Note: Descriptions are shown in the official language in which they were submitted.
~Z:~933
The present invention relates to a method and appara-
tus for testing transparent ma-terial sheets, particularly flak
glass, for flaws such as foreign substances or ~as bubbles trap
ped in the sheet, in which the material sheet is scanned with a
flying light spot over its width, and the transmitted and/or
reflected radiation is intercepted, converted into e:lectric siy-
nals, and evaluated.
In the present invention transparent material sheets
includes plastics, organic glass and particularly sheet glass.
Since sheet glass is mechanically produced in large quantities as
flat glass in the form of an endless belt, attempts are of course
made to keep the sources of flaws as small as possible. The
greatest need for testing equipment exists in the production of
flat glass. The invention relates particularly to the examina-
tion of flat glass but is not restricted thereto.
In the production of flat glass, for example, on float
glass machinery it still happens that fine, usually bright,
small stones infiltrate the glass sheet despite extreme precau-
tions. Gas bubbles which are in the melt in a finely divided
state constitute a further and also frequent flaw. On reaching
a certain dimension these two flaws result in a surface defor-
mation of the glass sheet although they are completely enclosed
by the glass. Surface deformations can be readily determined by
electrooptical testing devices and methods as described, for
example, in the German Offenlegungsschrift 24 11 407. However,
this does not apply to flaws which are so small that they do
not change the surface of the glass sheet, particularly in the
case of small core bubbles. These core bubbles are not detected
by conventional devices, particularly when the surface tested
is not 100~ clean.
The float glass is tested by scanning the entire width
of the continuously moving material sheet with a flying light
-- 1 --
spot. In order to obtain a high luminous density, the flying
light spot is usually produced by a laser, which is directed onto
a rotating polygon reflector so that because of the high speed
of the polygon the light beam sweeps the flat glass sheet at a
high speed and thus forms the flying light spo-t. ~ portion oE
the beam is reflected at the surface of the glass sheet, a fur-
ther portion enters the glass sheet and is reflected by the under-
surface of the glass sheet, and the major portion of the beam
passes through the glass sheet after reEraction.
It isan object of the present inventionto detect flaws
in a transparent material sheet, which do not result in a surface
deformation of the material sheet. Particularly, so-called core
bubbles, i.e., gas bubbles,which are more or less at the centre
of the material sheet and are very fine so that they are covered
by the layers of the material sheet thick compared with the
sizes of the gas bubbles,clre to be detected.
Accordingly, the present invention provides a method
of testing transparent material sheets for flaws occluded in the
sheet, such as foreign substances or gas bubbles, comprising
scanning the material sheet with a flying light spot over its
width, detecting transmitted and/or reflected radiation, detect-
.ing radiation emerging laterally from the sheet, converting the
detected radiation into electric signals, and evaluating the
sheet with the aid of said signals.
Flaw-testing devices of the kind described in the Ger-
man Offenlegungsschrift 24 11 407 scan with a flying light spot
the width of a material sheet moving usually at a relatively high
speed. The devices have an adjustable sensitivity, which makes
it possible to detect both a flaw on the upperside and a flaw
on the underside when scanning the material sheet from above.
Core bubbles trapped in the material sheet, even fine foreign
substances do not result in a beam deflection as strong as that
~g~3
caused by deformation of the surface. The signal produced by
these flaws is so weak that it corresponds to the signals caused
by fine dust particles deposited on the material sheet. However,
these signals are cut oEf in the evaluation station vf -the test-
ing device. The threshold of this section is vertically adjust-
able. The sensitivity th~ls is adjusted down to such an exte~t
that impuri-ties on the surface do not result in a flaw signal.
However, when a core bubble or a fine small stone is
present, the impinging light spot is deflected in the yas bubble
or on the surface of the small stone. In this case the material
sheet thus acts like a photoconductor. Since the core bubbles,
i.e., the small gas bubbles, as well as the trapped small stones
have a substantially spherical shape, the light beam impinging
on them or entering them is reflected in a manner dependant on
its la-teral motion and the angle of incidence changing therewith
at least once parallel to the scanning line of the material
sheet. It passes to the right or left lateral edge of the mater-
ial sheet, where it is visible as a briefly flashing bright light
spot.
