ILC - International Linear Collider

Ring to Main Linac


The ILC Ring to Main Linac (RTML) is the collection of beamlines which transfer the beam from the damping ring to the main linac on each side of the collider. In addition to transporting the beam between these two regions, the RTML beamlines perform all the manipulations in the beam conditions which are required to match the parameters of the beam extracted from the damping ring to the parameters required of the beam injected into the main linac. The parameter matching includes:

  • Completing the achromatic extraction of the beam from the damping ring
  • Moving the beam from the axis of the damping ring to the axis of the main linac (geometry match)
  • Collimation of beam halo generated in the damping ring
  • Transformation of the beam polarization direction from the direction required in the damping ring (nominally vertical) to the direction required by experimenters at the IP
  • compensation of the beam jitter introduced in the damping ring or during extraction from the damping ring
  • Compression of the bunch length from the equilibrium length of the damping ring to the shorter length required in the linac and at the IP.

In addition, the RTML must provide instrumentation and diagnostics sufficient to measure and control emittance growth, beam jitter amplification, spin dilution, and other beam quality reductions which would otherwise be introduced by the beamlines of the RTML; and the RTML must provide a set of dumps and stoppers which can stop the beam from entering downstream systems during beam tuning at upstream locations or during maintenance accesses at downstream locations. Last but not least, the RTML must be made as cost effective as possible within the constraints of achieving its technical goals.


A baseline configuration for the RTML has been selected.


Below is a description of the RTML, divided according to sub-beamlines in longitudinal (“S-position”) order. For most beamlines, only a qualitative description can be presented at this time, as the detailed design has not yet begun. The exception is the bunch compressor, which is in a relatively more advanced state of development.

Extraction Geometry and Betatron Match

This beamline completes the extraction of the beam from the damping ring, including closing the dispersion introduced by the extraction kickers, septa, and bend magnets; transfers the beam from the damping ring elevation to the main linac elevation, if the two are different (for example, if the damping ring is in a shallow tunnel and the linac is deep) by means of achromatic bends; and matches the betatron functions from the damping ring extraction to those required by the downstream emittance measurement system. The beam which is extracted from the damping ring is initially travelling in the direction opposite to the nominal linac direction; its direction of travel is reversed by the turnaround.

Emittance and Trajectory Measurement

This beamline uses a set of four orthonormal skew quadrupole magnets to couple all four xy coupling terms in the beam matrix; this allows coupling introduced by the process of beam extraction from the damping ring to be corrected globally. The coupling correction section is followed by an emittance measurement station, which checks to make sure that the emittance correction is properly performed. The current baseline calls for a system which can measure full beam matrix and extract the normal mode emittances and the coupling terms. The emittance measurement station must be capable of measuring emittances during multibunch operation, using many bunches to measure the emittance within 1 train, and must also be capable of measuring emittances during single-bunch operation, using many pulses (at 5 Hz) to complete one measurement. In the same beamline, a set of beam position monitors measures the bunch-by-bunch trajectory of the beam for feedforward correction. At the end of this beamline there is an insertable stopper which can be used to stop the beam from passing into the collimation section and other downstream areas. This stopper can probably only handle a fraction of the nominal beam power, and so can only be used for short trains, low repetition rates, or some combination of the two.

Transverse Collimation

This beamline uses a spoiler/absorber scheme to collimate halo particles which are generated in the damping ring. Collimation is 2 phases x 2 planes x 1 iteration, and both planes are nominally collimated at the same depth. The spoiler positions and apertures are adjustable; the absorbers may also need to be adjustable. The beam is sufficiently enlarged at the spoiler locations to prevent damage to the spoilers in the event of a direct hit from the beam core by a small number of bunches (and assuming that damping ring extraction is halted after that number of bunches); the spoilers, in turn, enlarge the beam core to prevent damage to the absorbers from a direct hit from several bunches of the beam.


This beamline reverses the direction of travel of the beam so that it is headed in the direction of the main linac. The purpose of the turnaround is to allow the bunch-by-bunch trajectory measurements to be fed forward over a shorter path length to the Trajectory Correction section. The exact configuration of the turnaround has not been selected and is likely at least partially site-dependent, but it is probably on the order of 170 meters in length and includes bending magnets, quadrupole magnets configured in a FODO lattice, and possibly sextupole magnets for matching and suppression of high-order dispersion. The turnaround will introduce emittance growth from synchrotron radiation: it is estimated that a “cat-bone” style turnaround (with a 90 degree bend followed by a 262 degree reverse bend) with 170 meter length can limit emittance growth to about 0.16 um, or about 2% of the emittance at extraction from the damping ring.

