This website is no longer maintained. Its content may be obsolete. Please visit http://home.cern for current CERN information.
See Luca Bottura's presentation at Chamonix 06 here and Mike's here. Roughly:
Sector Test [7-8]: [Possible August 2007]
450 GeV Calibration run: [November 2007]
As above for whole ring plus snapback. Operational tests September/October
2007
7 TeV - commissioning [Mid 2008]
The full Monty.
(Note: Call B versus I Calibration Curve to distinguish from Transfer Function (B/I versus I).
Issue: assuming for the moment we plan to perform linear interpolation from database look-up. How close do the points have to be? Is this accurate enough?
ML 13/4/06
December 2005 Quick and dirty implementation of FiDeL in Java under LSA. Coefficients to the database. Run model to produce transfer functions and harmonics(I) - write to look-up tables, which are read during settings generation. Ran through momentum - angle - MB78/I - b3 - MCS78B1/K - MCS78B1/I.
November 2005 Quick implementation of FiDeL in Mathematica - looks cool - could certainly do this in Java - Luca's team could supply fit coefficients and we could generate TF and coefficients on-line. Pictures here.
Pulling in transfer functions from magnet measurement database (first curves October 2005) - this only for the selection actually measured. For the rest need warm to cold correlations and use this to correct average of measured curves (10A measured warm is extrapolated to 5000 A - basically this point is used to shift the whole curve)
Magnetic Field Prediction for the LHC Control System - meeting 31/05/05 Notes
How the magnetic measurements and the Reference Magnet System (RMS) will be used for commissioning? LB Cham 05
Meeting Report about the architecture of the data acquisition system for the Reference Magnets April 04
Beam
Based & Reference Magnet Measurements to Complement the Magnetic
Measurements ML Cham 04
Reference
Magnets LD Cham04
The RMS will measure b1, b2 and b3 at least. The hall probe and coils will provide on-line (real-time) measurements at around 3-10 Hz. It will have a predictive model which will be able to anticipate persistent current drifts and snapback given the powering history and the length of time at injection.
automatic feedforward, continual monitoring to hold to 10-4 of SPS, care with respect to positioning because of longitudinal variation of b1. Used to track and keep energy constant via orbit correctors on the injection plateau.
5th June 2003 Meeting to discuss RT in the context of the Multipole Factory: Notes plus outline
b1,a2,b2,b3,b5
Need to distinguish need for off-line predictions from RMS and on-line feed
forward. There is always need for off-line predictions at the best possible.
Maybe "predictability level" should be "required RMS correction
level".
Commissioning
- prediction from off-line RMS should be enough to establish circulating beam (given orbit correction etc.)
- a prediction of the decay to within � 20% should be good enough - we can always measure the beam
(b2,a2,b3,b5..)
- snapback predicted following knowledge of decay amplitude
Nominal Operation
1. Multipole errors: ~1 minute after arrival at injection level - injection of pilot
- given reasonable variations on the nominal operation sequence, reproducibility
and the RMS (off and on-line) should be good enough to get close to operating
conditions; certainly good enough to get easily a circulating pilot without major trims. (Here HERA now manages to within � 2 units of chromaticity using look-up tables (with a
pre-cycle) without resort to reference magnets.)
- will reproducibility/off-line predictions be good enough?
- can on-line RM help in predicting the initial field components? Quantify.
I don't think we need to consider directly injecting a high intensity beam
2. Decay
- without beam instrumentation - here I assume you mean we can't rely in anyway on beam-based measurements. This is probably extreme: in normal operation we
can always measure and correct the effects of b1, b2, a2, b3 etc. on a pilot, and probably
an intermediate beam. This should allow one to disentangle the average decay rate and to confirm the RMS
predictions.
However let's imagine we can't measure too much with the nominal beam; this would mean at least 8 minutes on the injection plateau trusting to pre-loaded off-line RMS predictions.
- can the on-line RM input help control of PC decay during standard operation? Quantify.
Not sure of the relevance of the other BI scenarios
3. Snapback
b3/b5 should be taken care of by proposed fit. (prediction) - clearly indispensable. Given input from the corrections/measurements made during decay, predictions should take care of a large part of the snapback
- Use of on-line RM to input to improve accuracy during snapback?
