Observations on ILC LET and Ground Motion

December 6 2006 Ground Motion Group Global Design Effort 1

Observations on ILC LET and Ground Motion

Paul Lebrun

Fermilab CD/AMR

December 6 2006 Ground Motion Group 2

Overview

1. Brief description of Low Emittance Transport (LET) work…

2. First look at Ground Motion,1. Data vs ATL model

2. In CHEF, for a 5 to ~50 GeV, ILC-style LINAC

3. Future work:1. More realistic alignment

2. Ground Motion1. Modify, improve ATL ass need be

2. More data ! Longer distances, higher frequency..


LET: Far from final…

Overview of progress of LET at Fermilab.– Simulation work: Done in TD, CD, AD

• AD/CD : – CHEF Accelerator Simulation code, maintained by

Francois Ostiguy (AD) and Leo Michellotti (CD) – Alex Valishev : Main Linac Lattice and LET studies.

• TD: Working with LIAR and Lucretia (Kirti Ranjan), • CD: Valentin Ivanov, P.Lebrun, : LET, static and

dynamic, using private code and CHEF.

– Dynamical LET: • High on priority list.. With CHEF, just started a few weeks

ago. • Somewhat software intensive (GM code, vibrations,

beam jitter..) • CPU intensive!!


LET: The basics

Something we already do for RunII, in the TeV!– Re-adjusting orbits to preserve emittance.

• 2D Vertical phase space for ILC is ~ 50 times smaller then our “few pi” numbers..

• ILC is pulsed machine, not a relatively stable ring• No “orbit”, just a trajectory.

• Linacs – Dispersion (D) Free Steering( DFS) method.

• Given large uncertainties on BPM offsets, tune the dipole correctors to a given Dispersion function instead of a prescribed path. If so, the BPM offsets cancel out.

• If D is small or, preferably, set to zero, BPM scale error also don’t matter.


LET Benchmarks And Algorithms

• While the DFS method is in principle straightforward, there are numerous tricks to play to make it more realistic, and to optimize it in case of multiple sources of Dispersion. – Implementation often messy.

• LET performance may depend on the quality of tracking code Benchmark.– An agreed upon lattice (Tesla Main Linac)– BPM resolution – Wakefield– Misalignments and BPM offsets.– …. (That was the hard part!)


Benchmark, Fixed dipole setting

Non trivial emit emittance growth!

Ups and down because D taken out.. and coming back..

Agreement not perfect, sensitivity to small local difference in tracking.. Improves if Dispersion if corrected.


Determining Dipole Settings..

O.K. Performance agreement is o.k., but solutions are different…


What does it has to do with GM ?

• If multiple solutions to a given misalignment pattern give roughly the same performance, are these solution robust and stable.

• Preliminary results on the dynamical problem (initial set of misalignment, with beam jitter and ground motion) show that we are not able to converge towards a solution that has good performance, over time.

• So more work is needed!


Ground Motion Model and reality.

• Valentin Ivanov translated in C++ the “ATL” model from A Seryi.

• Integrated in CHEF• First order, naïve comparison with Jim Volk et

al data.– For October only.. No long time duration

studies!– From the MINOS hall.. Not the “good” Galena

Platteville dolomite.


From Jim’s web site..

Big long term motion! 20 microns swing is very, very likely to demand a complete retuning (i..e,DFS) re-adjusting of the LINAC. ~ 500 microns is also likely to justify a physical re-alignment. Lots of frequencies..

L2 (microns)


Yes, we see the moon tides…

L2 (microns)

Data taken starting Oct 12, 20:08

The ~ 12 hours periods seems to be there, visible for about 2 days. ~ 0.25 to 0.5 microns/hour.

Then the tide amplitude rises and long term (~week) motion also increases..


Data conversion: taking out the global tilt

Delta 2/3_1 = L2 – (L3 + L1)/2. Large fluctuations remains.. 40 microns, over a few days, over 60 meters..


Comparison with ATL, for 2 hours.

Oct 8, 10:04 A.M. Oct 3 4:00 A.M.


Why two hours ?

• At best, a complete DFS re-steering will take: – ~20 pulses per setting, to average over beam

jitter and finite BPM resolution. – x2, need off/On momentum to measure

Dispersion. – 20 iterations per local DFS section.– x 30 to 60 : the number of DFS section for the

entire LINAC. – ~80% uptime – > ~ 1 to 3 hours…


LET Perf. with Beam Jitter + GM

Vertical Emittance, not corrected for Dispersion, for

Emittance, corrected for Dispersion.


Will it “work” ?

• With only magnetic steering ? – At first try, DFS steering did “converged”,

albeit with looser a convergence criteria then in the static case. Yet, the emittance growth is large: 40 % of the budget in 2 km !

• And I ran for ~10 minutes.. (one day of CPU time!) – Need to run for longer periods…

• More realistic misalignements!!!!

– Things to try (Software): • Concurrent DFS steering across sections ?• Better DFS steering algorithms and parameters • Further check of Ground Motions..• Better Control software (??)


Further investigations...

• Movers on quadrupoles and/or cavity ? – Cavity tilts at ~ 10 GeV have really bad effects

on LET !! • Much smaller (<~ 1 nm. Rad) are predicted if the cavities

can be placed “laser straight”, one only has to correct for quadrupole displacements.

• Feed forward Steering..Laser-track the motion of the machine, and use Beam steering to “check”..

• Need a lot more instrumentation ! – HOM BPM– Synch. Rad. detectors ?


Better ground Motion Modeling ?

• Conversion of exiting data to formal ATL parameters

• Beyond minimal ATL– Understanding why drives these motions?– Tide related frequencies/phases well known! – Needed if feed-forward will be considered.– However, evidently, earth quakes are

notoriously hard to predict!


Better Ground Motion Data?

• Goal: improve ATL model, make it reliable. – Systematic error analysis on existing HLS

data.– Frequency: Need ~ 5 Hz

• 10 Hz probably.. • Not KHz (intra-train effect: something entirely different.

Probably not ground motion!)• Means Laser-tracker technology.

– Longer distance.. – MINOS or Aurora mine ? Or other site ?

Documents

Observations on ILC LET and Ground Motion