HEXAGON TITLE SLIDE - NCSL International

Report
Alternate Method to Certify Grid Plates and
Line Scales using a CMM with Chromatic
Confocal Sensor verses Traditional Methods
Hexagon Metrology
By: JHorwell
Sr. Metrologist
July 2013
What may I Learn or Why Listen?
• Alternate method for certifying Line Scales
• How to apply a white light confocal sensor for edge
detection
• Difference between displacement and intensity
• Density of the data
• Selection Criteria
2
2013 NCSLI International Workshop and Symposium
AGENDA
01 Overview of Requirements
02 Brief History
03 Test Piece
04 Results
05 Error Budget
06 Lessons Learned
3
2013 NCSLI International Workshop and Symposium
Overview
• Test the ability of a confocal chromatic sensor to find
edges
• Review existing Hexagon Hardware and firmware to
accomplish this task
• Compare distance measurements against certified
artifact
• Determine the feasibility to certify grid plates and
glass line scales on a high accuracy CMM with
confocal chromatic sensor
4
2013 NCSLI International Workshop and Symposium
History
• Currently most line scales are measured on special
purpose hardware consisting of an XY stage, Laser
Interferometer feed back for position and an optical
measuring device
• Others use comparative techniques from a certified
master scale to the scale under test
• Some have combinations of manual and automatic
systems
• In the past some used high accuracy CMM with an image
analysis system to certify grid plates
5
2013 NCSLI International Workshop and Symposium
200mm Glass Scale with Chrome Lines
• Task is to be able to find
the edges with a
standard deviation less
than 0.09um
• Calculate the distances
and compare these to a
certified value.
6
2013 NCSLI International Workshop and Symposium
Equipment Selection Criteria
• Small Measurement Uncertainty ≤ 0.3
+ L/1000
• Non Contact Sensor Availability
• Stability over time
• Hardware, Software and Firmware
available and capable for finding
edges
7
2013 NCSLI International Workshop and Symposium
Leitz PMM C 12107 Infinity
• Measuring Error MPE in [µm]
according to ISO 10360-2 (20106)
• E0
≤ 0.3 + L/1000
• E150
≤ 0.5 + L/1000
• R0
≤ 0.3
• Scale Resolution
• 4nm
• LSP –S4 Probe head
with integrated
optional Precitec LR
Sensor
8
2013 NCSLI International Workshop and Symposium
Precitec LR Sensor Fully Integrated into Leitz PMM C Infinity
•
•
•
•
•
9
Numeric aperture
Measuring procedure
Measurement angle relative to surface
Measuring distance
Diameter of measuring spot
2013 NCSLI International Workshop and Symposium
0.66
chromatic confocal
90° ±40°
6.5 mm
1.4 µm
90 degree Sensor
• We are continually developing
sensors and adaptors for our
systems
• The sensor to the right is similar to
the sensor we used but can
measure the line scales in the
vertical direction verses horizontal
• The sensor can be manually
rotated and our controller can
handle more than one sensor on a
single system
10
2013 NCSLI International Workshop and Symposium
Overview Glass Line Scales
• Find the edges of a Chrome Lines on Glass Line Scales using
the Precitec LR Chromatic Confocal Sensor on the Leitz infinity
12107 in RI
•
•
•
•
•
Build a reference system on the Glass Scale using tactile probing
Build an alignment system using the Precitec LR Sensor
Measure each reference line via 3-pts and constructing an axis
Intersect each reference line with the alignment axis
Calculate the distance from each line to the zero line and compare
with certificate of certification
• Test 2-methods for finding the edge
• Test alignment stability
• Run repeat test to determine standard deviations
11
2013 NCSLI International Workshop and Symposium
Test Part – Glass Line Scale
• Certificate of Calibration No 115-01538 by CIPM MRA
• Uncertainty of Measurement: U = 0.10µm + L/2000
• The distances between the graduations were measured with a photo
mask measuring system consisting of a precision x-y bearing table, a
two-axis differential plane mirror interferometer and a microscope. The
table moves the graduations to the focus of microscope while the
interferometer measures its position. The relative position of a
graduation within the field of view the microscope is determined by
digital image analysis.
