LOCATING LCD GLASS FOR ROBOTIC HANDLING
J. A. Smith
Advanced Engineering, Corning Incorporated, Coming, NY 14831
ABSTRACT. The display industry is trying to improve yields by refining the automation processes in
the manufacturing of Liquid Crystal Display (LCD) Glass. To help achieve better yields, it is
necessary to locate with better precision the LCD panels with respect to robotic tools.
Two
approaches will be discussed to make ranging measurements on LCD Glass: a laser confocal
displacement sensor and ultrasonic rangers. The knowledge gained from these measurements will
improve the positioning of the robotic tools with respect to the LCD glass during the manufacturing
process.
INTRODUCTION
Corning and Coming's customers are trying to improve yields by refining the
automation processes in the manufacturing of Liquid Crystal Display (LCD) Glass. To
help achieve better yields, it is necessary to locate with better precision the LCD panels
with respect to robotic tools. As LCD glass becomes thinner, the structural rigidity of the
glass goes down and the LCD panels deform more under the panels own weight during
process steps. Thus the need to locate LCD Glass in space to assist in the robotic handling
of glass during a manufacturing process is the motivation behind this research.
The unique characteristics of LCD glass makes optical and vision based sensing
challenging because LCD glass is optically transparent, optically smooth, thin (<lmm) and
particulate free. During the manufacturing of the large LCD glass sheets (>1 x 1 m), the
glass is moving, hot (>100 °C), and contained within a dynamic processing environment.
By combing the characteristics of LCD glass and the harsh manufacturing environment,
the sensing of position and orientation would be challenging using any measurement
method.
To protect the pristine nature of the LCD glass and be effective in a manufacturing
environment, additional requirements are placed in the measurement method. The sensor
must be non-contact and provide for good sensing contrast. The measurement system must
be robust, easy to use, easy to implement, easy to maintain, and be cost effective.
This paper discusses two approaches which are currently being pursued to make
ranging measurements on LCD Glass. The first approach is based off a laser confocal
displacement sensor. This optical sensor will be used offline to measure the mechanical
deformation of the glass sheets when supported on two edges. The second approach to
locating LCD glass is based on commercial ultrasonic ranging sensors and is intended to
be used as an in-process sensor. The knowledge gained from these measurements will
CP657, Review of Quantitative Nondestructive Evaluation Vol. 22, ed. by D. O. Thompson and D. E. Chimenti
© 2003 American Institute of Physics 0-7354-0117-9/03/S20.00
1705
improve the positioning of the robotic tools with respect to the LCD glass during the
manufacturing process.
LASER CONFOCAL DISPLACEMENT SENSOR
At many LCD display manufacturers, the LCD glass substrates are stored in
cassettes that support the LCD glass at the edges. The glass substrates deform under their
own weight as shown in Figure 1. The fact that the glass substrates deforms or sags is not
in of itself an issue. The true issue with sag is variability in LCD glass sheets between
batches and vendors. As LCD glass becomes thinner, the structural rigidity of the glass
goes down and the variability increases. Thinner glass compounds the issue.
A measurement bench has been designed to measure deformation at points through
out the glass sheet. The measurement bench will sample LCD glass sheets off line to track
variations in LCD glass deformation with time. The sensor is based on a confocal laser
displacement sensor. Figure 2 shows the design of the sensor and the salient specifications
[1]The active confocal principle works in the following manner [1]. A laser beam is
focused on the glass surface through an objective lens that vibrates up and down by being
attached to a tuning fork. The surface can be specular or diffuse. If the target is
transparent, either the front surface or the back surface can be used. The focussed laser
light is reflected off of the target surface and back into the sensor head. The light is then
redirected by half mirrors to converge on a pinhole in front of a photodiode. The
maximum voltage output from the photodiode determines the tuning fork's exact position
when the laser beam is focused on the target surface. The distance to the target surface is
then calculated.