However, data on the flaw per se cannot be provided as
yet, i.e., the size of the flaw cannot be defined on the basis
of this laterally flashing light spot. A decision as to whether
this particular material sheet must be discarded or whether its
use can still be justified because of the minimal size of the
flaw cannot be made either. Therefore, the radiation emerging
laterally from the sheet is intercepted, converted into pulses
and used for controlling the evaluation unit, i.e., that at the
instant when the light spot emerges on one or bo-th sides of the
material sheet to be tested, the flaw can be localized and its
size can be determined. The normal soiling of the surface of
the material sheet to be tested does not result in a flaw indica-
tion.
-- 3
`` ~2~33
The flying light spot is suitably produced by a laser,
since in this manner relatively high energies can be applied,
i.e., relatively broad material sheets can be scanned up to the
edge without power loss. According to a particularly preferred
embodiment of the present invention the flying light spot has
the colour of the material sheet to be tested. Conventional
sheet glass has a slightly yreenish coloration due to slight
traces of iron in the melt. However, this can be detected only
when examining the faces of the sheet glass. Because of the
relatively long path which the liyht spot must cover in the glass
before it is visible laterally, the glass reacts like a colour
filter, i.e., in the case of green coloration, incident red
light, after covering a specific length of path, disappears,
that is to say, it is filtered out. However, light having the
same wavelength, which corresponds to the coloration of the glass,
is not filtered out, but it is subject to the normal absorption
and thus reaches the side faces with a smaller loss.
By employing a light source which radiates at the same
colour as the material sheet to be tested it is possible to
select a lower output of the light source and thus to save energy
and material and simultaneously to ensure optimal efficiency.
However, all transparent materials absorb a certain
proportion of light passing through them. This means that, in
the case of the relatively broad flat glass sheets, whose width
frequently exceeds 3 metres, in the presence of a core bubble,
i.e., a gas bubble at the centre of the sheet, the light reflect-
ed by this bubble must cover a path of approximately 1 50 m to
one of the two sides before it can be absorbed by a photoelectric
converter, generally by a photomultiplier. A substantial absorp-
tion of light thus results, i.e., it is not possible to provideexact data on this flaw just detected without additional ampli-
fication of the pulse emitted by the photomultiplier. However, when
- 4 --
~2~3.;~
-the ligh-t beam travels towards -the edge o the material shee-t an
increasingly distinct flaw signal is obtained even when the size
of -the flaw de-tec-ted and i-ts location in -the sheet axe iden-tical
-to -the flaw at the centre oE the shee-t. ~'ur-therMore, -the ~act
-that the material shee-t -to be tested, i.e., the float yl~ss, is
never 100~ clean must also be considered. Thi~, rneans -that both
the top side and -the bottom side can have du6 t particles, which
can also cause a reElection of the beam into ~he ~lass. Fur-ther-
more, a certain noise level is always present, and this noise
level also varies. For example, when scanning the cen-tre of the
sheet it is substantially lower than when scanning the sheet
edge so tha-t-Elaws which are at the cen-tre of the sheet can be
within the noise level a-t the sheet edge. It thus is important
to suppress the noise level differentially to ensure tha-t a simi-
lar and equally large flaw in -the edge region of the material
shee-t emits the same pulses as a corresponding flaw in -the centre
region of the ma-terial sheet.
Therefore, according to the present
invention each electric lateral pulse as a function of the posi-
tion oE the flying light spot scanning the transparent material
sheet is compared with a selectable value assigned to this posi-
tion, and on exceeding this va]ue a flaw signal is emit-ted.
A suitable embodiment of the present invention provides
-tha-t the selectable value is fed into an electronic memory. This
embodiment is par-ticularly suitable when only one material, for
example, a single grade of glass, is -to be -tested so -that the
material shee-t is no-t subjected to changes wi-th respect to bo-th
the composition and the thickness. In this case it is sufficien-t
to plot an absorption curve of the ma-terial and to store i-t.