• The scale was cleaned, supported at the Bessel-points (distance
121mm) and measured with the graduation on top using episcopic
white light illumination. For each graduation the position of the trailing
edge was measured at the height of the collinear alignment marks
12
2013 NCSLI International Workshop and Symposium
What was missing from the hardware configuration
• While we could find the edge of the lines using
traditional analysis by automatically finding extreme
points with a standard deviation between 112nm and
250nm using a 30 repeat test. We determined that
these values were not as good as we believe our
equipment was capable up.
• Our Firmware Development Group made a flag within
our firmware that allowed us to receive the intensity
out verses displacement from the Precitec LR Sensor
• This was the break thru that we
needed to accomplish our task
13
2013 NCSLI International Workshop and Symposium
Alignment & Fixture Procedure For Glass Scale
•
•
•
The Glass Scale was place on 3-spherical pins which were near the airy points with 2
at one level and a single pin at the other airy point.
2-posts were used to align the scale along its axis and a single pin for the stop on the
zero end plane of the Glass Scale
Tactile alignment
1. Using the LSP-S4 probe head and ruby sphere probe we measured a plane on the top and
long side and a single point on the end face. This provided us with the rough alignment
before we picked the Precitec LR Sensor from our Probe Change Rack
•
Optical Alignment
1. Scanned two lines on the top surface of the Glass Scale, constructed a plane and defined the
Z-axis spatial vector and alignment from that.
2. We measured 3-points on the line between the 0 and 10mm lines and constructed axis
3. Measure same on the last line between the 190 and 200mm marks
4. Constructed an axis from these two points which define the rotational portion of our
Coordinate system
5. Measured 3-points and constructed an axis on the back side of the zero line at ±0.5mm from
the axis just created
6. Intersected this zero line axis with the line the defined the rotational portion of our Reference
System and this became the X-zero of our Glass Scale
14
2013 NCSLI International Workshop and Symposium
Alignment of Glass Scale using Precitec LR Sensor
Found 3-edge points at ±0.5mm from zero
alignment axis on the 0-position vertical line and
built an axis and the intersection of that axis and
the zero alignment line created the XY origin
0
10
190
Found 3-edge points at 0.5mm
spacing on each of the 2-horizontal
lines and constructed an axis between
then which defined the axial alignment
15
2013 NCSLI International Workshop and Symposium
200
Measurement Sequence
• We started our measuring sequence by measuring the Zero Line using 3points; constructing an axis and intersecting the newly measured axis with
the directional axis of our reference system and got a single point for each
line measured.
• We basically stepped and repeated this process until all lines were
measured
• We then calculated the distance in X-direction between each line measured
and the Zero Line.
• We repeated this process 10 times
16
2013 NCSLI International Workshop and Symposium
Finding the edge
• What are some of the issues with finding the real edge and then correlating
this to traditional methods.
•
•
•
•
17
Edge is not perpendicular to the reference plane
Edge is not straight
Does back lighting verses top lighting give two different edges
Can we find the same edge as traditional optical systems and measuring
microscopes
2013 NCSLI International Workshop and Symposium
Finding extreme points on the edge
Edge Detail
Pos 1
Pos 2
Mean
Δdist
18
15.17795
15.18129
15.17962
0.00334
2013 NCSLI International Workshop and Symposium
Summary of Results
Results from 30 Repeats (mm)
Element Y1 Point Y2 Point Y3 Point Y4 Point
Range
0.000635 0.000428 0.001127 0.000420
Stdev
0.000150 0.000112 0.000265 0.000123
19
2013 NCSLI International Workshop and Symposium
Current procedure
• The 3 slides following this show the results from finding the edge of a
single chrome line using ScanOnLine with the following point
spacing:
• 0.0001 = 10,000pts/mm
• 0.00005 = 20,000pts/mm
• 0.000025 = 40,000pts/mm
• I am hoping to have even lower standard deviations that I originally
obtain using the Precitec LR sensor on Infinity
• Currently I am finding the edge by selecting points based on light
intensity of 0.26 to 0.35
• When I am on the chrome surface I get saturation which is near a
value of 1
20
2013 NCSLI International Workshop and Symposium
Finding the Edge Alignment Line
• Top picture illustrates the edge as
seen by the CMM system in red
• The blue line is the change in
intensity as the system crosses the
edge
• The small red points on top of the
blue line are the points used to
define the edge based on intensity
value.