By placing the sensor in a 3D gantry system, the out of plane deformation can be
imaged as shown in Figure 3. The LCD sheets can then be sampled as they are
manufactured and the deformation of the LCD sheets can be tracked. When the variability
of the sheets reaches a threshold, this information can be fed back into the LCD glass
manufacturing process to reduce the deformation variability of the LCD glass.
FIGURE 1. When LCD glass is supported by the edges, the glass sheet deforms due to gravity forces.
1706
• 7 µm beam spot size
7 µm beamof
spot
•• Resolution
0.2size
µm
•• Resolution
7 pm beamof
spot
size
• Range ±1 mm 0.2 µm
• Range
Resolution
of 0.2pm
±1-mm
•• Stand
off
28 mm
• Stand
Rangeoff
±1-mm
28 mm ±7O
•• Angular
Acceptance
Stand off-28
mm ±7O
•• Angular
Acceptance
• Angular Acceptance ±7°
FIGURE 2. Diagram showing the laser confocal sensor and the salient specifications.
FIGURE 2. Diagram showing the laser confocal sensor and the salient specifications.
FIGURE 2. Diagram showing the laser confocal sensor and the salient specifications.
S ag E xample
S ag E xample
Sag Example
0
0
-2
-2
Sag (m m )
Sag (m m )
Sag ( m m )
-4
-4
-6
-6
-8
-8
XXSam
m ))
Sampple
le Dim
Dim ee n s io n (m m
X S a m p l e Dimension ( m m )
475
475
400
400
325
325
8
250
250
CM
0
175
175
LO
100
100
25
25
-10
-10
150
150
150
500
500
500
Sam
p le
Y YSam
p le
Dim
esnio
spie
io
YeSam
Dim
n
nn
(m(m
mm
) )
Dimension (m m)
FIGURE3.3.Graph
Graphshowing
showingthe
the measured
measured deformation
deformation of
two
edges
from
FIGURE
of aaLCD
LCDglass
glassplate
platesupported
supportedalong
along
two
edges
from
FIGURE
3. Graph
showing the
measured deformation of a LCD glass plate supported along two edges
the
laserconfocal
confocal
displacement
sensor.
the
laser
displacement
sensor.
from the laser confocal displacement sensor.
1707
ULTRASONIC RANGING
Over time during the manufacture of LCD Glass, the LCD glass can significantly
"walk" from its original starting position and robotic tools can loose their alignment. Thus
a sensing system that can locate the LCD glass in space will be useful to maintain the
robotic alignment to the LCD glass. The work presented in this section discusses the
development of a non-contacting ultrasonic ranger to locate LCD glass in a harsh
manufacturing environment.
Commercial transducers have been tested for this application and work well when
the LCD glass is at room temperature. When the glass is being measured during the
forming process, the glass is moving, vibrating and is at temperatures in excess of 200°C.
Under these conditions, the commercial sensors become unreliable. This section discusses
the initial development of a robust ultrasonic ranger.
Prior internal results show that the major causes of ultrasonic ranging instabilities
in commercial ultrasonic rangers are due to dynamic thermal gradients that cause
amplitude fluctuations in the return ultrasonic signal. The commercial sensors use a signal
threshold to determine the time of arrival of the ultrasonic pulse. If the amplitude of the
ultrasonic signal changes, the relative location of the threshold within the returning
ultrasonic signal will change. The change in the threshold location relative to the
ultrasonic waveform will cause an apparent change in target range. Thermal effects can be
reduced by having a smaller beam size and shorter stand off distances. However,
occasional spiking still can occur. To further reduce the thermal effects and enable larger
stand off distances, a more sophisticated algorithm that continuously tracks the reflected
ultrasonic signal will be discussed.
Figure 4 shows the experimental setup to develop an ultrasonic ranger that is robust
with respect to thermal gradients. The LCD glass is heated by laboratory hot plate. The
hot plate is mounted such that the "hot plate" is vertical and normal to the ultrasonic
propagation direction. A thermo-couple and volt meter is used to measure temperatures.
