CERN Accelerating science

LHC-CP sub project on RT controls for the LHC

Meeting no.4 : Friday 16-03-2001 - reported by Mike Lamont

Present : M. Jonker, A. Butterworth, P. Ribeiro, M. Lamont, O. Berrig,  O.Bruening, Q. King, J. Wenninger

Agenda :

• Logical architecture
• Local orbit feedback - Olav Berrig

Minutes :

Comments on the minutes

Michel questioned the need for the power converters to have functions pre-loaded. It had been stated in the previous meeting that this was necessary to enable the PCs control algorithm to see what was coming.

There followed a long discussion driven by a white board exposition by Quentin.

• The RST algorithm work with a given granularity which varies from converter to converter, typically of order 30 ms sampling. Downloaded points are interpolated between. Quentin - please write me one paragraph to explain the issues!

Mike showed a DFD which might be something like the foreseen logical architecture (here). Comments included:

• We will sum real-time trims at top level
• Possible exceptions could be correctors used in local feedback - hardware reservation, global orbit feedback still needs to know what's going on.
• Corrector failure clearly has to be managed properly.
• Discussion about cross feedback system interference: 1. frequency separation. 2. What about multipole factory versus Q' feedback. It was argued that the MF was feedforward on measurements from main dipoles, there shouldn't be interference.
• Care about global orbit feedback acting only on appropriate global structure with a few correctors should avoid it interfering with local feedback. This would also imply that problem pickups will only cause trouble for the local feedback systems.
• Somewhere, somehow the real-time knobs should have some intelligent damage limitation safeguards.

Local Orbit Feedback - Olav Berrig

• Very tight requirements on the position of the orbit in cleaning sections, of order 30 microns at collimators at 7 TeV (300 microns at 450 GeV, collimators stay out during ramp, come in during squeeze to follow tightening of aperture in triplets, transients will be the critical regions as regards orbit stability).
• Many corrector magnets need to be controlled simultaneously
• The angle of the beam at in these regions also needs to be controlled.
• Also: beam position at special (shottky, high sensitivity) pickups and wigglers in point 4.
• Also: beam position of single beam) at injection - points 2 & 8. Requires that regulation can select a bunch.
• 1 Hz movements of beam -> 10 Hz regulation system -> 100 Hz sampling rate. But factor 10 is generous.
• Feedback noise. Any additional system introduces noise at its closed loop frequency, Each time the corrector magnets are updated with new values, we introduce frequencies. If we update every 100 ms, we will introduce 10 Hz noise.

To implement such a system ROSY (Real-time Orbit Stabilization System) is foreseen. The hardware implementation under study foresees splitting the BPM signal directly from the optics fibre, a dedicated ROSY create, direct connection to power converter gateway, turn by turn acquisition and communication to the global feedback system.

Quite a few comments:

• Do the power converters involved have enough voltage to react fast enough? Superconducting correctors are 7H/8V which limit them to 5-10 Hz. Warm magnets will be faster.
• Big discussion about turn-by-turn data acquisition, precision, filtering, bandwidth, getting rid of 50 Hz, whether 100 Hz was realistic, whether 10 Hz with 20 ms averaging would do etc.etc.
• Issue to be revisited next meeting when Olav will presented his ideas in more detail. He will address which magnets are involved, available voltages (and thus small signal bandwidth), filtering of beam pipe...
• Are there poles in the left plane?

Olav also asked "Will the Global feedback system need regulation?" to prevent noise from the BPM measurements introducing noise to the beam.

• To damp out the noise from BPM measurements (better that the 20 ms integration) maybe regulation or filtering will be needed.
• The feedback system kicks the beam. What is the effect on beam emittance?
• Is there any possibility that the noise frequencies will hit some of the eigen frequencies of the beam or the corrector magnets?

It was felt, echoing a point raised above, that global orbit correction with few well chosen correctors will inherently not be able to attempt correction of noise, naturally directing corrections at global features.

Next Week (23rd March)

Meeting with MTA to discuss Multipoles Factory - pavilion 10:00. Not strictly real-time but come along if you're interested.

Next RT Meeting

Local Orbit Feedback revisted - Olav Berrig
Lessons from SLC/PEP-II control system - Thijs