3.2.4.1
`
`'Masking'
`
`For many of the procedures detailed, it is necessary to 'mask' the data. In essence, this
`removes unwanted data from the image. In the example below, a printed dot is shown,
`masked, and urunasked, Figure 3-9.
`
`Figure 3-9 - unmasked
`
`Figure 3-9 - masked
`
`Figure 3-9 - Masked and unmasked images
`
`Masking may be performed automatically, by height, co-ordinate, or by defining
`either a specified shape (rectangle, circle etc) or by defining an irregular polygon, as
`required. The technique is used to isolate cells (see Appendix l), or to isolate dots
`(see section 4.4.2.4) as well as to remove rogue data from measurements. For the
`analysis of dots, polygons were created to isolate the dot. In the case of cells, either a
`polygon or a circle was used depending on the size of the cells, and the amount of
`space surrounding the detail in question.
`
`3.2.5 Cell Volume
`
`Three different interferometers were available for use, and were used as appropriate.
`The first is a large, laboratory based interferometer. The interferometer itself is based
`on an air-table, minimising vibration, and offers the highest flexibil ity. Multiple
`internal magnifications are available, and several lenses may be mounted on the
`measurement turret. This interferometer was used in the analysis of diamonds.
`Although it is the most flexible in terms of measurement options, it is not possible to
`
`46
`
`FAST FELT 2010 , pg. 61
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`mount a gravure cylinder under the measurement turret. The second is a portable unit.
`Utilising no internal magnification, and a CCD camera of the same resolution as the
`laboratory based unit this device was used primarily for measurements taken on
`cylinders (cells or surface finish). The third device is another portable device, utilising
`O.Sx internal magnification. This device uses a lower resolution CCD, offering a
`lower horizontal resolution, and also has a lower vertical resolution. This device was
`used where excess vibration occurs, as it is less sensitive to vibration than the higher
`resolution portable device.
`
`As previously indicated in 3.2.3, traditional volume measurements are taken by
`measuring the length and width of a cell using optical microscopy, and inferring a
`volume from these measurements and the stylus angle. Microscope images of cells are
`shown in Figure 3-10- below.
`
`Figure 3-10- Microscope image of cell
`
`If the stylus angle, e,
`formula;
`
`is known, the volume of the cell is calculated, using the
`
`I
`~ume=
`(
`
`height x width) ( width
`)
`x
`2
`2tan(B/2)
`
`l
`x-
`3
`
`Equation 3.1 -Traditional volume calculation
`
`Results from this formula will be compared with results generated from the method
`that was developed based on actual cell geometry measurement.
`
`As a starting point, measurement of engraved cells made with the system as supplied
`is shown in Figure 3-ll. Due to low reflectance, optical noise and an inability of the
`
`47
`
`FAST FELT 2010 I pg. 62
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`lens to collect light back from the sloped cell sides, approximately 99% of the data
`
`from inside the cell is missing (left image).
`
`Figure 3-1 1 - Interferometric measurement of gravure cells
`
`In the example shown above (Figure 3-11), black indicates missing data. The only
`
`information available for calculation is the apex of the cell and the base corresponding
`
`to the roll surface. There is a requirement to interpolate the remaining data to
`
`calculate the volume within the cell. Also in Figure 3-ll , the right hand image shows
`
`a measurement taken with an improved optical system. Now 99% of the data from
`
`inside the cell is present, and calculation of actual volumes based on measurement can
`
`be performed, rather than inferring it. The optical configuration was improved by
`
`using a higher magnification lens. This allowed the measurement of sloped surfaces
`
`(such as the inside of a cell) and by minimising the number of cells being measured,
`
`the amount of optical noise is reduced.
`
`Because the sides of the cells are sloped, most of the light is not reflected back up into
`
`the interferometer, but bounces off the other walls, causing optical noise. By
`
`increasing the intensity, and decreasing the number of cells, the opportunity for noise
`
`is reduced, and more light is reflected from the cell walls, allowing measurement
`
`(Figure 3-l l, right hand image). With this much higher data rate, this allows accurate
`
`calculation of volume rather than the inferences, which are characteristic of previous
`
`works.