The LCD Glass (305 X 305 X 0.7 mm) is mounted vertically and also normal to the
ultrasonic propagation direction. Temperature is measured by placing the thermo-couple
between the glass and the hot plate. The thermo-couple is held in place by the pressure
between the glass and hot plate on the lead wire insulation. It should be noted that the
thermo-couple is not in direct contact with the glass or the hot plate.
The ultrasonic transducer is mounted to a translation stage. The mounting fixture is
designed to ensure that the ultrasonic beam intersects the glass plate and hot plate
normally. The beam is centered on the targets. The stand off distance is nominally 1094
mm. To minimize ultrasonic side lobe [2] reflections off of the optical table between the
transducer and the target, ultrasonic absorbing foam is placed on the table between the
transducer and the target.
Figure 5 shows a typical return ultrasonic signal. The return signal consists of a
pulsed 450 kHz sinusoid with an amplitude envelop that is approximately Gaussian in
shape. As a first step to improve the reliability of the ultrasonic ranger, a Gaussian is fit to
the return signal. The center location in time of the Gaussian envelope should be
independent of ultrasonic amplitude.
1708
LCD Glass
305 X 305 X 0.7mm
Hot
Plate
Ultrasonic
Ranger
1094mm
67.5 mm
67.5 mm
Acoustic
Absorber
Translation Stage
with Micrometer
FIGURE 4. Diagram of the setup for the ultrasonic ranging experiments on LCD glass.
Ultrasonic Ranging: Return Echo, 23OC
Ultrasonic Ranging: Return Echo, 23°C
4
RF Return
Gauss Fit
3.5
3
Voltage (V)
2.5
2
1.5
1
0.5
0
2.14
2.14
2.16
2.16
2.18
2.18
2.2
2.2
2.22
2.24
2.22
2.24
Time (ms)
2.26
2.26
2.28
2.28
2.3
2.3
2.32
2.32
Time (ms)
FIGURE 5. Plot of the return ultrasonic pulse that consists of a 450 kHz carrier frequency modulated with a
FIGURE
5. Plot
of the return
ultrasonic
that consists
of a 450 kHz carrier frequency modulated with
near Gaussian
envelope
and the
resultingpulse
Gaussian
fit.
a near Gaussian envelope and the resulting Gaussian fit.
The ranging results for the LCD plate at room temperature for various displacements
from the nominal standoff distance of 1094 mm is shown in Figure 6. The return echo timing
is shown to be a linear relationship with travel distance. The measured velocity of 346.6 m/s in
air is in good agreement with the theoretical value of 345.1 m/s at 23 OC in dry air [3]. The
0.4% difference might be due to higher humidity in the lab. Table 1 lists the salient statistics
from the room temperature testing. The displacement standard of deviation is an indication of
the resolution of the ranger. At room temperature using a Gaussian envelope to determine
1709
return echo time, the resolution of the ultrasonic
ranger is 0.2 mm.
Ranging results for the LCD plate at 331OC for various displacements from the nominal
standoff distance is shown in Figure 7. The return echo timing is shown to be much noisier
than the room temperature timing. The linear relationship with travel distance is not as
compelling but the measured velocity of 343 m/s in air is still in good agreement with the
theoretical value of 345.1 m/s at 23 OC in dry air. Since the transducer is being moved within
room temperature air, the measured velocity should match the theoretical at room temperature.
Table 2 lists the salient statistics from the 331OC temperature testing. At a 331OC temperature,
the resolution of the ultrasonic ranger is 0.7 mm. The displacement resolution at high
The ranging results for the LCD plate at room temperature for various
displacements from the nominal standoff distance of 1094 mm is shown in Figure 6. The
return echo timing is shown to be a linear relationship with travel distance. The measured
velocity of 346.6 m/s in air is in good agreement with the theoretical value of 345.1 m/s at
23 °C in dry air [3]. The 0.4% difference might be due to higher humidity in the lab.
Table 1 lists the salient statistics from the room temperature testing. The displacement
standard of deviation is an indication of the resolution of the ranger. At room temperature
using a Gaussian envelope to determine return echo time, the resolution of the ultrasonic
ranger is 0.2 mm.
