1. Automated Collimation Operation
LHC-MAC
M.Jonker
On behalf of the collimation project
(as the controls coordinator)
Invite Bernd Roberto Stefano Hermann

2. Collimation Controls Collimation Controls Steering team (Cocost)
Stefano Redaelli (high level control applications)
Maciej Sobczak (css, middle level)
Roberto Losito (low level control)
Alessandro Masi (low level control)
Ralph Assmann (project leader)
Michel Jonker (controls coordinator)
Invited
Philippe Gayet (plc’s)
Rudiger Schmidt (machine protection)
Bernd Dehning (beam loss monitors)

3. Outline
94 (up to 160 in final upgrade) collimators, to protect against machine damage and magnet quenches.
The collimation process is a multi-staged process that require precise (0.1 beam) setting of the jaws with respect to the beam envelope. Goal for positioning accuracy is 20 m (0.1 beam at 7 TeV).
Actual beam envelope (position and size) may change (from fill to fill ?, by how much?)

4. Collimation Optimisation
Adapt to changing beam parameters to guarantee machine protection and to keep good cleaning efficiency
There are 376 degrees of freedom (4 motors per collimator) (188 if not considering the angle of the jaws)
30 seconds per degree of freedom (a very efficient operator) still requires about 3 hours.
 We need automated tools and procedures
Add sketch
Side view at one end
Motor
Temperature sensors
Gap opening (LVDT)
Gap position (LVDT)
Resolver
Reference
Microphone
Vacuum tank
+ switches for IN, OUT, ANTI-COLLISION
CFC
Sliding table
Movement for spare surface mechanism
(1 motor, 2 switches, 1 LVDT)

5. Setup Procedures Beam probing Fast beam based setup
Determine beam positions and size at every collimator by touching the beam.
Required for initial setup of a machine optics (injection and top energies ), or after substantial changes in beam parameters.
Setup at with a low intensity beam 5 nominal LHC bunches (equivalent to the Tevatron Beam in stored energy). Extrapolation from 5 to 3000 bunches…(bunch train effects?)
Fast beam based setup
Position collimators based on loss patterns, not on measured beam positions and sizes
Further systematic optimisation with nominal intensity beam
Response matrix corrections
Correct collimator positions guided by loss patterns.
Initial setup, changes in the machine optics:
Determine beam positions and size at every collimator. (Beam probing)
Assume accelerator is reproducible at injection (orbit, beta beat, etc.)
Collimator positions can be further optimised systematically at the injection energy with a loss pattern optimisation (fast beam based set-up, or response matrix based corrections)
Understanding of the machine, reproducible energy ramp &  squeeze:
Translate observed changes at injection to changes at higher energy. Re-establishing absolute collimator positions at top energy can be avoided for each new fill. Complement with loss pattern optimisation.

6. Response matrix corrections
Fine tune and optimise the cleaning efficiency (at injection or top energy).
Collimator response matrices to translate a given beam loss pattern into an adjustment of multiple collimator positions: PC = Mblc  BL
Theoretical matrices have been calculated based on Fluka and Struct programs.
Ambitious procedure, commissioning of this process will require many machine studies.
Not for the beginning
Not discussed here
BLM loss pattern
Teoretical matrices have been calculated with Fluka and Struct
Collimator Settings

7. Beam Probing
Establishes the beam positions, angle and size by probing the actual beam. A traditional method:
Starts with producing a well-defined cut-off in the beam distribution.
Each collimator jaw is moved until the beam edge is touched. This step defines an absolute reference position for each jaw. (and angle if two motors are moved independently)
Beam Loss Monitor
Beam Loss Monitor
From back to the front, first the tungsten jaws are moved in, then the secondary and finally the primary
Operationally more robust, to protect the cold aperture in the arcs
Other techniques to repopulate the beam by blow up
Note: Best done from the last element in the cleaning insertion to the first
Collimators may stay in place
Machine is better protected against quenches
Disadvantages:
Only possible with low intensity beam (i.e. 5 bunches).
Slow if done manually (188 positions )
Delicate (e.g. moving a collimator too far changes the cut-off in the beam distribution).

8. Beam Probing in SPS MD
by Chiara Bracca
Beam probing was tested in the collimation MD at the SPS in 2006 with the collimation control system.
The jaws were driven in by the control application either manually or in repetitive steps.
The control application simultaneously displays jaw position and Beam Loss Data
In the MD, to speed up we used successively smaller steps, and while doing so we scraped the beam away bit by bit.

