Download presentation
Presentation is loading. Please wait.
1
University of Wisconsin-Madison
Cocoa ME+1 Blessing James N. Bellinger University of Wisconsin-Madison 6-April-2009 James N. Bellinger 20-March-2009
2
Data used 0T 3.8T PG Distancemeter 16-Nov average DCOPS 11-Nov event
Link Oct 3.8T Distancemeter 1-4 Nov average DCOPS Oct event Link Oct PG PG within disk UR-0058 (2006) (Oleg cleaned it up) Supplementary UR-0103 (2008) PG of disk UR-0124 (after Craft) James N. Bellinger 20-March-2009
3
Cocoa Fit Types Ideal 0T 3.8T Special
Ideal Geometry for Endcap+Link, default data 0T Data from 0T, Link fit geometry/data from 0T Transfer plates from PG, rest of Endcap ideal 3.8T Data from 3.8T, Link fit geometry/data from 3.8T Special Transfer plates from PG, initial chamber pos PG James N. Bellinger 20-March-2009
4
Cocoa Hierarchy: A CMS SLM Transfer Plate Chamber 3 type Laser
SLM Ref DCOPS Transfer Ref DCOPS Laser Z-sensor Chamber 3 type Outer DCOPS Inner DCOPS James N. Bellinger 20-March-2009
5
Cocoa Hierarchy: B CMS SLM-continued Chamber 2 type Link Ring
Outer sensorbox Outer ASPD DCOPS Inner sensorbox Inner ASPD Link Ring Link lasers MAB Z-sensor laser target James N. Bellinger 20-March-2009
6
Cocoa Ideal Fit vs DDD Only 6 entries. Cocoa Ideal minus DDD geometry
Ring 3 only Cocoa Ideal geometry fit is fine: “chi-squared” is with 872 “degrees of freedom” Cocoa pos – DDD pos Mean, microns RMS, X -17 69 Y -55 52 Z -7 1 James N. Bellinger 20-March-2009
7
Chamber Z deviations Cocoa 3.8T and 0T vs Ideal
0T Fit -Ideal 3.8T Fit-Ideal X Y Z ME+1/3/03 2.78 -2.00 -1.24 2.77 -1.97 -3.17 ME+1/3/09 2.87 -0.77 -4.50 2.73 -0.73 -4.70 ME+1/3/14 -1.12 0.37 -3.39 -0.98 0.50 -3.60 ME+1/3/20 -2.52 -1.13 1.36 -2.60 -0.84 ME+1/3/27 0.95 -2.22 2.54 1.33 -2.32 -0.62 ME+1/3/33 -0.43 -1.60 8.25 0.10 -1.09 -0.92 ME+1/2/02 1.30 -3.07 1.06 1.38 -3.35 -7.18 ME+1/2/08 2.94 -1.96 -0.68 3.00 -1.95 -8.04 ME+1/2/14 0.59 -1.21 -1.00 0.72 -1.08 -7.81 ME+1/2/20 -1.18 -0.46 2.36 -0.20 -5.87 ME+1/2/26 1.39 0.05 5.55 1.50 0.02 -3.11 ME+1/2/32 0.04 -0.26 5.21 -0.07 -0.40 -3.74 Cocoa 3.8T Cocoa 0T Cocoa Ideal Ideal fit uses ideal geom and nominal measurements HSLM6 bad due to blocked IR target James N. Bellinger 20-March-2009
8
Chamber center Z deviations
Cocoa Fit 3.8T - Fit 0T mm Fit 0T - P.G. ME+1/3/03 -1.93 -1.39 ME+1/3/09 -0.20 0.33 ME+1/3/14 -0.21 0.25 ME+1/3/20 -2.20 -1.81 ME+1/3/27 -3.16 -3.27 ME+1/3/33 -9.17 4.06 ME+1/2/02 -8.24 -0.90 ME+1/2/08 -7.36 1.94 ME+1/2/14 -6.81 2.82 ME+1/2/20 -8.23 -5.34 ME+1/2/26 -8.67 0.45 ME+1/2/32 -8.96 4.02 The Cocoa 0T fits are not far from the PG numbers The 1_2 chamber deviations with field agree w/ Celso's numbers The HSLM6 fits are bad because of a blocked IR target Rms=1.4 Rms=2.9 James N. Bellinger 20-March-2009
9
Fit Ring (average of all chambers) Position Deviations from Ideal
0T-Ideal X Y Z 3.8T- Ideal +1/3 .59 -1.17 -1.05 .65 -1.07 -2.59 +1/2 1.01 -1.33 1.46 1.07 -1.31 -6.40 PG (disk) .58 -1.37 0.57 NA James N. Bellinger 20-March-2009
10
ME+1/3 chamber tilts (mrad)
3.8T-0T ME+1/3/03 -0.40 2.43 2.83 ME+1/3/09 -0.71 1.85 2.56 ME+1/3/14 -0.88 0.89 1.77 ME+1/3/20 0.39 2.04 1.65 ME+1/3/27 -1.73 -0.55 1.18 ME+1/3/33 2.10 -2.10 AVERAGE -0.67 1.33 2.00 At disk top At disk bottom Tilts (mrad) determined from DCOPS Z positions at upper and lower ends of each chamber James N. Bellinger 20-March-2009
11
PG targets and Cocoa 0T Fits: Z of DCOPS dowels
XFER PG Pred 1/3Out 1/3In XFer Cocoa Coco- PG HSLM1 -0.79 -1.21 -1.37 HSLM2 4.53 8.48 1.80 HSLM3 2.32 2.12 0.51 HSLM4 -1.19 -0.83 -1.86 HSLM5 -0.53 6.39 -0.93 HSLM6 8.78 7.19 1.59 Uses the DCOPS PG targets to predict the DCOPS dowel positions for the Xfer DCOPS and the ME+1/3 DCOPS Different target holders at ME+1/3/09_outer and ME+1/3/27_outer?? Inconsistent James N. Bellinger 20-March-2009