Y-Axis Intensity
-15.17914
0.24
-15.17911
0.24
-15.17908
0.23
-15.17906
0.23
-15.17903
0.24
-15.17900
0.24
21
2013 NCSLI International Workshop and Symposium
• Lower picture shows the measured
edge from the CMM in Red
• The line in black shows the intensity
curve from the Crocodile box. The
change in intensity is from glass
surface to the silver plated surface
Density at 10,000pts/mm
Chrome Surface
Glass Surface
Measured Curve
Intensity Points
Selected
Scan Direction
22
2013 NCSLI International Workshop and Symposium
Results
Delta (microns)
Correlation with Original Certification Tolerance Bands +/- (0.3 + L/1000)um
y = 0.0004x
1.00
0.80
0.60
0.40
0.20
0.00
-0.20
-0.40
-0.60
-0.80
-1.00
25 run
+ Tol
-Tol
1st 10 runs
Linear (25 run)
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
Distance (mm)
• The red line represents average value from 10 repeats – the certified
value. I did not apply temperature compensation at the moment but
could and most likely will.
• The Blue dots represent the results from the 25 runs
23
2013 NCSLI International Workshop and Symposium
Results - numerical
Infinity w/Precitec 23-Apr-2013
Nominal (mm) Meas-Cert (µm) 2 sigma (µm)
0
0.00
0.123
10.0
-0.10
0.073
20.0
-0.02
0.101
30.0
-0.19
0.121
40.0
-0.23
0.131
50.0
-0.07
0.130
60.0
-0.08
0.138
70.0
0.00
0.131
80.0
-0.14
0.123
90.0
0.09
0.174
100.0
-0.04
0.151
110.0
-0.15
0.179
120.0
0.08
0.160
130.0
-0.07
0.119
140.0
-0.05
0.141
150.0
0.03
0.138
160.0
0.09
0.132
170.0
0.12
0.144
180.0
0.20
0.135
190.0
0.23
0.165
200.0
0.24
0.177
24
2013 NCSLI International Workshop and Symposium
Still not Satisfied!
Noise reduction algorithm affects the edge
detection!
25
2013 NCSLI International Workshop and Symposium
Raw Data From CMM with out Noise Reduction
26
2013 NCSLI International Workshop and Symposium
Selection based on Intensity Data Shown has NO Noise Reduction
1µm
The points inside this window
represents selected data based on
intensity:
Blue = 0.5 to 0.6 Intensity
Blue + Red = 0.5 to 0.71 Intensity
The points inside this window
represents original selection for
intensity:
Green = 0.5 to 0.6 Intensity 0.26 to
0.35
Intensity
Location
range
0.5-0.6
0.5-0.71
0.26-0,35
27
2013 NCSLI International Workshop and Symposium
(mm)
79.999732
79.999760
80.000352
Selection based on Intensity Data Shown has NO Noise Reduction
Affect of Intensite on Distance
Delta from Cert (um)
0.3
0.2
0.1
.5-.6
0
-10
-0.1
.5-.71
10
30
50
70
90
.26-.35
-0.2
-0.3
Distance (mm)
• The original testing presented in this paper used the 0.26 to 0.35 intensity selection
criteria and noise reduced data from the CMM
• We wanted to test raw data with no noise reduction and a steeper point on the
intensity curve.
