Overview
Hysteresis compensation on SFT flat top without dipoles pulsing on MD1. This requires the SFT injection to be set up for the lack of precycle
Other measurements
- Beam tune at SFT flat top without compensating quadrupole hysteresis , but after manually correcting dipole field w.r.t. B-Train during supercycle transitions with Q20 and Q26 LHCPILOT to validate Quadrupole measurements.
- Start with 3x SFT, then switch to SFT + LHC, stay 10 cycles to stabilize hysteresis, take 10 measurements on tune, then switch back to SFT, stay 10 cycles to stabilize, then take measurements.
- B-Train measurements on 26 GeV injection plateau without MD1.
- Take a 6 BP parallel MD cycle, set the main dipole current to 26 GeV equivalent across the full cycle, so without ramping to 200 GeV. Quadrupoles are allowed to ramp.
- Configure a supercycle SFT + SFT + MD and do measurements. By having beam on the SFT cycles we can also validate our eddy current parameter fits in post.
- This is a “wishful thinking” SSC configuration if we do not have a precycle.
- If possible, re-do with 14 GeV equivalent cycle.
MD Preparation
-
Train new model with Transformer LSTM on Dipole datasets v9 [completion:: 2025-08-10]
- Parameter tuning on Dipole datasets v9 [completion:: 2025-08-10]
- Pre-training on Jiles-Atherton data? [completion:: 2025-08-10] TFLSTMPRE-3 and TFLSTM-34
-
Double check Eddy current decay parameter fits [completion:: 2025-08-10]
-
Test predictions with new model with sps-app-hysteresis and sps-mlp-hysteresis-compensation [completion:: 2025-08-13]
-
Test predictions with eddy current decay, ideally with multiple trims, and trimming over multiple BPs [completion:: 2025-08-13]
Models to use
Suggested models:
MD Setup
MD Execution
Log
Logbook entries
https://logbook.cern.ch/elogbook-server/GET/showEventInLogbook/4363209