 7
Circuits Layout Version 2.1
18 kA
Q1 Q2a Q2b Q3
± 2 kA ± 2 kA
CC+
+ C+ C+ CC+
+
P1 P4 P2 P3 P1 P4 P2 P3 P1 P4 P2 P3P3 P2 P4 P1 P3 P2 P4 P1 P3 P2 P4 P1
± 35 A
DFHX DFHX DFHX DFHX DFHX Local Local
Fig.C.1:L
ayoutofthenew
HL
–LH
Ccircuits,V
ersion2.1
[40].
41
F New Q1A trim circuit of ±35A added for k-modulation:critical for accurate β∗ control.
F Q2A trim removed: Q2A/Q2B TF relative difference minimizedvia magnetic measurements and sorting.
Orbit drift from power converter noise
0
1
2
3
4
5
6
7
|x|y
IP
| [
10
-3/p
pm
]
Horizontal IP1
Vertical IP1
Horizontal IP5
Vertical IP5
RB
.A12
RB
.A23
RB
.A34
RB
.A45
RB
.A56
RB
.A67
RB
.A78
RB
.A81
RQ
F.A
12
RQ
D.A
12
RQ
D.A
23
RQ
F.A
23
RQ
F.A
34
RQ
D.A
34
RQ
D.A
45
RQ
F.A
45
RQ
F.A
56
RQ
D.A
56
RQ
D.A
67
RQ
F.A
67
RQ
F.A
78
RQ
D.A
78
RQ
D.A
81
RQ
F.A
81
RT
B8
RQ
X.L
1R
QX
.R1
RT
QX
A1.L
1R
TQ
XA
1.R
1R
TQ
X1.L
1R
TQ
X1.R
1R
TQ
X3.L
1R
TQ
X3.R
1R
D1.L
1R
D1.R
1R
D2.L
1R
D2.R
1R
Q4L1
RQ
4R
1R
Q5.L
1R
Q5.R
1R
Q6.L
1R
Q6.R
1R
CB
X1.L
1R
CB
X1.R
1R
CB
X2.L
1R
CB
X2.R
1R
CB
X3.L
1R
CB
X3.R
1R
CB
RD
4.L
1R
CB
RD
4.R
1R
CB
Y4.L
1R
CB
Y4.R
1R
CB
Y5.L
1R
CB
Y5.R
1R
CB
C6.L
1R
CB
C6.R
1
-10
-5
0
5
x|y
[
10
-3/p
pm
]
Horizontal (collimators)
Vertical (collimators)
Fig. 13: Beam 1 orbit variation at IP1 and IP5 (top) and at the primary collimators (bottom) underthe effect of one ppm error on each relevant circuit of HL-LHC with nominal 15 cm β∗ round optics(HLLHCV1.3). In the bottom graph the thicker error bars give the mean and r.m.s. values, while thethinner error bars indicate the maximum excursion over all primary and secondary collimators. Allvalues are normalised with respect to the local beam size.
D1 and D2. Their impact on the orbit is higher than the impact of the triplets5. The perturbation inducedby the other orbit correctors in the IR is about about 5 to 10 times smaller, while the impact of the 11 Tdipoles trim (RTB8) seems to be negligible. Note that the impact of the main bend and of some mainquadrupole PCs is also significant. Still, all contribution seems to be below one percent of beam sigmaboth at the IPs and at collimators.
During LHC operations it was observed an IP orbit stability during a fill of about half a beamsigma [48]. It is therefore expected that the PCs slow uncertainties have a negligible impact on orbitstability with respect, for example, to ground motion. For frequencies above a few tenths of Hz the orbitstability might play an important role due to beam-beam effects. Dedicated studies will need to be madeto quantify those effects. For the time being, it is recommended to adopt the same class of PC for RQXand RD circuits. Also note that the expected inductance of the RD1 and RD2 circuits is approximatelyone order of magnitude lower that the RQX circuits [3], therefore, according to Eq. (24), the RD1 andRD2 circuits will be about 10 times more sensitive to PC voltage ripple than the RQX circuits.