`
`48
`
`FAST FELT 2010, pg. 63
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`Previously several different possible techniques had been attempted, including coating
`
`the surfaces with carbon to reduce the reflectance, and minimise noise, and by
`
`removing filters from within the interferometer, to reduce the coherence length, and
`
`thus improve the sensitivity of the interferometer to measure the sloped cell walls.
`
`However, these methods had not provided data of sufficient quality to facilitate actual
`
`volume measurement.
`
`An analysis package was supplied with the interferometer to allow calculation of the
`
`specific volume of anilox rolls. This may, according to documentation, also be used
`
`for analysis of gravure cells. However, several assumptions are embodied in the
`
`measurement and data reduction. The analysis package's volume measurement system
`
`was re-evaluated with the new optical system. A much higher data rate was achieved,
`
`covering just one cell, however as the automatic analysis needs at lest 16 cells,
`calculations could not be performed using the supplied analysis package. It was also
`
`discovered on analysis of the automatic function that incorrect values (for example, an
`
`unrealistic cell volume) are often calculated for reasons that could not be established.
`
`Therefore a new methodology was required.
`
`c
`
`Surface
`roughness
`
`A
`
`Figure 3-12 - Dia.gram of Surface and Cell
`
`To facilitate accurate volume measurement. roll surface roughness needs to be taken
`
`into account, see Figure 3-12. If the volume was calculated below level 'c' this would
`
`include volume in the land surface of the cylinder in addition to the cell volume.
`
`Therefore before a volume measurement can be calculated, a series of surface levels
`
`49
`
`FAST FELT 2010, pg. 64
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`must be defined, the volume below which can be calculated. The total volume
`required will be the sum ofvolume V 1 and volume V2 in Figure 3-12 - i.e. from the
`apex of the cell to level 'C'. It is necessary first however, to define the 'land level' -
`i.e. the modal height for the unengraved, top surface of the cylinder, level 'A' in
`Figure 3- 12 which can be simply established from analysis of the data. However,
`using this datum line for volume calculation includes volume from the top surface of
`the cylinder in the measurement, which is not part of the cell volume. As a result of
`this the datum for volume measurement was redefined to level B, by stating that:
`
`Datum height = Average land height - Y2 Rz
`
`Equation 3.2 - Datum height calculation
`
`Part of the cell is formed below this datum (Figure 3-12 - volume V 1), and that
`anything above it is merely a part of the cylinders own rouglmess - barring the space
`above the actual cell (Figure 3-12- volume V2). Data above the datum line (indicated
`by line 'B' in Figure 3- 12) must be removed for volume measurement. This removes
`the land area from the measurement, leaving only the cell. Should the measurement
`contain other cells or parts of cells, these must also be removed, using the masking
`facility.
`
`To compute measured volume, the area of a single pixel is calculated, and multiplied
`by the depth from the datum height in the measurement. This result is then integrated
`across the entire measurement field to give a total volume (V1) for the cell. This is an
`automated function, built into the analysis package supplied. The correctness of its
`working was confirmed via the use of a stepped surface7
`, and was found to be
`accurate.
`
`This volume does not yet represent the volume of the engraved cell. The top part of
`the cell has been removed in order to remove the land area (volume V2), and this
`
`7 The stepped surface used was an extremely smooth and accurately defined step, used for calibration
`of the equipment, As the dimensions of the step are well specified by the manufacturers, the volume
`can be calculated malhematically and by the measurement technique.
`
`50
`
`FAST FELT 2010 , pg. 65
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`(
`
`needs to be added back in to calculate the final cell volume. This section is calculated
`using the open area of the cell (section 3.2.6). This is multiplied by the Rz to give
`volume V2, Figure 3-12, which when added to the cell volume previously calculated
`(V J) gives a total volume for the cell. A further section is missing around the very top
`section of the cell, indicated by the space between the volume V2 and the surface
`roughness in Figure 3-12. This was calculated as being between 0.3% for a small cell,
`to 0.1% for a larger cell. This is negligible, and was excluded in the analysis.
`
`Primary work was undertaken to calculate the number of cells that must be evaluated
`in order to provide an acceptable average. An example of this type of calculation is
`given in section 3.6 (for the number of analysed dots). It was found that the analysis
`of four cells was required. For each analysis of each type of cell, five were measured
`the additional cell being included to ensure that enough measurements were obtained
`in the event of a problem with one of the measurements, and results were then
`averaged. An example of calculation of all parameters is given at the end of this
`section.