- "HERA magnets use superconducting NbTi filaments with ~14 micron diameter, almost twice as large as the filaments in the Tevatron strands. Injects at a very low energy of 40 GeV. For these two reasons the hysteretic injection sextuple (and other allowed multipoles) are much larger than in the Tevatron."
LHC filaments 6-micrometres
- b3 drift at Tevatron +1.5 units over 10,000 seconds, 150 GeV
- around 0 geometric (ends compensate body)
| PC chromaticity H | Q'V | Change over 15 min | ||
| Tevatron | -140 | +119 | +8 | -7 |
| HERA | -275 | +245 | +13 | -11 |
| RHIC | -38 | +36 | +2 | -2 |
| LHC | -450 | +450 | +150 | -150 |
Qh 0.16 lower than expected, Qv is 0.16 higher than expected, use trim quads to correct, not understood. <b1> +1.4 in dipoles, operationally not a problem.
Coupling - creep in suspension.
decay and sb - problem operationally. Tune drift - feedforward - repeatable for same ramp history. Skipping the pre-cycle after a long store - changes.
sb - frist 6 seconds, worrying about tune. 0.015 units of tune. Alogrithm - FP duration ,
Coupling drift - measured min tune split. < 0.003. Coupling snapback not measured.
Tune & coupling drifts not explained. feeddown?
All studies with helix off.
Q' - 50 units drift, corrected to ~2 units. Nominal 4 units.
SB - FP duration into the alogirithm. RF frequency shifts - lot of beam time for measurements.
b3 alogorithms ofr drift compensation.
snapback time, snapback amplitude - FT time, BP time, FP time
DRY SQUEEZE???? Might we need it???
Q' - raamps with different frequencies.
RUN 1: ramp Tevatron 6 times.
HERA
HERA hysteretic sextupole at injection is -32.5 units (~275 Q'). Dynamic effects are also twice as large than in the Tevatron.
- "Prediction due to measurements in reference magnets +/- 2 units"
- decay 10 units in 30 minutes
- field/current at 40 GeV. 0.2267 T/244 A
- geometric b3 ~0.1 (sigma ~ 2.7) units (ref radius 25 mm (2/3))
- ~2/3.5 units of b3 decay in VAC/LMI conductor
- depends on manufacturer
"discrepancy between reference magnet and "average" magnet in the machine can be as much as 6 units of chromaticity which is ~1 unit of b3!
- delay of 300s before attempting to compensate the effect
pc 270 units decay 10%
Ex. review...
natural -44/-47
Always do pre-cycle...
Snapback +/- 30 units - major contribution predictable from Ref. Magnets. STILL USED ON-LINE
dependence on time at injections
beam energy at injection - non-problem
Q' drift - non-problem with b3 compensation ON
nearly non-problem - Q' snapback - corrections from empirical trim (operational experience) - feedforward. 5% effect. Measurement at 5 Hz, correction at 1 Hz - too slow.
Pitch effect needs to picked up by NMR probes - essential.
Real problem - prediction of Q' at injection
Tunes on ramp - tables + operator twiddling
Prediction not good enough: delta Q' +/- 6
Time dependence of manufacturers important decay constant 2*
Beam loss - injecting into head-tail instability... losses on ramp 100 ma 0.5 to 1%
RHIC
dry squeeze - hysteresis cycle
squeeze during the ramp, slow squeeze down to keep B'' under control.
design model versus trim model
Power converters didn't follow. Inductive lag. Reduce B'.
RHIC main dipoles 1.3 mm off-centre - incorporate into model.
Transfer function - error - coil calibration problem - bung in model.
Beta beat 5% in arc, 15-20% in IRs
Chromaticity measurement on ramp 200 micron modulation at 1 Hz plus PPL
ad hoc adjustment of model - 10 units on the DX
PLL, PLL, PLL, PLL.
On-line model, on-line model, on-line model.
LUCA
Ref magnets during sector test
I plan to have a model running with some average description of the magnets (i.e. a simulation of what magnets do given what their current and temperature was and is), and maybe some measurements from an unreal-time system running on one of the test benches (but this is questionable as it will impact on the testing schedule). If there is no strong request, I do not think we will be able even to have a prototype of the measurement system for the reference magnets.
Choice of reference magnets
(selection of reference magnets, mixing suppliers, sorting)