4.3 Beta beatingFigure 14 shows the impact of one ppm error for the main HL-LHC circuits on beta beating at IP1 andIP5 and along the whole machine for the 15 cm β∗ optics. The behaviour, as expected, reflects the impacton tune shown in Fig. 12b, and the main contributors are the main dipole of the ATS arcs and the maintriplet circuit PCs. The amplitude of each contribution (less than 6×10−4 beta beating per ppm) togetherwith the typically small “stability during a fill” values of the PCs (a few ppm, see Table 6) suggests anegligible impact on luminosity imbalance between IP1 and IP5 during a fill. However, this can becomerelevant on a longer time scale due to the drift accumulated by the PCs, i.e. due to their “long term fill-
5The effect on orbit triggered by an error on the triplet is generated by the crossing scheme.
33
����
Table 24: Expected HL-LHC beam 1 orbit stability at IP1/5 and collimators for 15 cm β∗ optics (HLL-HCV1.3) on a short time scale (i.e. assuming the “short term stability (20 min)”) and on a longer timescale (i.e. assuming “fill-to-fill repeatability”). The values are normalised with respect to the local beamsize assuming 7 TeV operation and 2.5 µm normalised emittance and 1.08 × 10−4 energy spread. Inparenthesis the values assuming to also upgrade the PC of main bends in the ATS arcs to class 0.
Orbit jitter [10−3 σbeam r.m.s.]Short time scale Long time scale
IP1/5 (x) 2.9 (2.8) 6.5 (6.2)IP1/5 (y) 2.2 (2.1) 4.7 (4.3)Collimators (x) 2.2 (2.2) 5.2 (5.1)Collimators (y) 1.7 (1.6) 3.5 (3.3)
Table 25: Worst HL-LHC beam 1 orbit and beta beating drift at IP1/5 for 15 cm β∗ optics (HLLHCV1.3)on a year time scale (i.e. assuming the “long term fill-to-fill stability”). The orbit values are normalisedwith respect to the local beam size assuming 7 TeV operation and 2.5 µm normalised emittance and1.08 × 10−4 energy spread. In parenthesis are the values obtained assuming a factor 2 improvement onmain dipole and quadrupole PCs as well as on D1/D2 and main triplet PCs, corresponding to performinga calibration of those PCs every 6 months.
Max |orbit drift| [σbeam] Max |∆β∗/β∗0|
IP1/5 (x) 0.9 (0.7) 0.04 (0.02)IP1/5 (y) 0.6 (0.5) 0.03 (0.02)
a luminosity imbalance between IP1 and IP5 becomes observable along the year, it will be possible toperform either a dedicated calibration of the relevant PCs6 during an HL-LHC Technical Stop (TS), oran additional optics measurement and correction.
4.5 Dynamic aperture perturbationThe main concerns for Dynamic Aperture (DA) are the high-frequency voltage tones. The voltage spec-trum tones assumed in [1], Chapter 2.3.2, are reported in Table 26. Those values are also compatible
Table 26: Main PC voltage spectrum tones assumed in [1, 49].
Frequency Amplitude [mV] r.m.s.50 Hz 3.2100 Hz 0.8300 Hz, 20 kHz 10600 Hz, 40 kHz 2.510 MHz 1others 0.5
with the CERN custom acceptance levels (see Fig. 6) and no updated values are available, yet.6This is possible also for PCs not equipment with a remote controlled calibration unit. In this case one needs about half a
day per PC.