Spin Rotator

This beamline uses 4 strong solenoid magnets to allow the beam polarization vector to be set to any orientation desired by the experimenters. The first half of the system contains two solenoids which are powered in series and separated by an Emma rotator (a beamline which performs a +I transformation in the horizontal plane and a -I in the vertical), to allow the polarization to be adjusted without introducing coupling from the solenoids. This is followed by an achromatic arc of approximately 8 degrees, which completes the turnaround of the beam trajectory; the arc is followed by another pair of solenoids separated by an Emma reflector. The combination of the two solenoid pairs and the bending system allows the polarization to be pointed in any direction required by the experimenters.

Trajectory Correction

This beamline is a simple FODO array with 2 horizontal intra-train dipole correctors separated by 90 degrees in betatron phase, and 2 vertical intra-train correctors with the same phase separation. Bunch-by-bunch trajectory information is measured in the upstream emittance measurement section, and fed forward to this location to correct the beam jitter generated in the damping ring and during extraction (i.e., by jitter in the extraction kicker amplitude).

Emittance Measurement and Coupling Correction

This beamline is basically identical to the emittance measurement and coupling correction system which follows damping ring extraction.

First Stage Bunch Compressor

The first stage bunch compressor is divided into the following subregions:

  • An RF section which generates the necessary correlation between longitudinal position and energy. This section contains 32 9-cell RF cavities arranged in 4 cryomodules of 8 cavities each, based on the assumption that this is the cryomodule configuration which will be used for the ILC main linac. Because the bunch is long in this section, relatively strong focusing is used to limit the emittance growth from transverse wakefields: quad spacing is 1 quad per cryomodule, with 90 degree phase advance per cell in x and y. The cavities are phased near the zero-crossing (-100 degrees is typical), and require gradients of up to 18.4 MV/m. There are no spare modules in this section.
  • A wiggler based on 6 90 degree FODO cells with chicanes placed in the space between each pair of quads (12 chicanes total). Each chicane contains 8 bend magnets, with bends 1, 4, 5, and 8 in each chicane powered in series by one power supply and bends 2, 3, 6, and 7 in each chicane powered in series by a second power supply; this second power supply is varied to adjust the R56 of the system. The wiggler also contains magnets for tuning the horizontal and vertical dispersion, instrumentation for measuring beam energy, and adjustable energy collimators which can tolerate being struck by several bunches without being damaged.
  • A longitudinal diagnostics section, which permits measurement of the central energy, energy spread, arrival time, and bunch length. Some or all of this instrumentation may be in an extraction channel and not the beamline which leads to the second stage compressor.
  • A short region which extracts the beam from the straight-ahead channel to a tune-up dump. This region is equipped with pulsed bends (which can driven by a DC power supply or else pulsed to take bunch trains out to the tune-up dump) and also with a set of kicker magnets (which can rise from zero to full strength in 100 nsec, which permits a subset of the train to be extracted).
Second Stage Bunch Compressor

The second stage bunch compressor is divided into the following subregions:

  • An RF section that generates the necessary correlation between longitudinal position and energy. This section contains 456 9-cell RF cavities arranged in 57 cryomodules of 8 cavities each. There is 1 quad per 2 cryomodules and a phase advance of 60 degrees is used in each plane. Of the 57 cryomodules, 3 are spare and 54 must be accelerating the beam. The phase of the RF is between -22 degrees and -58 degrees depending on the exact configuration, and the maximum gradient required in the accelerating sections is 27.9 MV/m.
  • A wiggler with optics identical to the wiggler in the first-stage bunch compressor, but with weaker bends.
  • A longitudinal diagnostics section, which permits measurement of the central energy, energy spread, arrival time, and bunch length. Some or all of this instrumentation may be in an extraction channel and not the beamline which leads to the main linac.
Launch into Main Linac

The launch into the main linac performs the final conditioning of the beam necessary for main linac injection. This region includes a station for the measurement of projected emittances, and collimators which protect the linac from beams which exit the bunch compressors with unacceptable trajectories. There is also an extraction line which permits DC extraction, train-by-train extraction (via pulsed bend magnets), or sub-train extraction (via kicker magnets with 100 nsec rise tme), similar to the system at the end of BC1. The extraction system is downstream of the diagnostic section and upstream of the protection collimators for the main linac.