9. Fast beam based setup Fast if implemented as an automated procedure:
Complements the traditional set-up method.
Adjust positions to reproduce known beam loss pattern.
Based on experience of other accelerators: Collimation efficiency is more closely related to beam loss patterns than to absolute collimator positions, which are sensitive to orbit deviations, beta beat, etc.
Move jaws in hierarchical order into the beam halo up to the point where a specified beam loss level is recorded in the adjacent beam loss monitors.
Beam Loss Monitor
Beam Loss Monitor
While we optimise one beam in IR3, we can optimise the other beam in IR7
Fast if implemented as an automated procedure:
Start at a fixed offset relative to a previously known position (only have to move short distances, no need to be retracted.
Two beam can be tuned in parallel in the two cleaning insertions IR3 and IR7

10. Fast beam based setup Procedure in practice: Timing implications:
The collimators are set at 1.5 σ retracted with respect to the last optimised value.
The jaws are optimised one by one in a precise order.
Optimization by moving in steps of 0.05 σ until the associated set of Beam Loss Monitors (BLM) detects a predefined value of beam loss. The BLM reference levels are found empirically and may be updated from fill to fill.
Timing implications:
Starting position –1.5 σ, step size of 0.05 σ ( GeV)
⇒ 30 steps/motor ⇒ 9600 steps in total (only position, no angles, final upgrade).
Available time 5 min. two rings in parallel ⇒ 60 ms per step (16 Hz)
@ 2mm/s 50 μm ⇒ 25 ms per step needed for motor movement
=> 35 ms for driving, data collection, reading BLM, deciding

11. Controls Architecture
LHC tunnel
Underground, low radiation area
Surface support building
Control room
Control room software:
Management of (critical) settings (LSA)
Preparation for ramp
Assistance in collimator tuning Post Mortem data collection and Analysis
Based on standard LSA components
Dedicated graphical interface for collimator control and tuning
Collimator Supervisor System (CSS):
Support building, VME / FESA
Fesa Gateway to Control Room Software
Synchronization of movements
Beam Based Alignment primitives
Takes action on position errors (FB)
Receives timing, send sync signals over fiber to low level (Ramp & Beam Based Alignment)
Synchronization and communication with BLM
Low level control systems
3 distinct systems
Motor drive (PXI)
Position readout and survey (PXI)
Environment Survey (PLC)
Central Collimation Application
Ethernet
Controls Network
Data Base
Actual Machine Parameters
Critical Settings
Machine Timing
Machine Timing Distribution
BLM system
Local Ethernet Segment
Collimator Supervisory System
(one or two per LHC point)
Synchronisation
Fan out
How can we implement this automated optimisation
Motor Drive Control
PXI
Position Readout and Survey
Environment Survey
PLC
Beam Permit
. . .

12. Fast Optimisation Primitives
Collimator Supervisory System (CSS)
Send a trigger to adjacent BLM system on every motor movement
BLM system sends a short “transient” data to the CSS
Optimization primitive command (on CSS)
Move until BLM-level
Parameters
Motors and step size
BLM signals and limits
Repetition frequency
Maximum steps
Example:
Move Jaw-left in steps of 10 um every 30 ms until BL signal reaches 103
This optimization primitive can be used by a central application for
Beam Probing
Fast beam based optimization
Collimator Supervisory System
(one or two per LHC point)
BLM system
Synchronisation
Fan out
Local Ethernet Segment
Motor Drive Control
PXI
Position Readout and Survey
Change colors green on blue
Beam Loss Monitor

13. Fast Optimisation Primitives
During optimization, positions are continuously measured,
If the position gets out of tolerance, the procedure will be interrupted.
Collimator Supervisory System
(one or two per LHC point)
BLM system
Synchronisation
Fan out
Local Ethernet Segment
Motor Drive Control
PXI
Position Readout and Survey
15 µm
Change colors green on blue
~ 25 µm mechanical play
R. Losito et al

14. BLM Transient in SPS MD
Adjacent BLM triggered by collimator movements.
Collected data:
Transient Data Buffer (2.5 ms sampling 80 ms for BLM based FB).
Post mortem data (40 us sampling over 1.7 seconds)
(For analysis)
Data needs carefull interpretation.
LHC acquisition chain tested, including link with collimators (trigger and data transmission).
Plots from Daniel Kramer
Motor movement
10 ms (20 m)
Motor movement
10ms (20m)

15. Conclusion Fast collimator optimisation is technically possible.
The controls architecture contains the necessary elements to deal with these requirement (synchronisation lines, BLM data acquisition and connection)
The principles have been tested during an SPS MD in 2006
Fully automated steering application and procedures to be developed and tested (2008)
However, the real challenge…
beam dynamics…
SPS md: BLM responds to collimator movement over time scales of 100th of ms

16. Conclusion
Motor movement
10ms (20m)
During the SPS MD, not able to make clean cut in the beam distribution
Re-populatution of tails over 100th of ms.
Motor movement
50ms (25m)
Long tails after collimator movement,
Large noise components
(70 Hz)
50, 150, 300, 450 &
600 Hz noise
Plot from maciej for timing response
Plot with LVDT reading (stop movement when becoming unsafe)
Message: control system which is powerfull
we can remove reliable
We get fast BLM signal
Tools are available in the control system,
Limits are in the understanding of the beam dynamics
Top level manula beam movement for beam probing
Testing and implementation
Need good understanding of beam dynamics
SPS MD in 2007 to investigate the origins of these problems
If these effect are also present in the LHC, optimisation will me more challenging.
Loss tails with echo
12 sec