CERN Accelerating science

Use Case: Global orbit feedback through the cycle

Sequence

1. Pre-injection plateau -  prepare feedback systems:
   • Twiss
   • Matrix
   • Energy
   • References
   • Internal parameters

   
   

2. Ramp to injection energy, no beam yet, feedback system off.
3. Inject pilot. Acquire trajectory, check injection oscillations
4. Establish circulating beam, acquire orbit.
5. GOFB on. Correcting to pre-established reference.
6. Take another pilot - acquire trajectory, check energy offset, injection oscillations with respect to closed orbit. Correct as necessary.
7. 2x10¹⁰ (trigger reflections, sensitivity of BPMs).
8. PPIPDDDD... loop remains on. GOFB acquires beam intensity at 1 Hz and makes parameter adjustments as necessary. So although we will be dumping over pilots etc., it should be able to track changes in beam intensity.
9. Prepare ramp I_ch/cv(t) -> power converters. FB system needs to know what correctors will be doing in the ramp.
10. Start ramp. HANDS OFF
11. [Change ref]
12. End of ramp signal.
13. Prepare matrices before and as close to squeeze as possible. -> controller  [take into account BPM dropouts, have to recalculate]
14. FB -> OK
15. Start squeeze. Optics changes to GOFB.
16. End Squeeze
17. Collide - separation bumps using functions. Controller changes settings? Clearly doesn't want to get involved with bringing separation bumps off..
18. Physics: GOFB on with appropriate limits and correction strategy.

Issues

• Correction strategy in common regions. E.g. orbit stabilization at the TDI
• How to apply energy correction with horizontal orbit correctors.  Discrete trims, pre-loaded functions?
• Need to be able to get trajectory for each injection whilst performing RT orbit acquisition
• Data management with Service Unit
• Who's going to do it?
• Testing?
• Correct reference and down-load to feedback system rather than pause correct and re-start.
• Re-use of business layer code as far as possible. Java not necessarily out.
• De-gauss of MCS???
• Centralised deal with MCS hysteresis...  plus interaction with chromaticity feedback - one brain.

Data

• Transfer functions
• Beam intensity at 1 Hz
• Limits
• Absolute current in correctors
• Energy at 1Hz
• Optics, Twiss. Would pre-load twiss, GOFB notified of optics changes.
• Reference orbits versus optics
• Number of beams
• Control loop parameters
• Control flow
• To business layer: trim history

Division of real time functionality between service unit and controller. Service unit interacts with business layer to get required data for controller. Imagines SU and controller running on separate RT machines. FESA suggested and nominally agreed as being the appropriate framework under which to develop these components.