RTML Footprint

The figure shows the footprint of the possible RTML configurations: (a) Baseline design; (b) Alternate design with no turnaround; © Alternate design with single-stage bunch compressor; (d) Alternate design with shorter two-stage bunch compressor.

Parameter Tables

Beam Parameters
Parameter Nominal Value HighLumi Value LongDRBunch Value
init momentum 5 GeV/c
init espread 0.15%
init emit 8 um x 20 nm
init x jitter 1.0 sigma?
init bunch length 6 mm 6 mm 9 mm
final bunch length 0.3 mm 0.15 mm 0.3 mm
final momentum 15.0 GeV/c 13.0 GeV/c 15.0 GeV/c
final espread 1.1% 2.5% 1.6%
final x jitter 0.1 sigma?
ISR emit growth 0.34 um 0.18 um 0.34 um
emit growth budget 1.0 um x 4.0 nm?

Note: the magnet counts and system lengths for the two bunch compressor stages are based upon an existing lattice which has been studied, whereas all other beamlines are based on scaling from lattices used in the NLC or TESLA designs over the past decade. Thus, the values for all beamlines except the bunch compressor are very approximate, while those for the bunch compressor are merely somewhat approximate.

Magnet Counts
Region Bends Quads Sextupoles Dipoles Kickers Solenoids
DR Extract 10 40 0 80 20 0
Emittance 1 2 30 0 60 20 0
Collimation 0 15 0 30 0 0
Turnaround 100 100 10 200 0 0
Spin Rotator 6 30 0 60 0 4
Emittance 2 0 30 0 60 4 0
BC1 100 30 0 60 20 0
BC2 100 55 0 110 0 0
Linac Launch 2 8 0 16 20 0
Total 320 338 10 676 84 4
RF Components in Bunch Compressors
Region Cavities Modules Klystrons
BC1 32 4 1? 2?
BC2 456 57 19
Total 488 61 20? 21?
Instrument Counts
Region BPMs Wires BLMs Phase Monitors
DR Extract 40 0 0 1
Emittance 1 30 6 0 0
Collimation 15 0 0 0
Turnaround 100 0 0 0
Spin Rotator 30 0 0 0
Emittance 2 30 6 0 0
BC1 30 0 1 1
BC2 55 0 1 1
Linac Launch 8 4 0 0
Total 338 16 2 3
System Lengths
Region Length
DR Extract 85 m
Emittance 1 70 m
Collimation 110 m
Turnaround 170 m
Spin Rotator 80 m
Emittance 2 70 m
BC1 400 m
BC2 1000 m
Linac Launch 30 m
Total 2015 m


The justification for a section which matches the extraction geometry and betatron functions is self-evident.

The emittance measurement station immediately downstream of extraction is required because of the well-known sensitivity of the extracted beam emittance to beam position in the damping ring septum magnet. Only an emittance measurement station immediately following the extraction can be used to determine the optimum extraction orbit for emittance preservation.

During SLC operation, calculations of likely beam halo populations due to linac scattering processes did not explain the large observed beam halo, which was removed by collimators at the high-energy end of the SLAC linac. Because of the end-linac ILC beam parameters (energy, power, and emittance), it will be quite difficult to collimate the beam halo at the high-energy end of the linac. The risk of intense halo formation in the damping ring is mitigated with a relatively simple collimation section in the RTML, where the energy and beam power are relatively low and the geometric emittance relatively large.

The intra-train jitter requirements for the extraction kicker are extremely tight (0.07% RMS), and represent a luminosity risk for the ILC. In addition, the tight vertical beam jitter requirement imposed at the IP by the strong disruption (0.05 sigmay) also represents a luminosity risk. The jitter measurement and feedforward permit that risk to be mitigated. The turnaround is required to delay the arrival of the beam sufficiently for the beam jitter data to be processed and the jitter correction to be applied to the magnets.

