results from beam-based collimator alignment

LHC Collimation Working Group – 02 April 2012

Results from Beam-Based Collimator Alignment

G. Valentino, R. W. Assmann, R. Bruce, F. Burkart, M. Cauchi, D. Deboy, L. Lari,

S. Redaelli, B. Salvachua

Outline

Gianluca Valentino

2

• The collimator setup application

• Fast BLM data acquisition • Software issues encountered during the setups

• Results comparison: 2011 vs. 2012

• Qualification of the collimator settings

• Summary

Reminder: collimator setup application

Gianluca Valentino

3

Setup Task Sequencer

GUI

User

Parallel Setup Algorithm

Loss Threshold Selection

BLM Spike Recognition

Fast BLM Data Acquisition

BLM Feedback

Structure as proposed for March 2012 setups

12.5 Hz BLM data acquisition

Gianluca Valentino

4

cs-ccr-logging2

cs-ccr-dev

Thanks to: R. Assmann, V. Baggiolini, A. Bland, B. Dehning, S. Jackson, S. Redaelli, C. Zamantzas

CCC Setup Application

Fast BLM Data Logging

~3900 BLMs in all crates

12.5 Hz and 1 Hz BLM signals

Left jaw aligned

Right jaw aligned

Issues encountered during setups

• Some software issues were encountered in the 450 GeV setup, which were fixed by the 4 TeV setup:

- Fast BLM data was transmitted from the low-level with channels & crates in reverse order for each IR.

- GUI: multi-threading issues prevented full functionality of GUI.

- Sequencer not jumping correctly through the setup stages (TCP, parallel, sequential).

- Setup Sheet: left and right jaw positions not saved correctly.

• Outstanding issues:

- Loss spike recognition: All clearly optimal spikes (e.g. TCP IR7) were classified correctly.

- Main issues were for “grey area” spikes: in the classifier design, it was preferred to err in classifying an optimal spike as non-optimal rather than classifying an non-optimal spike as optimal.

- When detecting a non-optimal spike, the jaw is moved in once again until an optimal spike is classified. This could cause dumps at 4 TeV if the jaw is at the beam and is moved in erroneously again.

- Algorithm assumes a simple loss spike model with a low noise level, no other spikes after main spike, etc.

- An improved spike recognition algorithm will be developed and tested in an MD.

• Beam instabilities: possible impedance effects with too many collimators close to the beam (N. Mounet)

Gianluca Valentino

5

Alignment results: setup time

• Setups in March 2012:

• Comparison in setup time 2010 – 2012 for the full system:

Gianluca Valentino

6

2010 (Manual) 2011 @ 1 Hz 2012 @ 8 Hz

450 GeV 3.5 TeV 450 GeV 3.5 TeV 450 GeV 4 TeV

# Fills 3 5 6 3 3 1

# Dumps 1 4 2 0 0 0

# Hours 11 28 18.5 17.7 10 7.5

Setup Type Injection Flat Top Colliding Squeezed

Date 21st March 29th March 30th March 31st March

Alignment results: beam-based centres

• Beam-based centre comparison to 2011 for injection orbit.

Gianluca Valentino

7

µ = 0.177 mm σ = 0.033 mm

µ = 0.229 mmσ = 0.076

mm

Delta: 450 GeV 2011 vs. 2012 B1 and B2 Delta: Flat Top 2011 vs. 2012 B1 and B2

∆ TCLIA.4R2 = -12.920mm∆ TDI.4L2 = +7.048mm

Alignment results: beam-based centres

Gianluca Valentino

8

µ = 0.238 mmσ = 0.111

mm

µ = 0.183 mmσ = 0.059 mm

• Beam-based centre comparison between 450 GeV and 4 TeV in 2012 (injection orbit).

Delta: 450 GeV vs. 4 TeV B1 Delta: 450 GeV vs. 4 TeV B2

Alignment Results: Beam Size Ratios

• Beam Size Ratio comparison: 2012 vs 2011, 450 GeV

Gianluca Valentino

9

IR2 IR3 IR7 IR8 IR3IR7

TDI TDI

Alignment Results: Beam Size Ratios

• Beam Size Ratio comparison: 2012 vs 2011, Flat Top

Gianluca Valentino

10

IR3 IR7 IR3IR7

Collimator Settings Qualification

Gianluca Valentino

11

B. Salvachua

LIC BLMs

LIC BLMs

Collimator Settings Qualification

Gianluca Valentino

12

B. Salvachua

Summary

• Collimator setup time reduced from 17 hours in 2011 to 7.5 hours in

2012 at flat top, 1 fill.

• Limit for setup time not yet reached: improvements can still be made.

• Further tests will be carried out in an 8 hour slot in April MD.

Future Work:

• Only GUI will run on CCC console: BLM feedback and jaw movement

requests to be moved to a server.

• Distributed application will be developed based on Java RMI and JMS,

running on the server and the CCC console.

• More complex beam loss recognition system to be designed.

Gianluca Valentino

13