36
rem
ote
cal→
no
rem
ote
cal→
Remote calibration systems for D1&D2 PCscost 200 kchf but it is judged unnecessary
Tune ripple from power converter noise
RB
.A1
2R
B.A
23
RB
.A3
4R
B.A
45
RB
.A5
6R
B.A
67
RB
.A7
8R
B.A
81
RQ
F.A
12
RQ
D.A
12
RQ
D.A
23
RQ
F.A
23
RQ
F.A
34
RQ
D.A
34
RQ
D.A
45
RQ
F.A
45
RQ
F.A
56
RQ
D.A
56
RQ
D.A
67
RQ
F.A
67
RQ
F.A
78
RQ
D.A
78
RQ
D.A
81
RQ
F.A
81
RQ
X.L
1R
QX
.R1
RT
QX
1.L
1R
TQ
X1
.R1
RT
QX
2.L
1R
TQ
X2
.R1
RQ
4.L
1R
Q4
.R1
RQ
5.L
1R
Q5
.R1
RQ
6.L
1R
Q6
.R10
0.5
1
1.5
2
2.5
3
|Q
| [1
/pp
m]
10-5
Horizontal
Vertical
(a)
RB
.A1
2R
B.A
23
RB
.A3
4R
B.A
45
RB
.A5
6R
B.A
67
RB
.A7
8R
B.A
81
RQ
F.A
12
RQ
D.A
12
RQ
D.A
23
RQ
F.A
23
RQ
F.A
34
RQ
D.A
34
RQ
D.A
45
RQ
F.A
45
RQ
F.A
56
RQ
D.A
56
RQ
D.A
67
RQ
F.A
67
RQ
F.A
78
RQ
D.A
78
RQ
D.A
81
RQ
F.A
81
RT
B8
RQ
X.L
1R
QX
.R1
RT
QX
A1
.L1
RT
QX
A1
.R1
RT
QX
1.L
1R
TQ
X1
.R1
RT
QX
3.L
1R
TQ
X3
.R1
RD
1.L
1R
D1
.R1
RD
2.L
1R
D2
.R1
RQ
4L
1R
Q4
R1
RQ
5.L
1R
Q5
.R1
RQ
6.L
1R
Q6
.R10
0.2
0.4
0.6
0.8
1
|Q
| [1
/pp
m]
10-4
Horizontal
Vertical
(b)
Fig. 12: Variation of Beam 1 tune per ppm of current variation (with respect to Irated) for each of themain circuits of LHC with nominal 40 cm β∗ round optics (a) and for HL-LHC with nominal 15 cm β∗
round optics (b). The difference in between RB circuits for HL–LHC is due to the features of the ATSoptics.
30
� �
F Increased β-functions in the ATS arcs magnifies power converternoise, increasing tune ripple.
F The 4 ATS dipole PCs are proposed to be upgraded to class 0 toreduce tune jitter from 4.1×10−5 to 2.7×10−5.
Upgrade of 4 dipole PCs to class 0
'
&
$
%Cost: 600 kchf
Turn-around-Time
Phase Time [minutes]Old baseline New baseline
Nominal (Ultimate)
Ramp-down 60 40Set-up, injection 55 65Ramp & Squeeze 25 25Flat-top, Squeeze 30 5 (10)Adjust/collide 10 10TOTAL 180 145 (150)
Faster ramp-down and Ramp & Squeeze have considerably reducedturn-around-time.
Further improving turn-around-time?
LHC current ramp-down
Fig. 2: Typical BEAMDUMP, RAMPDOWN and the following SETUP sequence for a few main mag-nets in LHC. The top figure is the measured current in the circuits as a function of time. The middle andbottom figures are the first and second time derivatives numerically computed, respectively. The verticaldashed lines delimit the different beam processes.
2.2.3 Maximum ramp and acceleration ratesThe maximum and minimum values for current and ramp rates observed during fill #5848 are sum-marised in Table 3. The values are divided between processes with beam (INJPROT, INJPHYS, PRE-RAMP, RAMP, FLATTOP, SQUEEZE, ADJUST, STABLE) and without beam (BEAMDUMP, RAM-PDOWN, SETUP). A comparison of the different processes divided by circuit type is also available inAppendix A. Note that the sampling frequency of the data used for the analysis is 2 Hz, therefore fasterchanges are not detectable here.
Table 3 shows that the maximum and minimum ramp rates are exploited during the SETUP andRAMPDOWN processes. The energy ramp and the optics gymnastic during the squeeze also require fastramp-up rates. The speed of the D1 and D2 dipoles, as well as the corrector circuits, is also dominated bythe SETUP and RAMPDOWN processes when degauss cycles are performed for some of these circuits.
Figure 4 shows the use of the orbit correctors during the beginning of the analysed fill. For thewhole duration of the beam cycle the LHC orbit feedback [12] is active in order to keep the orbit close tothe “golden” orbit defined by the operators. Due to the tight constraint imposed by the quench protectionsystem, the RCBX correctors are not used by the orbit feedback in order to avoid undesired beam dumpstriggered by the Quench Protection System (QPS) [13]. During STABLE-beam operations the orbitcorrectors are also used for luminosity optimisation. The histograms of orbit corrector current deviation
7
Main dipoles
Triplet quads
Fig. 2: Typical BEAMDUMP, RAMPDOWN and the following SETUP sequence for a few main mag-nets in LHC. The top figure is the measured current in the circuits as a function of time. The middle andbottom figures are the first and second time derivatives numerically computed, respectively. The verticaldashed lines delimit the different beam processes.