12
DCOPS from PG vs Cocoa 0T Fit Summary
DCOPS Dowel positions: 0T Cocoa fit – predicted from PG of DCOPS targets Reference: mean= 0.87, rms=2.21mm ME+1/3_outer: mean= 2.99, rms=3.86mm ME+1/3_inner: mean= -0.37, rms=1.34mm HSLM6 is not included RMS is large, and at least partly attributable to PG problems “Reference” = reference DCOPS on transfer plate James N. Bellinger 20-March-2009
13
Deviations from Ideal Chamber mounting errors: should not exceed a few mm PG measurement errors: supposedly 300 microns but I don’t believe that anymore Cocoa fitting errors Real distortions because of the field James N. Bellinger 20-March-2009
14
Cocoa Estimated Errors
Cocoa returns some estimated errors for quantities in the coordinate system of the mother volume (Cocoa uses a hierarchical system description) If I assume that off-diagonal entries are 0, I can transform this to the CMS coordinate system I have no sense of how well Cocoa estimates errors James N. Bellinger 20-March-2009
15
3.8T Cocoa ME+1/3 Chamber Centers
mm, Cocoa errors X Y Z ME+1/3_03 ± .15 ± .28 ± .09 ME+1/3_09 ± .30 ± .17 ± .09 ME+1/3_14 ± .24 ± .21 ± .09 ME+1/3_20 ± .12 ± .30 ± .09 ME+1/3_27 ± .30 ± .12 ± .09 James N. Bellinger 20-March-2009
16
Now Compare Cocoa to DDD
Cocoa errors and chamber mismounts both contribute to this Remove overall disk rotation and translation to get a picture of the internal shifting Only 6 chambers available for ME+1/2 Only 5 chambers for ME+1/3 (PT6 bad) Does NOT display chamber tilts James N. Bellinger 20-March-2009
17
Expect Z shift of ring due to disk bending will be gone
Rotation of disk will be gone Chamber mismounting, sensor mismeasure, and Cocoa fit error will remain James N. Bellinger 20-March-2009
18
ME+1/3 deviation changes with field
5 measured centers Overall rotation and translation is removed No more than a few dozen microns difference between the patterns found with field off and field on Max dev =1.6mm Animated cm James N. Bellinger 20-March-2009
19
Cocoa Estimates Cocoa vs Ideal deviation RMSs are comparable to and smaller than (on the average) PG vs Ideal deviation RMSs: next slide’s table Cocoa better than PG? Deviation averages aren’t always 0 because of missing measurements BUT Cocoa may be biased to finding things close to the ideal, since the ideal geometry is one of the inputs! James N. Bellinger 20-March-2009
20
“Cocoa(0T) vs Ideal” vs “PG vs Ideal” Variation of Deviations
PG Apin ME+1/2 ME+1/3 X devs 0 ± 1.2 0 ± 0.8 0 ± 0.7 Y devs 0.1 ± 0.7 0.2 ± 0.6 0.2 ± 0.9 1.1 ± 1.5 Z devs 0 ± 0.4 -0.5 ± 0.8 3.1 ± 6.0 1.8 ± 5.5 James N. Bellinger 20-March-2009
21
Check for Bias Create a new 0T SDF file using PG measurements instead of Ideal geometry as the starting point for chamber positions Compare fits from this special run to the normal 0T run James N. Bellinger 20-March-2009
22
Special (PG) 0T – normal 0T
X Y Z ME+1/3_03 2.13 0.39 0.01 ME+1/3_09 ME+1/3_14 -0.76 0.73 0.13 ME+1/3_20 -2.64 -0.7 0.31 ME+1/3_27 -0.39 -1.86 -0.01 ME+1/3_33 0.52 -0.66 PG not available Rms=.15 James N. Bellinger 20-March-2009
23
Special 0T – normal 0T: notes
The difference between using PG and Ideal geometry as a starting point has little effect on the Z fit: 10 microns in most places HSLM2 did not have good PG measurements for the alignment pins, so the Special run used Ideal measurements X and Y are not well constrained without the presence of the Transfer Lines. The fact that the Z measurement is bad at PT6 is irrelevant to this comparison, which studies fit stability James N. Bellinger 20-March-2009