The polarization of the beam is rotated into the vertical to preserve it during storage in the damping ring. The polarization at the IP has to be completely adjustable and tunable, and the adjustment/tuning is not permitted to dilute the emittance. The most straightforward method identified is to use solenoidal spin rotators with the lattice properties described above to cancel out the emittance growth from xy coupling that the solenoids would otherwise generate. Since the solenoids rotate the polarization from vertical to horizontal, the 8 degree arc between the two solenoid pairs is required to rotate the polarization from horizontal to longitudinal; thus the first solenoid pair plus the arc puts the polarization in the longitudinal; if the first solenoid pair is turned off and the second pair is turned on, horizontal polarization is generated; if both solenoid pairs are turned off, the polarization remains oriented in the vertical.

Since the coupling correction of the paired solenoids is not likely to be perfect, the coupling correction and diagnostic lattice is required to globally correct any residual coupling.

Bunch compression in the ILC is a necessity, given the opposing requirements of the damping ring (where long bunches are needed to limit collective effects) and the IP (where short bunches are needed to match the small values of betay which are mandated by the high luminosity goals). This compression is complicated by the large longitudinal emittance generated by the damping rings, which means that bunch compression leads to large energy spread after compression. Because of the energy spread, a single stage for compression from 6 mm to 0.3 mm RMS length was already marginal, and shorter bunches such as 0.15 mm RMS, which are required in the parameter tables, are not achievable in a single stage. The two-stage system works around this by accelerating the beam between stages of compression to limit the maximum fractional energy spread at any point in the ILC. The large and complex configuration of the wigglers is driven by the requirements of flexibility in the initial and final bunch lengths and by the requirements of dispersion tuning quadrupoles which do not introduce betatron mismatches or x-y coupling.

The final emittance measurement station is required to tune the emittance of the large energy spread beam generated by the compressor, prior to injection into the main linac. The collimation in the main linac launch is primarily machine protection segmentation: it ensures that a mistuned or mis-steered beam in the RTML will not result in a machine protection incident in the main linac. This segmentation is vital to simplify design of the active MPS.

Required R&D

None of the beamlines described in this document have been designed at this time. There is an optical design for the two-stage bunch compressor which does not include longitudinal diagnostics or beam stoppers and dumps, and which could be advanced to an acceptable level for costing with minor effort. Although there are no designs for the other regions, they are largely magnetostatic transport lines and similar lines have been designed for the NLC, TESLA, and LCLS, and designs for these areas can be developed without “starting from zero.”

Assuming that the above-described configuration is accepted as the baseline, the following R & D steps are required in order to produce a complete design for which the cost can be estimated:

  • Construction of a complete set of lattice files (“decks”) for the RTML and its associated extraction lines, including one iteration of optimization for minimum system length. This will probably require on the order of 2 man-months.
  • Tuning and tolerance studies (analytic and simulation). Since tuning studies are often fairly sensitive to the details of the simulation, it is necessary that this step be taken by at least two different people/groups working semi-independently. For this reason the tuning studies probably require at least 4 man-months. The exact length is hard to estimate as nobody has yet made a serious attempt at studying the combined transverse and longitudinal tuning of any RTML lattice for any linear collider design. Since the RTML tuning is in a much less mature state than the main linac, unpleasant surprises are a possibility that can't be ruled out.
  • Development of component and system tolerance specifications. This step will take approximately 1 man-month and is partially dependent on the results of the tuning and tolerance study.
  • Review of component and system tolerance specifications by qualified engineers. This step will take approximately 2 man-months, and depending on the outcome may indicate the need for further iterations of the design, tuning, and tolerancing studies listed above.

It is worthwhile to note that, compared to the other regions of the ILC, the RTML pushes the limits of technology in very few places. The electromagnets can easily be designed to fall within the limits of existing accelerator magnet technology; the RF components are duplicates of the main linac versions, and in fact are generally down-rated in their power and gradient requirements; the pulsed kickers for extraction of a runaway beam are based on similar technology to the system at the end of the main linac, and benefit from the much lower beam energy compared to the latter system; the bunch length monitors can be based upon very successful systems in use elsewhere for the measurement of much shorter bunches. The main technological issue for the RTML is likely to be the required RF system phase stability, which is a few percent of 1 degree of L-band. This phase stability must be maintained for a period which is long enough for a beam-based feedback to determine that an unacceptable phase change has occurred, as indicated by variation in the beam arrival times at the IP; thus, a stability period of a few seconds is probably sufficient.