In the present case -the -term electronic memories means
semiconductor memories whiich differ according to the circuit
technologies used. Memory types such as RAM, ROM, PROM, or EPROM
-- 5
33
can be used in addition to shift registers. It has been found
that the PROMs - programmable read only memories - are particular-
ly suitable. The PROMs are fixed-value mernories whlch are pro-
vided with the desired bit pattern after the production pxocess
This can be done, for exarnple, by burning specific connectiorls
within the memory. Thus, the program cannot be erascd, i.e.,
that after writing a program into a PROM it is no lvnger p~ssible
to change the information and an accidental change of the kestiny
program is not possible. In a second possibility of programming
the PROMs the highly insulated gate electrodes are u-tilized.
The gate electrodes can be discharged by irradiation with UV
light and recharged by once more applying a correspondingly high
voltage, i.eO, they can be programmed.
In the logic of the evaluating unit the value of the
input pulse is compared with the ~alue of the stored pulses cor-
responding to the scanning position oE the scanning beam and on
exceeding the pulse value a flaw signal is released. Of course,
it is also possible to employ several PROMs which correspond to
different curves. This means that different material sheets can
also be tested after preselecting the suitable memory. In this
case RP~Is are also suitable as memories. These RAMs are not
capable of retaining stored data over a lengthy period but they
can be programmed as desired.
A preferred embodiment of the present invention is
characterized in that a reference beam is separated from the scan-
ning beam forming the Elying light spot and passed over a flaw-
less reference strip of the transparent material strip to be test-
ed. The light emerging at the lateral faces oE the reference
strip is picked up and converted into pulses. These pulses are
compared in their height with pulses obtained fxom the material
to be tested.
This embodiment of the invention assures that exact
-- 6 ~
-` ~IL2~33
values are always attained since the identical ma-terial sheet is
used as the reference strip. Of course, it is also possible to
use a reference strip which substantially corresponds to the
material sheet to be tested, but is no-t completel~ identical
thereto. However, in that case the fact that -the ~eference pulses
obtained do not correspond 100~ lo the pul.~es obtain~d from ~h~
material sheet to be tested must be accep-ted, i.e., a certain
tolerance limit must be accepted.
The lateral pulses are evaluated by means of a trigger
threshold. For this purpose the absorption curve is stored in
a PROM. This process is absolutely independent of the speed,
and it is also free from overshots. By employing several PROMs
it is possible to program also the testing of colored glass.
The lateral pulses are preferably evaluated by opposi-
tion of polarity of the voltage obtained.
Any position of the flying light spot on the ma-terial
sheet, provided the sheet is flawless, produces a pulse which
is identical to the pulse produced from the reference strip by the
separated reference beam at identical position of the flying
light spot. In the flawless state the pulse values thus cancel
each other. Therefore, no deflection results, whereas in the
presence ~f a flaw the pulse values differ. Since the absorption
of the glass is eliminated by the opposition of polarity of the
voltages obtained, the extent of the flaw can be read from the
extent of the pulses, i.e., that equal flaws now also result in
equal flaw signals irrespectively of their position, i.e., re-
gardless of their distance from the edge of the material sheet.
A device for carrying out the process suitably com-
prises at least one testing instrument, which scans -the material
sheet with a flying light spot, a receiver absorbing the reflect-
ed and/or transmitted light, and an evaluating station assigned
to the receiver. The evaluating station has at least one photo-
-- 7 --
33
multiplier disposed laterally of the material sheet to be tested.
The additional installation of a single photomultiplier
on one longitudinal side of the transparent material sheet to
be tested permits the detection o~ core bubbles. The devi~e use~
heretofore on a large scale, for example, a de~ice accordiny to
the German Offenlegungsschrift 2~ 11 407, can be controlled via
the additional photomultiplier and then ~sed ko detect core
bubbles and occlusions in the glass sheet. The photomul-tiplier
is then favourably disposed at the level of the scanning line
traced by the flying light spot on the material sheet to be t,est-
ed since the light beam entering the glass sheet is reflected to
different sides, but the path parallel to the scanning line is
the shortest path so that of all the points at which the light
beam emerges in the region of the lateral edge of the material
the region of the scanning line has the greatest brightness
value and thus yields the strongest and clearest pulse.
A preferred device for carrying out the test is charac-
terized in that the photomultiplier assigned to the face end of
the material sheet to be tested is disposed above the longitud-
inal edge of the material sheet to be tested and half a reflec-
tor surface therebelow.