`
`3.2.6 Cell length, width depth and open area
`
`With the measurement area masked for the volume calculation (see 3.2.5), the length
`and width of the cell was also calculated. Length is determined as the distance
`between the point at which the diamond enters the copper to engrave the cell, to the
`point at which it leaves. This is indicated by length 'a' in Figure 3-13. The width is
`defined as the maximum distance between points in the cross-cylinder direction. This
`is indicated by length 'b' in Figure 3-13. It was taken to be a simple linear measure,
`the curvature of the cylinder making little difference to the length on these scales. The
`dimension 'A' is however, not determined by diamond geometry, but by the speed at
`which the diamond is pushed into the copper, and the speed of rotation of the cylinder
`during the engraving process.
`
`51
`
`FAST FELT 2010 I pg. 66
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`Figure 3-13 - Width and Length calculation
`
`If all data above the datum line is masked, as in 3.2.5 (volume calculation) the number
`of measurement points making up the cell can be measured. By multiplying the
`number of pixels by the area of each pixel, the open area of the cell may be calculated
`(Equation 3.3)
`
`Open_Area = LN11 xP"
`
`Equation 3.3 - Open area calculation
`
`The measured depth of a cell is the vertical distance from line 'A' to the base of the
`cell in Figure 3-12. The actual depth of the cell must also include the 0.5 R2 that was
`subtracted by the manual masking. (see section 3.2.5). Adding this 0.5 Rz on to the
`depth ensures that the top and bottom of the cell are m the same place for both volume
`and depth calculations.
`
`3.2. 7 Cell offset
`
`Due to the electromechanical engraving technique used in the production of the
`cylinder the lowest point in the cell is not precisely in the middle of the cell. The
`offset angle defines the angle between a vertical line from the bottom of the cell and
`the line joining the bottom of the cell to the geographical centre of the cell. The offset
`is directly related to the diamond and its holder used in the engraving. A typical
`
`52
`
`FAST FELT 2010, pg. 67
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`example of the type used is shown below (Figure 3-14). The diamond is rotated into
`the copper about the pivot point. As a result, the diamond tip does not execute a
`straight line as it enters the copper, but a curve, which shows itself as the measured
`offset value.
`
`Pivot Point
`
`Cylinder
`
`Path of diamond
`
`Figure 3-14 - Engraving Diamond
`
`This can be seen most clearly if the cell data is inverted, Figure 3-15.
`
`Figure 3-15 - Inverted cell, showing offset
`
`The offset is indicated by the manner in which the cell 'leans' to the right.
`
`3.2.8 Cell wall roughness
`
`Once a cell has been measured, it is possible to select only a small portion of the cell,
`and flatten it in the measurement plane through the application of a mask (section
`3.2.4.1 ). From this, it is possible to calculate the roughness of the wall inside the cell.
`
`53
`
`FAST FELT 2010 , pg. 68
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`An example is shown in Figure 3-16. A preliminary investigation was performed.
`Although
`it highlighted internal cell features,
`
`it did not affect cell volume
`significantly and so was not considered to be important in the context of this study .
`
`oP------------------------..
`
`"
`..
`"'
`"
`
`,.,
`,..,
`
`1<10
`
`. ...
`
`OJI)
`
`<lSI
`
`·I«<
`
`·I !I)
`
`•Uil
`
`·!&J
`
`Figure 3-16- Cell wall quadrant, showing roughness
`
`3.2.9 Example calculation
`
`A typical measurement is shown below, Figure 3-17. Several cells are included in the
`measurement, and the measurement must be masked to remove the additional cells.
`
`54
`
`FAST FELT 2010 , pg. 69
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`~ Welsh Centre for
`~ Printing and Coating
`
`Surface Data
`
`Mag 181 X
`Mode VSI
`
`Date IOIIIn002
`Tune 10 5645
`
`:m
`
`2SJ
`
`:Ill
`
`ISl
`
`0111
`
`Sl
`
`Swface Statistics:
`Ra· 4 'l7 urn
`Rq 5 05 urn
`Rz 20 IOurn
`Rt 10 46 urn
`Rku 3 02
`Rsk -1 13
`
`Set-up PanJM1US:
`S.u 480 X736
`Samphng $42 39 nm
`Pnocessed OplioftS:
`T enns Removed
`Ttlt
`Flltenng
`Median
`
`Title: db70.1
`Note:
`
`so
`
`JD
`
`10
`
`· 10
`
`-3J)
`
`.. J)
`
`·l Jl
`
`ofO
`
`•UD
`
`·IJO
`
`•1$1
`
`Sl
`
`m
`
`1lD
`
`:m
`
`:I!D
`
`Figure 3-17- Raw measurement of engraved surface
`
`Figure 3-18 shows the individual cell to be analysed. This has been masked, removing
`the extra cells,. The shape of the mask is not critical, and here a simple circle has been
`used to isolate the cell.