Ranging results for the LCD plate at 331°C for various displacements from the
nominal standoff distance is shown in Figure 7. The return echo timing is shown to be
much noisier than the room temperature timing. The linear relationship with travel
distance is not as compelling but the measured velocity of 343 m/s in air is still in good
agreement with the theoretical value of 345.1 m/s at 23 °C in dry air. Since the transducer
is being moved within room temperature air, the measured velocity should match the
theoretical at room temperature. Table 2 lists the salient statistics from the 331°C
temperature testing. At a 331°C temperature, the resolution of the ultrasonic ranger is 0.7
mm. The displacement resolution at high temperature is a factor of three worse than the
room temperature resolution.
o
Measured
MeasuredRelative
RelativeDisplacements
Displacementson
on23
23CCLCD
LCDGlass
Glass
160
166
Relative Travel Distance (mm)
UT Ranging
Linear Fit
Error Bars
y =y -=346.6*x
+ 2226
- 346.6*x
+ 2226
140
140
120
120
100
100
8080
6060
4040
2020
00
-20
-20
-40
-40
66
6.05
6.05
6.1
6.1
6.15
6.15
6.26.2
6.25
6.25
6.36.3
Time
TimeofofEcho
Echo(ms)
(ms)
6.35
6.35
6.46.4
6.45
6.45
6.56.5
FIGURE
data
taken
from
LCD
glass
at at
room
temperature.
TheThe
speed
FIGURE6.6.Plot
Plotshowing
showingthetheultrasonic
ultrasonicranging
ranging
data
taken
from
LCD
glass
room
temperature.
ofspeed
soundofinsound
air is in
determined
by the slope
of slope
the line
to fitted
the measured
data points.
air is determined
by the
of fitted
the line
to the measured
data points.
TABLE 1. Table showing the measurement statistics for ultrasonic ranging on LCD glass at room
temperature.
Parameter
Stand off distance
Round trip return time
Standard of deviation (s)
Measured velocity
Standard of deviation (m)
Gaussian Fit
Measured Value
1094 mm
6.42 ms (average from 5 trials)
0.9 µs
346.6 m/s (theoretical 345.1 m/s)
1710
0.2 mm
0.5 Max RF envelope value
SUMMARY
To help achieve better yields in the manufacture of Liquid Crystal Display glass, it is
necessary to locate with better precision the LCD panels with respect to robotic tools. As LCD
glass becomes thinner, the structural rigidity of the glass goes down and the LCD panels
TABLE 1. Table showing the measurement statistics for ultrasonic ranging on LCD glass at room
temperature.
Parameter
Measured Value
1094mm
Standoff distance
6.42 ms (average from 5 trials)
Round trip return time
Standard of deviation (s)
0.9 jis
Measured velocity
346.6 m/s (theoretical 345.1 m/s)
Standard of deviation (m)
0.2 mm
Gaussian Fit
0.5 Max RF envelope value
contact measurement are three-fold: 1) for laser systems, nothing, such as a sticker, needs to be
placed on the sample – this made 3D measurements a possibility – in the same vein, we do not
add any mass whatsoever and change the sag magnitude; 2) we do not deflect the sample and
change
sag magnitude
would happen
with
contact probe;
process is
TABLEthe
2. Table
showing theas
measurement
statistics
foraultrasonic
ranging 3)
on the
LCDmeasurement
glass at room 331°C.
more efficient since there is no sample prep other than cutting the glass to the desired size.