The light passed through a core bubble or an occlusion
~Z~33
in the material sheet emerges at the untreated material sheet edge,
where it is scattered. Thus it passes ~owards the side as well
as upwards and downwards so that in the ca~e of a p-~rel~ lateral
arrangement of the photomultiplier it will be di~icult to pick up
the light. However, by simply arranying a reflector surfa~e belo~
the longitudinal edge a large proportion of the arnounk of light
emerging from the untreated face end of the material sheet is still
picked up by the mirror and reflected to the photomultiplier dis-
posed above the material sheet. The photomultiplier also is
additionally acted upon with light emerging directly upwards. The
picking-up of light thus is substantially improved.
In a favourable embodiment of the present invention a
reference strip of a material, which corresponds in thickness,
coloration,and composition to the transparent material sheet to
be tested or is identical thereto and extends over the entire width
of the transparent material sheet to be scanned is assigned to
the testing device. On each of its narrow sides the reference
strip is provided with a photoelectric converter. The narrow
sides of said reference strip extend parallelly to the longitudin-
al edges of the transparent material sheet to be tested but incontrast thereto they have been treated so that no undefined scat-
tering occurs. Since the strips also are relatively narrow, the
entire light emerging on these narrow sides can be picked up by
the photoelectric converter.
However, the light enters this reference strip only after
it has been subjected to a special preliminary treatment, i.e.,
in thenormal case, the light would pass through the glass as in
the case of any flat glass and would not be so passed into -the
glass that a substantial percentage thereof would emerge on the
front ends, i~e., in the present case on the narrow sides of the
reference strip.
Therefore, in a preferred development of the present
g
invention the re~erence strip is provided with notches er~ually
spaced apart and extending in the longitudinal direction of the
material sheet to be tested. The notches are a~vantageously sp~c~
ed apart by 5 to 10 mm. Because of -this arrangement pulses which
correspond to the number of notches are ob-tained. Each pulse has
a different value corresponding to its dis-tance frosn the centre
of the material sheet since on approachiny the edge of the refer-
ence strip, i.e., its narrow side, less light is absorbed and a
higher signal thus is near the photoelectric converter.
By so arranging the notches that they are equally spac-
ed apart this signal can be used simultaneously for determining
the position of flaws and a mutual distance of 5 to 10 mm assures
a substantial accuracy in the evaluation of the flaws.
In a further favourable development of the present inven-
tion the reference strip is provided with a delustered line ex-
tending over its entire length. This delustered line is suitably
a sand-blasted area or a transparent adhesive strip applied to
the reference strip. Both the sand-blasted strip and the trans-
parent adhesive strip suitably applied below the reference strip
permit the light to enter the reference strip and thus to be pass-
ed on the narrow sides and to be absorbed in the photoelectric
converter. However, in departure from the notches described above
no current pulse is obtained in the converter, but when the light
beam impinges on the reference strip, a specific voltage which
varies in its value is obtained. This voltage has the lowest
value when the flying light spot reaches the centre of the refer-
ence strip, where the absorption is most intense. For this reason
the reference strip is provided with photoelectric converters on
both sides since only small amounts of light travel from one ma-
terial sheet edge to the other. For each of the two converters acurve established by the converters begins at the centre of the
material sheet with a value which is slightly above zero and with
-- 10 --
~2~3;3
increasing proximity of the scanning light point it increases
towards the sheet edge. For estimating the entire sheet wid-th
the results of the two photoelectric converters, which ~omple-rrlent
each other to the full curve, must be taken into account.
The present invention ~ill now be described in rnore
detail, by way of e~arnple only, with reference to the accornpany~
ing drawings, in which:
Figure 1 is a diagrammatic sketch of a testing device;
Figure 2 shows the -testing apparatus with the reference
strip being scanned;
Figure 3 shows the mirror arrangement at the edge in
detail;
Figure 4 is a chart showing the individual pulses re-
corded by the lateral photomultipliers; and
Figure 5 is a chart showing the straight line resulting
from superimposed curves of opposite polarity with a flaw signal.
The material sheet 1 is reciprocated under the testing
device -2 by means of rollers 8 driven by an electric motor 9.