24
Special (PG) 3.8T – Ideal 3.8T X Y Z ME+1/3_03 2.136 0.388 0.016
-0.003 0.000 -0.002 ME+1/3_14 -0.759 0.725 0.134 ME+1/3_20 -2.638 -0.703 0.319 ME+1/3_27 -0.385 -1.865 -0.012 ME+1/3_33 0.099 -1.085 -0.917 James N. Bellinger 20-March-2009
25
Cocoa Input All measurements to be blessed use
PG initial starting positions for transfer plate PG initial starting positions for chambers Link fit position for link disk Link fit position for MAB laser line Calibrated ASPD positions wrt P4 Calibrated DCOPS CCD positions wrt dowel James N. Bellinger 20-March-2009
26
Cloud on the horizon: ME+1/2 chamber Z centers
Fits My B=0 Celso B=0 Diff B=0 My B=3.8 Celso B=3.8 Diff B=3.8 ME+1/2_02 .86 .88 ME+1/2_08 .78 .77 ME+1/2_14 .67 .61 ME+1/2_20 -.09 -.15 ME+1/2_26 .51 .48 ME+1/2_32 .21 .23 James N. Bellinger 20-March-2009
27
Why the difference? Not sure yet Change with field is the same
James N. Bellinger 20-March-2009
28
Conclusions Cocoa fit for ME+1/3 chambers is stable with respect to initial conditions in Z Photogrammetry includes spurious outliers Cocoa deviations from the ideal are tighter than PG deviations, even if PG values were the starting point James N. Bellinger 20-March-2009
29
Blessing for ME+1/3 chamber Z
0T Pos mm 0T Tilt mrad 3.8T Pos mm 3.8T Tilt mrad ME+1/3_03 -0.40 2.43 ME+1/3_09 -0.71 1.85 ME+1/3_14 -0.88 0.89 ME+1/3_20 0.39 2.04 ME+1/3_27 -1.73 -0.55 Average -0.67 1.33 Δ from nominal -0.94mm -0.67mrad -4.08mm 1.33mrad James N. Bellinger 20-March-2009
30
Photogrammetry errors for the Z of the alignment pins are not 300μ
Evaluate the PG Photogrammetry errors for the Z of the alignment pins are not 300μ Loveless says the pins were not inserted to nominal depth James N. Bellinger 20-March-2009
31
PG targets on chambers Targets on DCOPS (not used in next slide)
Targets on alignment pins Coded targets on chambers Use alignment pins to define chamber axis Use X/Y of coded target to predict a Z Compare predicted w/ measured Z James N. Bellinger 20-March-2009
32
Coded Target Z – Predicted Z
ME+1/3 chambers Alignment pins used to predict Z of coded target given its X/Y Rms=1.4mm Looks like a single distribution, NOT a narrow one with a few typos mm James N. Bellinger 20-March-2009
33
Crosscheck coded targets
Oleg says some were on wrong chambers Use his corrected table Look at deviation of coded target from alignment pin axis line Nothing looks badly wrong; largest deviation is 145mm from axis (min 75mm) James N. Bellinger 20-March-2009
34
DCOPS targets DCOPS on Transfer Plate, chamber 3 outer and chamber 3 inner have three 1.27mm PG targets on top. These were included in the survey. In the following table the three measurements were averaged for each of the 18 visible DCOPS James N. Bellinger 20-March-2009
35
Variation of PG Z for DCOPS
Ref Ave Rms 3 out 3 in HSLM1 0.169 0.097 0.193 HSLM2 0.037 0.385 0.198 HSLM3 0.054 0.067 0.197 HSLM4 0.040 0.099 HSLM5 0.082 0.737 0.148 HSLM6 0.092 0.238 0.302 PG target position 3-point ave/rms James N. Bellinger 20-March-2009
36
DCOPS PG Variation Along Line
HSLM1 HSLM2 HSLM3 HSLM4 HSLM5 HSLM6 Ave Z Rms Z James N. Bellinger 20-March-2009
37
Evaluation of DCOPS targets
Consistency of measurement: The Transfer Plate DCOPS are measured significantly better than the rest HSLM5 outer DCOPS are not very consistent Consistency along line: Chamber mounting variations contribute! HSLM2 and HSLM5 show unreasonably large fluctuations James N. Bellinger 20-March-2009
38
Chamber surface Z’s from PG
Apin outer inner Coded DCOPS 3 outer 3 inner Diff outer Diff inner HSLM1 HSLM2 NA -0.84 HSLM3 -696.3 HSLM4 -0.09 HSLM5 HSLM6 -700.4 Rms=.53 Rms=.37 James N. Bellinger 20-March-2009