Lattice Files

The lattice files for the baseline configuration's two-stage bunch compressor are available here.



Elimination of the Turnaround and Feed-Forward

One possible cost savings would be the elimination of the turnaround and the trajectory feed-forward which it supports. This would permit a reduction in site length of approximately 170 meters (actually slightly less because the 8 degree bend in the spin rotator would need to be compenstated by an 8 degree reverse bend elsewhere), and the elimination of about 100 bend magnets and quadrupoles. In addition, this would make the LET footprint a straight line in the xz plane, as shown in sub-Figure (b) of the footprint diagram, which could potentially simplify the layout of the overall accelerator on the site. On the downside, this would introduce a performance risk from the intra-train stability of the damping ring and its associated extraction systems. In order to safely proceed with this change, it would be necessary to demonstrate that the intra-train kicker stability of 0.07% can be met without feed-forward, and that the vertical orbit in the damping ring is also adequately stable bunch-to-bunch and train-to-train. Note that, although this alternative would reduce cost for the ILC, the downside risk to the integrated luminosity from beam jitter is sufficient that the Low Emittance Transport working group at Snowmass does not favor pursuing this alternative.

Single Stage Bunch Compressor

Sub-Figure © of the footprint diagram shows the footprint of the RTML assuming that the two-stage bunch compressor is replaced by a single-stage compressor. Note that although the figure implies a site savings of 1 km per side, this is partially compensated by the fact that the beam energy at the end of the RTML is reduced from around 15 GeV to 4.4 GeV, and thus the linac must be lengthened in this option; as a result the actual net savings is about 0.5 km per side, plus all 100 of the BC2 bends and about 30% of the BC2 quads and RF elements. Such an option can only be pursued if the emittance tuning strategies for the RTML and main linac can be shown to function reliably in the presence of an RMS energy spread in excess of 4%, and if the parameter sets assuming a longer bunch in the damping ring or a shorter bunch at the IP are eliminated. Because this change to the design would reduce the parameter reach of the ILC, the Low Emittance Transport working group at Snowmass 2005 does not favor pursuing this alternative.

Shorter Two Stage Bunch Compressor

Sub-Figure (d) of the footprint diagram shows the RTML footprint if a more compact two-stage bunch compressor is used. This design has compression capabilities comparable to the baseline design (ie, both the nominal 300 micrometer RMS bunch length and the shorter 150 micrometer RMS length can be achieved), but uses a single chicane for each stage of compression, rather than the 12 chicanes used in the baseline design. This design eliminates about 190 bends per side from the total BC1/BC2 system, along with about 25 quads per side. The site-length savings is about 700 m per side relative to the baseline. Since this design eliminates the emittance-tuning features of the wiggler in the baseline, and eliminates the symmetries of design which make those features possible, there is a risk of unacceptable emittance dilution which must be studied. In particular, the tuning strategy and installation tolerances of the shorter system must be carefully reviewed. Because the shorter two-stage bunch compressor design has not been studied to the same degree as the longer design, it is the longer design which has been selected as the baseline. However, it is highly recommended that this alternative be studied, since a design does exist and since, if this design proves tractable from the point of view of emittance tuning, it would permit a significant cost savings without sacrificing performance or parameter reach.

Lattice files for the shorter two-stage bunch compressor can be found here.

Required R&D

If the RTML is to be built without a turnaround and feed-forward, the required damping ring kicker stability must be demonstrated.

If the RTML is to be built with a single-stage bunch compressor, more intense studies of emittance preservation in the presence of a larger RMS energy spread will be required. In addition, it will be necessary to verify that the damping ring can achieve its required stability stability with short (6 mm RMS) bunches, and to verify that IP conditions will be tolerable with long (0.3 mm RMS) bunches.

If the RTML is to be built with the shorter two-stage compressor, more intense studies of emittance preservation will be required, concentrating on the absence of dedicated dispersion tuning quadrupoles.

bcd/ring_to_main_linac/ring_to_main_linac_home.txt · Last modified: 2005/11/29 16:04 by ring_to_main_linac
Recent changes RSS feed