`
`55
`
`FAST FELT 2010 , pg. 70
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`~ Welsh Centre for
`~ Printing and Coating
`
`Surface Data
`
`Mag- 18 I X
`Mode- VSI
`
`Date 10/1112002
`Tsme 10 56 45
`
`31l
`
`:m
`
`2!ll
`
`2ll
`
`lSI
`
`1111
`
`<II
`
`Surface Statistics:
`Ra 381 urn
`Rq 473 urn
`Rz 19 04 urn
`Rt 1976 urn
`Rku 4 01
`Rsk -I 48
`
`S.t.ap Para.eters:
`SIU 480 X 736
`Semphng 542 39 nm
`Pncessecl Optilnas:
`Terms Removed
`TilL
`Flltenng
`Medtan
`
`Title: db70.1
`Note:
`
`Figure 3-18- Masked measurement of engraved surface
`
`A measurement screen was set up to allow calculation of depth and volume, Figure
`3-19. This incorporates both a 20 image of the cell, (top left), a volume chart (top
`middle - unused), the volume information (top right) and a histogram, (bottom).
`
`56
`
`FAST FELT 2010, pg. 71
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`Volume Calculations
`
`Results
`VaNd Points
`Pixel SIZe
`Volume
`
`rm· II.OO'f.ofP.V
`Pis Below: 110.00% of Tote I
`Vol: &.38e+004 um3 100.00% of Tote I
`
`Histogram
`
`8021 7
`542 39 nm
`63789 um3
`
`A
`
`...
`
`..
`
`ltcljll
`
`..
`
`Figure 3-19- Developed calculation screen
`
`The histogram (at the bottom of the screen) shows the number of points at any given
`height. The 'Volume' number is the number of cubic microns of fluid which would be
`required to submerge the sample to the highest point on it. This is equivalent to the
`volume described in section 3.2.5 as the volume below line 'C' (Figure 3-12) The
`'Valid Points' variable, gives the number of pixels in the measurement.
`
`Histogram
`
`Jl ...
`
`...
`
`.,
`
`...
`
`·•
`
`..
`
`Ktlg ll
`
`·•
`
`·•
`
`Figure 3-20- Histogram indicating peak
`
`57
`
`FAST FELT 2010 , pg. 72
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`rn Figure 3-20 the highest point indicates the most common height for pixels - in this
`case, relating to the height of the land area.
`
`..r
`~ r-- ()(30om ]-- - - -
`=
`
`y 2553 0
`
`-= =
`--= ---=
`= ---
`=
`--=
`.-
`
`...
`
`Histogram
`X 188um
`
`-----
`
`(
`
`lt •158 .....
`Y• 4j)
`
`----
`
`-
`
`T
`I
`
`\ I
`
`j
`
`--
`
`__ __,r
`
`-
`
`•tl
`
`. ..
`
`.. Htlglt
`
`·l
`
`Figure 3-21 - Histogram, indicating cell depth
`
`The height from the land area to the lowest data point gives the depth of the cell,
`mjnus the 1/ 2 Rz, which for this cylinder = 0 .8Jlm above the mean land level, thus the
`total depth of the cell is 19.6Jlrn, Figure 3-21 (18.8jlm as indicated, + 0.8Jlm =
`19.6Jlm)
`
`Htstogram
`
`O~h~ J -------------
`
`v
`
`....
`
`I'Ll ~ .... •,1
`
`------
`
`Figure 3-22 - Histogram indicating masking point and direction
`
`The cursor has now been moved 0.8Jlrn to the left (down), and a 'Mask Right' is
`about to be performed. This removes all data to the right (above) of the cursor, Figure
`3-22. This leaves the part of the cell below line 'B' m Figure 3-12. As stated above,
`
`0.8~ is equivalent to 'h Rz for this cylinder.