39
Z’s from PG vs data HSLM5 outer chamber 3 DCOPS measurements are clearly out of line The DCOPS readings from HSLM5 correspond to corrected values shown at right. Not much variation XFer 3 Out 3 In 2 18.98 16.72 17.10 18.26 mm, corrected data values James N. Bellinger 20-March-2009
40
PG Conclusions Assuming the Alignment pin and coded target errors are comparable, the variation on these is 1mm and not 300 microns. If coded error=300μ, Apin error is 2mm If the variation is due to random errors: for a DCOPS target at Transfer Plate: 140μ Outer chamber edge: 470μ Inner chamber edge: 350μ Disregard PG measures with large disagreements with either other PG measurements or with data? James N. Bellinger 20-March-2009
41
Displays Omitting lines illustrating chamber surface
Triangles show the slope well enough PG information not displayed Diagram is very cluttered already HSLM1-5 are animated to show 0 to 3.8T shifts HSLM6 has bad data for the DCOPS at 3.8T and bad Z information for the distancemeter James N. Bellinger 20-March-2009
42
Distancemeter and dists Chamber surface estimates Red=Real Green=Sim
DCOPS dowels Chamber surface estimates Red=Real Green=Sim ME12 ASPD IR target MAB ASPD ASPD P4 Animated James N. Bellinger 20-March-2009
43
Animated James N. Bellinger 20-March-2009
44
Animated James N. Bellinger 20-March-2009
45
Animated James N. Bellinger 20-March-2009
46
Animated James N. Bellinger 20-March-2009
47
3.8T is bad IR target obscured, Z is bad
James N. Bellinger 20-March-2009
48
Animated James N. Bellinger 20-March-2009
49
Animated James N. Bellinger 20-March-2009
50
Animated James N. Bellinger 20-March-2009
51
Animated James N. Bellinger 20-March-2009
52
Animated James N. Bellinger 20-March-2009
53
3.8T is bad IR target obscured James N. Bellinger 20-March-2009
54
Blessing for ME+1/3 chamber Z
0T Pos mm 0T Tilt mrad 3.8T Pos mm 3.8T Tilt mrad ME+1/3_03 -0.40 2.43 ME+1/3_09 -0.71 1.85 ME+1/3_14 -0.88 0.89 ME+1/3_20 0.39 2.04 ME+1/3_27 -1.73 -0.55 Average -0.67 1.33 Ave Δ from nominal -0.94mm -0.67mrad -4.08mm 1.33mrad James N. Bellinger 20-March-2009
55
Caveats The average Z position omits hSLM6, and can therefore be biased wrt disk rotations I have no error estimate for the tilts yet Error estimates for the Z’s follow James N. Bellinger 20-March-2009
56
Error Estimates Cocoa estimates the error in Z to be 90 microns
Probably the usual artificial error fitters’ return I can’t trust the PG values for Alignment pins There are problems with PG values for the DCOPS targets The PG coded targets require some convolution to get to the appropriate Z location: show later James N. Bellinger 20-March-2009
57
Error Estimates: outer Z
The outer position Z measurement is limited by MAB laser line uncertainty Assumed perfect Tolerances in machining 50 microns is what we usually claim Tolerances in carbon fiber thickness 100 microns maximum → 30 microns rms Paper target thickness 50 microns? Tolerances in transfer plate 50 microns Measurement error Seems to be negligible Add in quadrature for 90 microns James N. Bellinger 20-March-2009
58
Error Estimates: Inner Z
Inner Z measurement uncertainty driven by mismatch with Celso’s fit We agree to a few dozen microns for ASPD positions, but not on the chamber center! This is an artificially large error estimate Assume Link system Z estimates are perfect Ratio of distances scales the effect on ME1/3 chambers by .36 James N. Bellinger 20-March-2009
59
Error Estimates: Inner Numbers
Average disagreement w/ Celso on chamber center 470 microns (call this systematic) RMS of disagreement 350 microns Contribution to error on ME+1/3 center is 36% (ratio of distances) 130 micron plus 170 micron systematic error James N. Bellinger 20-March-2009
