FP420 Alignment WithBeam Position Monitors
Jo Pater (Manchester)
14-16 July 2008
July 2008 J.Pater - BPM-based Alignment 2
FP420 Alignment Plan
• LHC button-style BPMs on fixed beampipe + Wire Positioning System (WPS)– Alignment wire is absolute reference
• Similar but modified (larger-aperture) BPM on each Hamburg pipe– Referenced to detector by knowledge of mechanics
• Offline track-based alignment using exclusive dileptons (M.Albrow et.al)
July 2008 J.Pater - BPM-based Alignment 3
Hamburg pipeHamburg pipe
BP
M
BP
M beam
LHC beampipe
Alignment wire
WPS sensors
bracket
Fixed BPMs + WPS
Overall accuracy of ~10 challenging: tolerances of individual components add up quickly:– WPS sensors: known to be accurate to < 1– Mechanics: <10 tolerances possible but not easy !– Complicated by the moving bit– BPMs: need micron-scale accuracy and resolution
July 2008 J.Pater - BPM-based Alignment 4
BPM issues for FP420
– Preliminary study (JP) suggests that BPMs are capable of it (see following slides)– Will need carefully designed readout electronics (see following slides)– UCL engineer (A.Lyapin) on board, experienced with BPMs for linear colliders
• Electrode design could be tailored to give better performance if necessary– Two larger electrodes (instead of four smaller) would give better performance but only
in one dimension
Workshop April 2007: what BPMs are the best choice for us?
– 1st choice: LHC BPMs (electrostatic button-type)
• Already used in large numbers in LHC– minimises integration issues
• Can be optimised to special diameters – e.g. to mount on Hamburg pipe
• Micron-level precision/resolution believed possible, although not demonstrated, by LHC team:
July 2008 J.Pater - BPM-based Alignment 5
Manchester, Cockcroft Alignment Test Bench• Damped, floating optical table 1.2m wide x
3m long (see next slide)• 3 WPS sensors + wire + readout• 2 LHC BPMs (i.e. the fixed BPMs)
– Horizontal setup: • beam wire stretched by hanging weight over
pulley• ends of wire on micro-positioners
– Read out with network analyser (courtesy of Cockcroft Laboratory):
• 40 MHz CW on beam wire• Electrode signals (unamplified) compared
internally with reference• calculated offline
– NB: as yet no large-aperture BPMs (i.e. the ones that will move with the Hamburg pipes)
• 2 Schaevitz LVDTs and signal conditioners– Read out via DMM + Labview
July 2008 J.Pater - BPM-based Alignment 6
LHC BPMs in FP420?Preliminary Study (JP, July 2008)
• Resolution– A quick study:
• 6 repeated measurements of under identical conditions, at two different wire positions, yields standard deviations of 0.00025 and 0.00035.
• Corresponds to spatial displacement of the wire of about 5 microns.
• Taken as a measure of what the BPM itself is capable of, this can be considered a ‘worst possible’ resolution as specialised readout electronics can only help.
• Linearity– See next slides
July 2008 J.Pater - BPM-based Alignment 7
LHC BPM Linearity± 6mm either side of centre
July 2008 J.Pater - BPM-based Alignment 8
LHC BPM Linearityin 50 steps ~2.5mm from centre
July 2008 J.Pater - BPM-based Alignment 9
LHC BPM Linearityin 10 steps around centre
July 2008 J.Pater - BPM-based Alignment 10
LHC BPM Linearityfurther from the centre
July 2008 J.Pater - BPM-based Alignment 11
LHC BPM Performance
• Based on preliminary study– Resolution (~ a few microns) looks acceptable– Linearity
• Very good over short distances near centre of BPM• Less good further away from centre
– As expected!– Should be repeatable and therefore correctable
• Further work needed to determine– Repeatability correctability accuracy– Other corrections, e.g. temperature dependence
July 2008 J.Pater - BPM-based Alignment 12
Possible Hardware Solutions for BPM Processing Electronics
A. Lyapin (UCL)
• Narrow bandwidth electronics commercially available (i-tech) high resolution as noise is rejected (down to a few um) gain/offset drifts compensation implemented (stable over hours and days!) averaging over a few hundred consequent bunches
• Wide bandwidth electronics single bunch measurement poor single bunch resolution (LHC electronics: ~100 m) Averaging turn-by-turn could improve resolution by sqrt(N) e.g. standard LHC front-end electronics + custom next-level board need to take care of drifts
LHC frontend electronics + specialised next-level board
July 2008 J.Pater - BPM-based Alignment 13
BPM tests: next steps
Have in hand front-end LHC readout boards1) Commission them
• need to bricolage connection to power supplies (don’t have the custom backplane)
2) Test them• Use e.g. LabView to simulate averaging over individual
bunches
3) If that works well, AL to design next-level board.