`
`58
`
`FAST FELT 2010 , pg. 73
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`Volume Calculallons
`
`Results
`Valid Points
`Pixel stze
`Volume
`
`nm·· 0.00% of P-V
`pta Below: 110.00% of Total
`Vol: 5.62e+D04 um3 110.00% of Total
`
`Histogram
`
`31144
`542 39 nm
`56225 um3
`
`·•
`
`·•
`
`..
`
`·• N•IJII
`
`Figure 3-23- Entire data for masked cell
`
`The volume (V 1) for the cell may now be calculated using data taken from the
`analysis package, Figure 3-23. In this case;
`
`• Volume = 56225um3
`
`• Valid Points = 3l l44 points
`
`The area of this open area must now be calculated. The total measurement area is
`equal to
`
`342 urn x 260um = 88920urn2
`
`using a camera of resolution 736x480 pixels. Thus I pixel equals;
`
`342 x 260 = 0.252 um2/pixel
`736 X 480
`
`59
`
`FAST FELT 2010 , pg. 74
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`Thus
`
`Total Volume= Volume of cell+ Volume of top section
`
`Total Volume= Volwne of cell+ (Open area x R,)
`
`Total Volume= Volume of cell + (Valid Points x Area/pixel x 1.6)
`
`Total Volume= 56225 + (31144 x 0.252 x 1.6)
`Total Volume= 68,782um3 for this cell
`
`The cell dimensions may be established using a similar strategy, Figure 3-24.
`
`\I !i
`
`X Profile
`x 103.8 um
`
`\
`--~.,..,__,. -- -..-,
`..
`..
`
`Y Profile
`lt 1238um
`
`..
`...
`
`1.!
`!!
`
`!l!!l!!
`e-
`1!1!111!
`br%1!
`""•
`
`-
`
`1 J
`
`""
`I. 0
`1 "' ~
`
`..
`
`!,!
`
`1.!
`1!
`b
`
`.,,_
`
`~jf ...
`l!E•
`:IH~
`·lfllt ...
`
`., .....
`
`?9 .. f.J ._
`
`~J!
`·2D.•
`-1!!.91-
`
`,..,_
`
`4J ....
`l'l!.
`'8:•
`-UJ-..
`
`-
`
`..;
`
`X
`y
`!rt
`
`Wll
`19"
`.. ,
`
`~
`Tide: db70.1
`Note:
`
`---
`
`~ u .....
`ca U2-
`l!!M
`l!l!lf
`~~~ •lr?J: :S
`
`""· -ma-
`
`Figure 3-24 - Measurement of depth and width
`
`Here the length and width of the cell may be calculated, by finding the north, south,
`
`east and west extremes of the cell (indicated by the triangles in the cell picture, top
`
`left). The width is defined as the measured distance in the 'X Profile' (103.8um) and
`
`the length as the measured distance in the 'Y Profile' (l23.8um).
`
`60
`
`FAST FELT 2010, pg_ 75
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`X Profile
`~.2~
`
`...
`
`Y Profile
`lt 123 8 um
`
`._,-
`··'·~
`
`U:~•
`:!" !!!l
`•lt,71-
`
`!S
`1&
`II.
`k
`b
`
`J ...
`;.:,
`0
`"t ~
`
`,.
`
`!:!!!! m.lli!-"
`c .....
`·1&.H ...
`tapy
`VtP
`·2J:I•
`laldl
`,mn-
`
`.....
`
`...
`
`J. ~
`c
`T 0.
`~
`
`.gJ ...
`
`1.)1-
`1'22•
`'2!.
`
`-9~-
`
`!5
`1&
`II
`k
`b
`
`!:!!!!
`c .....
`z..
`.....
`bUll
`
`Ul..&
`t l2 -
`Ita
`•l21-
`·) ) ) l f -
`
`-
`
`W81
`
`. -. -..
`
`Title: db70.1
`Note:
`
`Figure 3-25- Measurement of offset
`
`The offset is calculated in Figure 3-25. The top of the cell is the point at which the
`
`diamond enters the copper (engraving from top to bottom) in this example, so the blue
`
`line (Y Profile) is lined up on the northmost point of the cell. The lowest point in the
`
`cell is found, and the distance in the X direction (east-west) is calculated from this
`
`point to the line coming down from the diamond entry point, in this case 8.2um.