•
•
Selected 0.5 to 0.6 Intensity
Selected 0.5 to 0.71 Intensity
• Results are taken weeks apart with completely new setups and the standard
deviations for this newest procedure has a maximum σ of 0.55µm
28
2013 NCSLI International Workshop and Symposium
Intensity Selection Comparison
Distance
(mm)
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
29
Effect of Intensity Selection
Sigma (mm)
Δ Cert (um)
0.260.50.50.260.5-0.6
0.5-0.7
0.35
0.6
0.7
0.35
0.000024 0.000014 0.000025 -0.15 -0.14
-0.11
0.000036 0.000039 0.000053 -0.06 -0.05
-0.03
0.000047 0.000042 0.000046 -0.22 -0.21
-0.20
0.000043 0.000040 0.000049 -0.26 -0.25
-0.21
0.000038 0.000051 0.000061 -0.14 -0.14
-0.05
0.000044 0.000041 0.000056 -0.08 -0.08
-0.05
0.000042 0.000051 0.000047 0.04 0.05
0.02
0.000046 0.000044 0.000066 -0.11 -0.11
-0.11
2013 NCSLI International Workshop and Symposium
Straightness Plot of the Scan (YZ plane)
30
2013 NCSLI International Workshop and Symposium
Verification results are equal in the + and – Scan Direction
31
2013 NCSLI International Workshop and Symposium
Verification results are equal in the + and – Scan Direction
32
2013 NCSLI International Workshop and Symposium
Optiv Performance 2z443
• Measuring Error MPE in [µm]
according to ISO 10360-7
• EX , EY
≤ 2.2 + L/125
• Standard Scale Resolution
• 0.5µm
• Optional Scale Resolution
• 0.1µm
33
2013 NCSLI International Workshop and Symposium
OPTIV Results
Results from OPTIV
5
4
3
Delta from Cert (um)
2
1
Optiv Data
OPTIV Uc
0
Optiv -Uc
Infinity
-1
-2
-3
-4
-5
0
20
40
60
80
100
120
Distance (mm)
[2.2 + L/125]µm
34
2013 NCSLI International Workshop and Symposium
140
160
180
200
Error Budget
Average difference NIST vs NPL
NPL Calibration Uncertainty NIST Uncertainty Infinity Accuracy Limits
0.0012
Average Deviation (mm)
0.0007
0.0002
Avg
NIST
AVG
NPL
-0.0003
-0.0008
-0.0013
Nominal Step
• Environment 20°C±0.1°C 10%CTE of 8.3*10-6/°C = 0.016µm
• Reproducibility – actual 25 repeats plus realignment = 0.083µm
• CMM Accuracy Parallel to X extrapolated from physical measurements [0.2 +
L/1500]µm
• Other considerations
•
•
Quality of edge
Squareness of measured lines to center axis
• UC = [0.3 + L/1000]µm with coverage factor of 2
35
2013 NCSLI International Workshop and Symposium
Summary
• We wanted to show that one could calibrate a Glass Line Scale and or a
grid plate via a non traditional method such as presented within this
paper. We hope that the community will consider this alternate method
and hope to engage in some round robin testing to verify that a high
accuracy CMM with chromatic confocal sensor could be used and that
is process could allow users to verify their working standards without
the need for special purpose systems. It also adds validity for owning
instruments of this caliber.
• We are not claiming that this method is faster than a tradition noncontact CMM, but we do feel that we can reduce the measurement
uncertainty for calibrating longer artifacts with a significant lower
measurement uncertainty and without the need for a dedicated single
purpose system.
36
2013 NCSLI International Workshop and Symposium
Lessons Learned
• It is possible to find an edge using a chromatic confocal sensor
using only displacement
• Light intensity changes are a better method to find the edges
than using displacement values only
• Quality of the edge has a significant impact to our standard
deviation
• Use of noise reduction algorithms has some impact on the
standard deviation, but changing the selection criteria had very
little impact
• Correlation to alternate calibration methods are easy to reach
with the Leitz Infinity with integrated Precitec LR chromatic
confocal sensor
37
2013 NCSLI International Workshop and Symposium
The End
Thank you!
38
2013 NCSLI International Workshop and Symposium

similar documents