The testing device 2 contains a receiver 3 for reflected radia-
tion and a receiver 3' for transmitted radiation. The two re-
ceivers are connected to an evaluation unit 4, which is also
acted upon by the photomultipliers 5, 5' laterally disposed
with respect to the material sheet 1.
The laser 14 in the testing device 2 is provided with
a beam splitter 30, which reflects two partial beams 31 and 32
on the rotating mirror drum 15. The partial beam 31 is reflected
as the light spot 10' and the partial beam 32 as the light spot
10. Because of the rotation of the mirror drum 15 the light
beam 32 is passed as a scanning beam 16 over the entire width
of the material sheet 1. The partial beam 31 shown as the spot
10' is simultaneously passed over the reference strip 21 and
enters it through the notch 24. A photoelectric converter 23 or
-- 11 --
` ~2~3~
23' is assigned to each of the narrow sides 22 of the reference
strip 21. The photoelectric converter receives the liyht emerg~
ing from the reference strip 21 and passes i^t on ~ ~e evaluation u~i-t ~.
If the material. sheet 1 contains a Elaw in the ~orrtl of
a core bubble 13, the scanning beam 16 no lonyer reache.s the re-
ceiver 3 as a reflected scanning beam 16', but is deflected as
the light beam 11 and 12 and passed on the scanning line 7 to the
face end 6 of the material sheet 1, where it enters the photomul-
tipliers 5 and 5', which pass the pulse received to the evalu~tion
unit 4. The photomultipliers 5, 5' are connected to the evalua-
tion unit 4 by cables 17 and 18, and the photoelectric converters
23, 23' anologously by cables 33 and 34. Furthermore, an electric
lead 19 extends from the receiver 3 to the evaluation unit 4.
As mentioned hereinbefore, when the scanning beam 16
impinges on a core bubble 13, the light is deflected from the core
bubble 13 and emerges inthe region of the face end 6 of the ma-
terial sheet 1. In the embodiment of Figure 3 there is disposed
in the edge region o the material sheet 1 a substantially hori-
zontally adjusted mirror 27 and a substantially vertically ad-
justed mirror 28. 'rhe mirrors are arranged on a slidable support29 and the two mirrors are so aligned that they deElect light
directed onto them into the photomultiplier 5 disposed above the
edge region of the material sheet.
The partial beam 31 produces the light spot 10' on the
rotating mirror drum 15. The reference beam 20 formed by the
light spot scans 10' the reference strip 21 and enters it at the
notches 24. A pulse is thus produced in the photoelectric conver-
ter 23 by each notch 24. This pulse is recorded in the evaluation
unit 4 and compared with the corresponding values determined by
the photomultipliers 5, 5'. When the material sheet 1 is free
from flaws the values determined are identical, i.e., they do not
differ from each other. When using a reference strip 21, which has a
- 12 -
~2~33
delustered line or an adhesive strip, the evaluation is analoyous~The light passes along the delustered line or adhesive strip an~
enters the reference strip 21 and leaves it on i~s narrow side~
22, where it is received by the photoelectric converters 23. How-
ever, a voltage which varies with the travel o the reference beam
and can be plotted as a curve is obtained instead of a pulse.
Figure 4 shows an absorption curve 35, which was plotke~
along the spikes of the individual pulses 36 produced by the notch-
es 24 in the reference strip 21 by the reference beam 20. The
noise level curve 37 has been plotted below the absorption curve
35 in the same Figure. This noise level curve is caused substan-
tially by impurities on the upper side and bottom side of the
glass and is of no importance in the process accordiny to the
present invention. However, it is clearly evident from the curve
that the noise level is substantially higher in the edge region
of the sheet so that flaws which can occur in the centre region
of the sheet are superimposed by said noise level.
Figure 5 shows the opposite polarity of the absorption
curve 3S, which was obtained via the reference strip 21 with
underlying adhesive strips, with the scanning curve shown there-
above. The scanning curve 38 has a flaw signal 39, which in its
absolute height is smaller than the values in the edge region of
the curve. The opposite polarity straight line 40 results from
the opposite polarity. The flaw signal 39 distinctly projects
from said straight line.
- 13 -
