G.Sirri – INFN Bologna videconf 2008-03-06 1/14 Fiducial Mark Finding with ESS Lateral Mark Finding [FlexMapX]: Recommended release 42b (2007-10-25)

G.Sirri – INFN Bologna videconf 2008-03-06 1/14

Fiducial Mark Finding with ESS

Lateral Mark Finding [FlexMapX]:

Recommended release 42b (2007-10-25) successfully used by several labs despite bugs and lack in integration with the CI.

Next release 52 (with many improvements) declared as recommended (2007-12-21) withdrawn for bugs.

The new development version 57 available. Test in progress.

What’s New:

New Lateral mark finding strategy

Search of “Cloned” vertical strip

Unified Fiducial Mark finding module for the ESS,Front X-ray Spot + Optical Spot + Lateral X-ray + Edge Finding tools

More details in: http://www.bo.infn.it/opera/scanning/lateralmarks/

G.Sirri – INFN Bologna videconf 2008-03-06

FlexMapX v57 (Development)

*News*- fit of the strip segments with Parametric Line Fit - search of "clone" vertical strip [to do: take decision] - new map strings "mapid:" for identical transformation and “edge”- auxiliary init map window

*bugs fixed* - product operator matrix * vector - map string: mark side flag are "1" [top] or "2" [bottom]- "Computer freeze during monitor creation" bug fixed

*source code improvements* - Migration to Visual Studio 2005 SP1- Separation between steering code, hardware macros and user interface. Steering code without MFC/COM.

2/14


New Lateral Mark Finding Strategy

1. Scan along a perpendicular path(a)

2. Scan alternately on one and on the other side of S position(b)

5

2

4

NominalNominal edge edge

Segment S found here

ExpectedExpected Mark Position Mark Position

XX

1

6

7

(a) The trajectory lays between the nominal edge and the expected mark position. The path length is “Scan Transverse Tol.” parameter.

(b) The number of segments to be searched for is “Max Strip candidates” parameter (S included).The interdistance between segments is “Strip Sampling Distance” parameter.

2 14 3

8

9

Y strip case

3

3. Segments are fitted to a line.

Same for the other strip. Mark position: intersection point between fitted lines.

3/14


Automatic clone check: after the first segment of a vertical strip is found (2 or 3), the system will search for the other one.

If both strip (2 and 3 in the picture) are found, the closest-to-the-edge is automatically taken.

If only one is found … to be decided!

1.Pause the process, wait for the user

2.Go On, but put a warning (where?)

Now is simply “Go On”.

The “Clone” problem

4/14


What is expected to work differently

FlexMap3 evaluates the Nominal-Stage Transformation with the 3 marks nearest-to-the-fragment Different transformations in different plate positions [or only 3 marks for the whole emulsion plate].

FlexMapX fits the transformation parameters with all available marks. One global transformation valid for the whole emulsion plate.

Features not implemented in FlexMapX (yet):

"Sheet is rotated by 180°“

Recovery File

Map load from a *.map fileyou MUST write the mark map string in the initialization window:

Spot Mark Finding [FlexMapX <-> FlexMap3]

5/14


FlexMapX FlexMap3

not implemented

not implemented

Implemented inside the code

6/14


“To Do” List

Tool for creating MARK MAP [referred to the edge or to a previous map].

- refresh monitor before start a new mark map search- focus at each lateral mark site- 180 degree sheet rotation and recovery option

- new parameter: minimum number of mark to be found in order to validate the affine- new parameter: maximum number of degree-of-freedom of the affine

speed up the lateral mark searching:- during the first strip segment search: stop the segment search after one candidate is found- optimize parameters (number of segment vs segment interdistance)

speed up the spot mark searching:- focus at each site and scan only one image for trial [at present: scan a thickness every trial]

7/14


Mark Finding ReproducibilityPreliminary Test

8/14


Lateral Mark - Test

2 marks

Test: 72 measurements (2008-03-03) 2 failure (clone)

Mark 1 : RMS_x = 3 µm RMS_y = 3 µmMark 2 : RMS_x = 2 µm RMS_y = 2 µm

X (µm)Y (µm)

Y (µm)X (µm)

9/14


Lateral Mark - Test

Problems of stability during the test (temperature?)

Mark 1 - X (µm)

Mark 2 - X (µm)

Mark 1 - Y (µm)

Mark 2 - Y (µm)

#Trial

103

103

103

10/14


Lateral Mark – Test

Spread of the predicted position of the 4 corners:

#1: <X> = 3689.79 X_RMS = 4.5 <Y> = 4808.38 Y_RMS = 4.2#2: <X> = 128839 X_RMS = 2.8 <Y> = 4484.8 Y_RMS = 3.3#3: <X> = 3947.58 X_RMS = 5.0 <Y> = 104648 Y_RMS = 3.6#4: <X> = 129097 X_RMS = 2.3 <Y> = 104324 Y_RMS = 2.4

1 and 3 are far from the mark position....

12

34

11/14


Optical Mark - Test

6 marks

Test: 114 measurements (2008-03-04)

#1: <X> = 17703 X_RMS = 0.6 <Y> = 16369.1 Y_RMS = 0.6#2: <X> = 17483 X_RMS = 0.2 <Y> = 96766.6 Y_RMS = 0.6#3: <X> = 52745.5 X_RMS = 0.3 <Y> = 96889.1 Y_RMS = 0.6#4: <X> = 122876 X_RMS = 0.3 <Y> = 97085.8 Y_RMS = 0.7#5: <X> = 123141 X_RMS = 0.2 <Y> = 16632.6 Y_RMS = 0.5#6: <X> = 70472.7 X_RMS = 0.3 <Y> = 16542.8 Y_RMS = 0.5

X (µm)Y (µm)

Y (µm)X (µm)

X (µm)Y (µm)

X (µm)Y (µm)

X (µm)Y (µm)

X (µm)Y (µm)

12/14


Optical Mark – Test

Spread of the predicted position of the 4 corners:

#1: <X> = 8060.8 X_RMS = 0.4 <Y> = 6102.8 Y_RMS = 0.6#2: <X> = 133177 X_RMS = 0.2 <Y> = 6147.2 Y_RMS = 0.5#3: <X> = 8020.9 X_RMS = 0.4 <Y> = 105914 Y_RMS = 0.6#4: <X> = 133137 X_RMS = 0.3 <Y> = 105959 Y_RMS = 0.7

12

34

13/14


Conclusions

The development of FlexMapX is going on Bugs fixed and new lateral mark finding strategy Integration with Spot Mark Next step: assistant to create mark map

Feedback is welcome

Test of mark finding reproducibility Optical map RMS ~ 0.5 µm

no problems with 6-marks transformation

Lateral map RMS ~ 2.5 µm Further test to optimize parametersprediction spread larger in the opposite side w.r.t. the marks

14/14

G.Sirri – INFN Bologna videconf 2008-03-06 1/14 Fiducial Mark Finding with ESS Lateral Mark Finding [FlexMapX]: Recommended release 42b (2007-10-25)

Documents