2.2.3 Maximum ramp and acceleration ratesThe maximum and minimum values for current and ramp rates observed during fill #5848 are sum-marised in Table 3. The values are divided between processes with beam (INJPROT, INJPHYS, PRE-RAMP, RAMP, FLATTOP, SQUEEZE, ADJUST, STABLE) and without beam (BEAMDUMP, RAM-PDOWN, SETUP). A comparison of the different processes divided by circuit type is also available inAppendix A. Note that the sampling frequency of the data used for the analysis is 2 Hz, therefore fasterchanges are not detectable here.
Table 3 shows that the maximum and minimum ramp rates are exploited during the SETUP andRAMPDOWN processes. The energy ramp and the optics gymnastic during the squeeze also require fastramp-up rates. The speed of the D1 and D2 dipoles, as well as the corrector circuits, is also dominated bythe SETUP and RAMPDOWN processes when degauss cycles are performed for some of these circuits.
Figure 4 shows the use of the orbit correctors during the beginning of the analysed fill. For thewhole duration of the beam cycle the LHC orbit feedback [12] is active in order to keep the orbit close tothe “golden” orbit defined by the operators. Due to the tight constraint imposed by the quench protectionsystem, the RCBX correctors are not used by the orbit feedback in order to avoid undesired beam dumpstriggered by the Quench Protection System (QPS) [13]. During STABLE-beam operations the orbitcorrectors are also used for luminosity optimisation. The histograms of orbit corrector current deviation
7
In HL-LHC upgrading IR2 and IR8 triplet PCs could reduce TaT by15 minutes, increasing integrated lumi by 2-3%.cost? diode option cheap: 80 kchf?
Challenges for optics control in HL
F β∗ leveling: ≈50 optics need fine commissioning
F Arc errors enhanced without local quads for correction.
F β∗ accuracy with k-modulation challenged by tune jitter
F HL-LHC non-linear magnetic errors affect: DA, Landaudamping, β∗ and coupling. All changing Vs crossing angle.Beam-based measurements are mandatory.
F We have no experience in correcting b5, b6 and a6.
β∗ leveling
0
25
50
75
100
125
0 2 4 6 8
Nominal
Ultimate
64
41
β*
at IP
1&
5 [cm
]
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0 2 4 6 8
Lum
inosity a
t IP
1&
5[1
034cm
-2s
-1]
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
0 2 4 6 8
Bunch inte
nsity [10
11]
Time [h]
0.5
1.0
1.5
2.0
2.5
3.0
0 2 4 6 8
ρ- = 0.80 mm-1
ρ- = 1.20 mm-1
Peak lin
e p
ile-u
p d
ensity
at IP
1&
5 [m
m-1
]
Time [h]
#bunches=2748
Xsing=500 µrad
ε = 2.5µm
HL-LHC arc errors correction simulation
050
100150200250
coun
t
horizontal HL-LHC β∗ = 15cmLHC β∗ = 60cm
0.0 0.1 0.2 0.3 0.4 0.5Maximum ∆β/β
050
100150200250
coun
t
vertical
With current tools we expect 10-20% β-beating in HL-LHC.Collimation & β∗-reach request 5% (as LHC).
Flat and round ATS optics MDs (βarc×4)
0.2
0.1
0.0
0.1
0.2
0.3∆βx/β
x
5000 10000 15000 20000 25000Longitudinal location [m]
0.05
0.00
0.05
∆βy/β
y
IP1IP2 IP3 IP4 IP5 IP6 IP7 IP8
����
LHCB1 β∗x/y = 15/60cm @ IP5
IP2 IP3 IP4 IP5 IP6 IP7 IP8 IP1s [m]
−20
0
20
∆βx/β
x[%
]
LHCB1 β∗ = 10 cm
N-BPM����
∆β/β not under control for ATS large βarc
Measured tune jitter in MDs
Courtesy: Sergey Antipov
What is this 100s oscillation? How large will it be in HL-LHC? Itcould impair β∗ measurements with K-modulation.