60
Error Estimates: Fit Summary
Quadrature of inner mismatch, outer error, cocoa fit 180 microns with 170 micron systematic error Assumes perfect Link results for MAB and LinkDisk James N. Bellinger 20-March-2009
61
Error Estimates: Coded Targets
Coded targets are not at center Use fit Z and tilt to predict Z at coded target position (about 110 micron change) Mean difference = .387mm (ignore hSLM6) RMS = 1.47mm Known uncertainties: Post-craft disk rotation .6mm Post-craft disk position .5mm Nominal coded target Z .3mm Worse than DCOPS target ests James N. Bellinger 20-March-2009
62
DCOPS PG targets For a DCOPS target at
Transfer Plate: 140μ Outer chamber edge: 470μ Inner chamber edge: 350μ Chamber center variation of 410 μ Exaggerated because of DCOPS PG errors James N. Bellinger 20-March-2009
63
Error Summary From Cocoa and boundary estimates From Coded Target PG
180 microns plus 170 microns systematic From Coded Target PG 1470 microns plus 390 microns systematic From DCOPS target PG 410 microns The Cocoa + boundary estimate is probably too conservative I don’t know why the coded target PG is so large James N. Bellinger 20-March-2009
64
Tilt Error No estimate as yet James N. Bellinger 20-March-2009
65
BACKUP MATERIAL James N. Bellinger 20-March-2009
66
Method for Predicting Z from PG
Get PG (X,Y,Z) wrt disk center from UR-0058 or UR-0103 Rotate disk as specified in UR-0124 Translate disk as specified in UR-0124 (Post-Craft numbers from UR-0124) James N. Bellinger 20-March-2009
67
PG Issues Photogrammetry is not always correct
James N. Bellinger 20-March-2009
68
PG errors and chamber mismounts
PG deviations from Ideal include PG error, typos, and wrong targets Real chamber mismount Overall shifts and rotations of the disk Subtract the overall shifts and rotations to get a better picture of the PG errors and mismount errors In what follows PG Chamber centers are derived from alignment pin locations James N. Bellinger 20-March-2009
69
PG vs DDD, ME+1/2 Chamber centers Overall rotations and translations
are removed Deviations combine PG error and chamber mounting Max x/y dev is 2.2mm cm James N. Bellinger 20-March-2009
70
PG vs DDD, ME+1/3 Chamber centers Overall rotations and translations
are removed Deviations combine PG error and chamber mounting Max x/y dev is 2.6mm Still a tilt? cm James N. Bellinger 20-March-2009
71
PG to DDD summary Deviation of PG from standard geometry in the X/Y plane is at most 2.2mm for ME+1/2 and 2.6mm for ME+1/3. RMS for X deviations is .7 for ME+1/2 .8 for ME+1/3 RMS for Y deviations is .9 for ME+1/2 1.5 for ME+1/3 RMS for Z is about 6. and 5.5mm James N. Bellinger 20-March-2009
72
Z’s from PG vs data The HSLM6 outer Z seems out of line with the rest in the line, but agrees with the alignment pin estimate Data shows O(4mm) deviation at 3 Outer also PG deviation is OK XFer 3 Out 3 In 2 18.32 15.79 21.32 23.45 mm, corrected data values James N. Bellinger 20-March-2009
73
Comparisons of Ideal with Cocoa Rings
James N. Bellinger 20-March-2009
74
0T ME+1/2 Cocoa vs Ideal 6 measured centers Overall rotation
and translation is removed cm James N. Bellinger 20-March-2009
75
0T ME+1/3 Cocoa vs Ideal 5 measured centers Overall rotation
and translation is removed cm James N. Bellinger 20-March-2009
76
3.8T ME+1/2 Cocoa vs Ideal 6 measured centers Overall rotation
and translation is removed cm James N. Bellinger 20-March-2009
77
3.8T ME+1/3 Cocoa vs Ideal 5 measured centers Overall rotation
and translation is removed cm James N. Bellinger 20-March-2009
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.