The ramp will be driven by current, voltage and frequency as functions of
time, pre-loaded to the power converters and RF, with the start of ramp
triggered by a timing event. Before the start of the ramp we retract TDI and
injection collimators and leave the cleaning insertion collimators where they
are. Real-time corrections from operator-controlled knobs, feed-forward from
reference magnets, and beam-based feedback is eventually foreseen, but these
tools are unlikely to be in place for the first attempts. The main challenge
will be anticipating the depth of the snapback and attempting to deal with
associated swing of beam parameters. The strategy will be something like:
1. Establish length of time on injection plateau
2. Reference model or measurements to establish depth of snapback.
3. Predict required corrector functions and incorporate into the machine
settings
4. Load functions to hardware
With the larger tolerances, input from the offline RMS, tune and orbit
measurements and some rudimentary corrections we should be able to get some beam
through the snapback.
Beam 1, Beam 2
Prerequisites: Stable pilot at injection, nominal cycle, predictions from FiDeL, control system for parameter control during ramp in reasonable state.
Exit: Reasonable transmission of pilot through first 2 minutes of ramp. [26 GeV ~ 100 s.]
This will essentially be a hands off operations (we hope). Predicted functions will have been downloaded to the power converters. Tune feedback, orbit feedback and RMS feedforward will be not be operational yet. Operators should have real-time knobs available to control Q and Q'. Key thing here is:
Would aim to make a programmed stop just above the end of snapback - to be test before without beam. Start worrying about reproducibility, full re-cycle after each attempt. Fixed length on pre-injection plateau.
Gradually increase intensity.