Measured tune jitter in MDs
0 1 2 3 4Simulation
0
1
2
3
4
5
6M
easu
rem
ent
Tune jitter [10 5]
bal
listi
c
β∗ =
40cm
β∗ =
30cm
β∗ =
25cm
β∗ =
60/1
5cm
β∗ accuracy, K-modulation and tune jitter
F ATLAS/CMS Lumi imbalance should be below 5%
F From power supply ripple in current baseline we expect:
tune jitter=4.1×10−5 → β∗ accuracy=10% → Lumi imb.≈20%
F If we upgrade 4 arc dipole PCs to class 0:
tune jitter=2.7×10−5 → β∗ accuracy=6% → Lumi imb.≈12%
F Further noise reduction techniques, statistics would still berequired to achieve the 5% goal in lumi.
Non-linear errors: Landau damping (in LHC)
What you want
What you get withoutIR non-lin corr.
Non-linear correction is critical for Landau damping.
Non-linear errors: Landau damping and
skew octupoles (a4)
0.320
0.322
0.314 0.316 0.318
Qy
Qx
Increasing Jx
Increasing Jy
|C-|=0.000000 Qx − Qy = 0
IR errors: model vs measurements a3 & b3
-1
0
1
-1 0 1
b 3 R
1 [1
0-3 m
-3]
b3 L1 [10-3 m-3]
-1
0
1
-1 0 1
b 3 R
5 [1
0-3 m
-3]
b3 L5 [10-3 m-3]
Model corrections
Beam-based corrections
-10
-5
-10 -5
a 3 R
1 [1
0-3 m
-3]
a3 L1 [10-3 m-3]
IR errors: model vs measurements b4
-1.0
-0.5
1.0 1.5
b 4 R
1 [m
-4]
b4 L1 [m-4]
Model corrs
-1.0
-0.5
1.0 1.5
b 4 R
5 [m
-4]
b4 L5 [m-4]
Beam-based corrs
In LHC discrepancies between corrections from magneticmeasurements and from beam measurements can be very significant.Sources are: meas. uncertainties, orbit errors and magnetmisalignments.
(I have asked E. Todesco for uncertainties in magnetic multipolemeasurements).
Non-linear errors: Feeddown
HL-LHC WP2 meeting, 19th December 2017
Impact on linear optics can become considerably more serious forsmaller β∗
e.g. simulation studies of HL-LHC (15cm, 295µrad)
Machine ProtectionLimit
Peak Δβ/β
0
5
10
15
0 0.1 0.2Δβx/βx from sextupoles
Machine ProtectionLimit
0
2
4
6
8
0 0.01 0.02Δ|C-| from sextupoles
HL-LHC targetHL-LHC target
Also need to consider effect on linear coupling
Direct impact due to feed-down
Ability to measure
IRNon-linear errors plus crossing angle heavily affect linear optics.It might be more important to correct for feeddown than for DA!Strategy to be defined.
Non-linear errors: DA
F DA without non-linear correction is 5σ at β∗=15cm
F This challenges optics measurements which use ≈2σ oscillation
F Iterative corrections linear↔non-linear together with
F 1st guess from magnetic measurements will be critical
Accurate magnetic and alignment measurements arefundamental
F Ideal correction for DA gives 9σ
F What will be the DA value when correcting for feed-down?
cost / benefit
F
F
F
F
F
Possible AC dipole review in 2018
F AC dipole is fundamental for linear and non-linear opticscommissioning
F It is limited to 1 measurement per minute to allow for cool-down
F Tunes away injection/collision tunes requires intervention
F AC dip. amplifier breaks about once per year
F Review in 2018 to check possible improvements or upgrades
Back-up
Optics control: LHC Vs HL-LHCLHC HL-LHC
unit β ∗ = 40 cm β ∗ = 15 cmCMS/ATLAS luminosity imbalance [%] 5 5toleranceTune jitter (rms)
[10−5] 2-4 4.1
Assumed tune measurement uncertainty[10−5] 1.5 2.5
β ∗ accuracy:rms tolerance for lumi imbalance [%] 2 2rms achieved or expected [%] 1 4
Peak β -beating after correction [%] 5 10-20β -beating from crossing angle [%] 2 20(without non-linear IR correction)|C−|:
Tolerance for instabilities[10−3] 1 1.0
Tolerance for K-modulation[10−3] 1 0.6
7 month drift[10−3] 3 12
∆|C−| from crossing angle[10−3] 2 20
(without non-linear IR correction)Dynamic aperture:
Before IR correction [σ ] 10 5After IR correction [σ ] 12 9
Table 6: Tolerances and achieved or expected values for LHC and HL-LHCoptics control related parameters. Tune jitter values come from [16]. Theassumed tune jitter of 2.5×10−5 requires upgraded power supplies for thetelescopic arc dipoles. LHC DA values are taken from [84] and rescaled tothe HL-LHC emittance of 2.5 µm.