Parameter
Measured
Value ranging sensors to locate LCD
The second approach discussed is based
on ultrasonic
1094mm
Standoff distance
glass and is intended to be used as an in-process sensor. A commercial ultrasonic ranger in the
Round trip return time
6.45 ms (average from 5 trials)
presence
of temperature gradients has been modified to reliably determine the return echo time
Standard of deviation (s)
4.0 {is
by fitting a Gaussian envelope to the return ultrasonic signal. This modified transducer has
Measured velocity
343 m/s (theoretical 345.1 m/s)
been
shown
to have (m)
a resolution of 0.7 mm 0.7mm
with a LCD glass temperature of 331OC at a
Standard
of deviation
standoff
of 1 m. The knowledge gained
measurement
Gaussiandistance
Fit
0.5from
Max these
RF envelope
value systems will be used
to improve the positioning of the robotic tools with respect to the LCD glass during the
manufacturing process.
oo
MeasuredRelative
RelativeDisplacements
Displacementsonon331
331CCLCD
LCDGlass
Glass
Measured
20
20
Ranging
UTUT
Ranging
Linear
Linear
FitFit
Error
Bars
Error
Bars
Max
Signal
>1.0V
Max
RFRF
Signal
> 1.0
V
y = - 343.2*x + 2213
Relative Travel Distance (mm)
15
15
10
10
5
5
5
i.
0
-5-5
-10
-10
6.4
6.4
6.41
6.42
6.41
6.43
6.44 6.43
6.45
6.42
6.45
Time
of
Echo
(ms)
Time of Echo (ms)
6.46
6.476.47
6.44
6.46
O
FIGURE
ranging
data
taken
from
LCD
glass
at 331
C. TheThe
speed
of sound
in
FIGURE7.7.Plot
Plotshowing
showingthetheultrasonic
ultrasonic
ranging
data
taken
from
LCD
glass
at 331°C.
speed
of sound
air
by the
of the
to thetomeasured
data points.
in isairdetermined
is determined
byslope
the slope
ofline
the fitted
line fitted
the measured
data points.
1711
SUMMARY
To help achieve better yields in the manufacture of Liquid Crystal Display glass, it
is necessary to locate with better precision the LCD panels with respect to robotic tools.
As LCD glass becomes thinner, the structural rigidity of the glass goes down and the LCD
panels deform more under the panels own weight and during processing steps. Two
measurement systems have been presented in this paper to sense the position of LCD glass.
Both measurement sensors are non-contact to protect the pristine nature of the LCD glass.
The first approach discussed is based off a laser confocal displacement sensor.
This optical sensor will be used offline to measure the mechanical deformation of the glass
sheet when supported on two edges. The sensor has been integrated into a 3D gantry to
image the out of plane deformation with a resolution less than ten micrometer. The
advantages of non-contact measurement are three-fold: 1) for laser systems, nothing, such
as a sticker, needs to be placed on the sample - this made 3D measurements a possibility in the same vein, we do not add any mass whatsoever and change the sag magnitude; 2) we
do not deflect the sample and change the sag magnitude as would happen with a contact
probe; 3) the measurement process is more efficient since there is no sample prep other
than cutting the glass to the desired size.
The second approach discussed is based on ultrasonic ranging sensors to locate
LCD glass and is intended to be used as an in-process sensor. A commercial ultrasonic
ranger in the presence of temperature gradients has been modified to reliably determine the
return echo time by fitting a Gaussian envelope to the return ultrasonic signal. This
modified transducer has been shown to have a resolution of 0.7 mm with a LCD glass
temperature of 331°C at a standoff distance of 1 m. The knowledge gained from these
measurement systems will be used to improve the positioning of the robotic tools with
respect to the LCD glass during the manufacturing process.
ACKNOWLEDGEMENTS
The author would like to thank Brian Strines and Monroe Marlowe for developing
the sag measurement bench and for providing the sag data. The author would also like to
thank Lori Hamilton, Display Technologies and Corning Incorporated for proving the
support necessary to write this paper.
REFERENCES
1.
2.
3.
Keyence, Laser Confocal Displacement Meters, LT Series, Cat. No. LT-2.5M-699,
Keyence Corporation of America, 50 Tice Blvd., Woodcliff Lake, NJ 07675, 1999.
Kinsler, L. E. , Frey, A. R., Coppens, A. B., Fundamentals of Acoustics, John
Wiley & Sons, New York, 1999.
Nave, C. R., HyperPhysics, Georgia State University, http://hyperphysics.phyastr.gsu.edu/hbase/sound/souspe3 .htmlffcL 2000.
1712