July 2008 J.Pater - BPM-based Alignment 14
Potential BPM Calibration Scheme• On bench:
– Attach fiducials to outside of BPM– Survey --> position of fiducials wrt WPS sensor and
beam-wire, fold in BPM response– Must be temperature-dependent (e.g. BPM expansion)
• In-situ:– Mount some BPMs on positioners
• calibrate them by offsetting a known amount• Cross-calibrate the others by fitting the orbit
– Inject pulse to compensate for gain/offset drifts (it should last for at least one normal fill) - method studied by T-474 at SLAC ESA (A.Lyapin)
FP420: two ofour BPMs
already move
July 2008 J.Pater - BPM-based Alignment 15
• WPS sensors use a capacitive measurement technique along 2 perpendicular axes.
• On each axis the wire lies between 2 electrodes
• Proven resolution ~0.1-0.3 microns
• LEP energy spectrometer study
• Reproduced on Manchester bench
Wire Positioning Sensors
July 2008 J.Pater - BPM-based Alignment 16
The Moving BitNeed to relate detector position precisely to alignment wire, while allowing detector (on Hamburg pipe) to move freely
–LVDT is an obvious potential solution, but off-the-shelf examples not accurate enough:
• best are ~0.25% of full scale i.e. ~100 on 4cm
–Schaevitz® designed special (rad-hard, very accurate) LVDTs for LHC collimator alignment (see next slide)
• 0.1-0.04% of full scale i.e. 16-40 on 4cm• Compact package (20cm)• Rad-hard to 50 MGy, very good temperature stability• Company confident they can provide shorter version with significantly
better accuracy (at least at one end of stroke.)• Have 2 examples of LHC device at Manchester…
July 2008 J.Pater - BPM-based Alignment 17
July 2008 J.Pater - BPM-based Alignment 18
LVDT study at Manchester
• Resolution
• Accuracy
• Temperature dependence and compensation
July 2008 J.Pater - BPM-based Alignment 19
LVDT Resolution• As expected, resolution is a function of
displacement – (plots show resolution in volts; 10V=25mm)
At 25mm, ~115nm
At 10mm, ~60nm
At centre, ~30nm
July 2008 J.Pater - BPM-based Alignment 20
LVDT Accuracy:• Calibrate by scanning across length of
LVDT, plotting voltage against nominal x position; fit a line to this data.
July 2008 J.Pater - BPM-based Alignment 21
Can get better accuracy near centre by fitting to central points
More work needed: e.g. calibrate at constant temperature
July 2008 J.Pater - BPM-based Alignment 22
LVDT Temperature Dependence• Data taken at several displacements…
– several days per displacement– Tracking room temperature
• …shows clear temperature dependence– Probably more than one effect, e.g.
• Difference in CTEs of support components
• Effects of temperature on electrical characteristics of LVDT (e.g. wire resistance)
July 2008 J.Pater - BPM-based Alignment 23
LVDT Temperature dependence (2)
• Should be correctable. First try:
• Needs more work– Different correction factors for e.g. dT/dt– Better temperature control --> better calibration
• Have programmable ‘oven’ at Manchester
July 2008 J.Pater - BPM-based Alignment 24
Other BPM-based Alignment Jobs
• Integration– Must put together working group to integrate alignment
hardware. Action JP to coordinate, someone from each relevant area.
• DAQ requirements (inputs/outputs) need to be defined
Reserve Slides
July 2008 J.Pater - BPM-based Alignment 26
July 2008 J.Pater - BPM-based Alignment 27
Resolution/Precision/AccuracyA. Lyapin (UCL)
• Resolution – the smallest change of the measured value an instrument can see– depends on the sensitivity, noise/adc bit
resolution• Precision – if multiple measurements of the
same value are taken, how far they fall from each other– mainly depends on resolution and scale
calibration• Accuracy – how far the averaged measured
value is from the true value– depends on the offset calibration, drifts and
non-linearities• Precision and accuracy are usually defined
over some period as they degrade
High accuracy, low precision
From Wikipedia:
High precision,low accuracy
July 2008 J.Pater - BPM-based Alignment 28
Narrow- vs. Wide-Band Electronics
At 420m, will individual bunches be…– …same size, same orbit as overall beam?
• Then use narrow-band solution, nothing to be gained from bunch-by-bunch analysis
– …smaller than beam, each bunch having a stable individual orbit?
• Could win by using wide-band electronics, averaging over ~hundreds of turns for each bunch
• At IP, individual bunch orbits vary by ~1 (for an r.m.s. beam size of 16)
July 2008 J.Pater - BPM-based Alignment 29
Gain/offset drifts compensation A. Lyapin (UCL)
• T-474 experiment at SLAC ESA active monitoring system sending CW burst
into processing electronics when there is no beam induced signal
clear gain drifts have already been observed compensation hasn’t yet been demonstrated method under study