Both experiments and simulations suggest that peak β -beating will be about20% in HL-LHC, specially appearing in the arcs used for the telescopic squeeze.
The non-linear errors will pose severe challenges even for the linear opticscommissioning via their feed-down to β -beating and coupling and by reducingthe available DA for optics measurements with the AC dipole. Iterative correc-tions alternating the target between linear and non-linear orders will be required.A broad spectrum of techniques to measure and correct IR non-linear errors areemerging but a substantial effort is required to demonstrate their feasibility. Astrategy based on these techniques should be defined and verified with simula-tions of realistic scenarios for optics commissioning in HL-LHC.
50
Baseline: DA validation
62.30 62.31 62.32 62.33Qx
60.30
60.31
60.32
60.33Q y
4.0
4.0
5.0
HL1.3; I=1.2e11; * =15cm;Xing/2=250 rad; Q'=15; IMO=-300; Min DA.
3
4
5
6
7
8
9
DA [
beam
]
DA = 6σ in a small region close to Qx = Qy . Tune and couplingcontrol become critical. Further details in Nikos’ presentation.
IR non-linear correction
LHC IR non-linear correction at β∗ = 14 cm in ATS MD:
99.7
99.8
99.9
100.0
0 100 200
Su
rviv
ing
fra
ctio
na
l in
ten
sity
Time [s]
LHCB1
Before b4 correction After b4 correction
99.7
99.8
99.9
100.0
0 100 200
Su
rviv
ing
fra
ctio
na
l in
ten
sity
Time [s]
LHCB2
Before b4 correction After b4 correction
Figure 29: Surviving fractional intensity versus time, calculated from BCTdata. The fractional intensity is calculated from ∼ 2minutes prior to appli-cation of the b4 correction (blue), and for 2minutes from the time of MCOXtrim completion. The time period during which the b4 correction is beingapplied is ignored, as feed-down to tune causes transient losses.
beam. Furthermore the β ∗ imbalance generated by the IR sextupoles in mostcases is operationally intolerable.
Having the facility to compensate such errors will be essential for the HL-LHC, but may require a serious revision to the linear optics correction strat-egy. While application of nominal commissioning methods may be possible, inthe LHC linear optics has always been commissioned with flat-orbit and correc-tion with crossing scheme applied is entirely untested at low β . Furthermore, ifcrossing-angle bumps are to be varied during operation (to provide luminosity orpile-up leveling or to limit energy deposition in the triplets) changing feed-downwill dynamically alter the β ∗-imbalance during leveling unless local sextupolecorrections are implemented.
Feed-down to coupling also represents a significant challenge. Figure 31shows a histogram over the target error table seeds, of the linear coupling gen-erated by sextupole feed-down alone for β ∗ = 15 cm, 295 µrad. Feed-downfrom the nonlinear errors in the experimental IRs has the potential to generatevery large shifts to the linear coupling during the squeeze, up to 0.025. In theLHC |C−| ≈ 0.004 has been observed to cause instabilities and a tolerance of|C−| ≤ 0.001 is estimated for HL-LHC [17]. Therefore, correction of the IR cou-pling from feed-down during the squeeze will be essential. Further, allowing fora residual |C−| at the 10−3 level, the majority of seeds in Fig. 31 would gen-erate enough coupling to cause HL-LHC beams to become unstable under the∼ 60 µrad crossing-angle manipulations proposed for leveling during HL-LHC
36
F Losses without IR correction of 4%/h at β∗ = 14 cm.
F Lifetime recovered thanks to beam-based corrections
F HL-LHC has larger IR non-linear errors → Challenge ahead!
Concluding remarks
F New baseline scenario meets goals at 50% efficiency
Pushed: optics, collimation, impedance, beam-beam, DA, etc.New: Q1A trim, remote alignment, PC class 0, etc.
F A slightly flat optics increases performance by 2-4%
F The largest threat is e-cloud, 8b4e reduces performance by 25%
A mixed filling scheme 25ns/8b4e could mitigate loss
F Not having CCs would result in 7-10% lower luminosity with25% larger ρ