`
`It is also possible to measure the roughness of the internal cell walls. First, the first
`
`quadrant to be examined must be isolated, Figure 3-26. Once the quadrant is isolated,
`
`selecting the 'terms mask' allows the section isolated to be used as the measurement
`
`plane.
`
`61
`
`FAST FELT 2010 , pg. 76
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`"
`
`p.a
`::J
`<'~Joe~ Jc:-~wo
`lr p.,.
`r-....!l!lJ~l
`_J _,___j _J
`~_J....i.._jl
`~ ...,.E_ r
`P ... l;p .. II
`__J
`y...
`u .. .
`,- RM
`~01&0
`t"" D .. &. M•lY f'~
`r
`..;
`(' Ha1k
`
`Figure 3-26 - Cell wall roughness, isolating the quadrant
`
`Once the measurement area has been isolated, curvature must be removed, effectively
`
`rendering a level plane.
`
`Once this has been selected, the quadrant is levelled, and the rouglmess, in terms of
`
`Ra, Rq etc, may be read from the analysis screen, Figure 3-27.
`
`Slllface Statistics:
`Ra: 611 .41 run
`Rq: 837 33 run
`Rz: 4.98 urn
`Rt· 10. 10um
`Rku: 9.78
`Rsk: 1.04
`
`Set-vp Parameters:
`Size. 480 X 736
`Sampling: 54239 nm
`P.rocessed OptiDJtS:
`T enns Removed:
`Curvature & Tilt
`Filtering:
`Medi81l
`
`76
`
`6D
`
`5D
`
`4D
`
`JD
`
`2D
`
`ID
`
`DD
`
`-10
`
`-25
`
`Figure 3-27- Flattened quadrant
`
`(
`
`62
`
`FAST FELT 2010, pg. 77
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`3.2.10 Comparison with traditional methods
`
`A series of comparisons benveen different cell models was performed, to see if it was
`
`possible to use the microscopy measurements to establish the cell volume.
`
`Calculations were performed on an 80% cell (the largest cell with no channel present),
`
`engraved with a 130° stylus. Average cell dimensions were used (137.9um wide,
`
`153.3um long). The measured volume was compared with both the traditional,
`
`pyramid shape approximation, and a new, average circumscribing conical model
`
`where;
`
`1r (width+ length )
`VO ume =-X
`I
`3
`4
`
`2
`
`width
`X --:-----:-
`2 tan(B / 2)
`
`Equation 3.4- Conical model
`
`The possibilities of user error are much larger when using the image analysis
`
`techniques than in the measurement technique, described previously due to the
`
`necessity for user intervention when defining the width and length of cells. This may
`
`lead to inaccuracies ofup to approximately 3j.lm in each dimension, which will have a
`
`significant effect on the results obtained.
`
`In addition, the stylus angle must be assumed to be perfect for this calculation. Since
`
`it is shown in Appendix 1 that the diamond angle changes with wear, this adds a
`
`further inaccuracy to the calculation. Clearly this error in stylus angle is not relevant
`
`to the interferometric technique described as this measures the volume of the cell
`
`itself, without inferring the internal shape.
`
`3.2.11 Blade load measurement
`
`(
`
`The effect of blade load on ink release was explored with the project (see figure 3.1)
`
`and a means of quantifying this is required. Practically the doctor blade is loaded on
`
`the cylinder pneumatically. The load is controlled using a gauge on the press, which
`
`63
`
`FAST FELT 2010 I pg. 78
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`monitors the pneumatic pressure supplied to the pistons. Thus strain gauges were
`
`glued to the doctor blade, and the bending of the blade monitored. However this does
`
`not allow a direct measure of load due to the friction that is present between the blade
`
`holder and the press. This friction may be variable and therefore blade load may not
`
`be indicated consistently via the pneumatic supply pressure. The strain induced in the
`
`blade by the application of the blade to the cylinder was monitored with a PC based
`
`datalogger. This was used to ensure that the same load was applied for each trial,
`
`rather than establishing actual load levels. The latter is particularly difficult to achieve
`
`and requires the application of a calibration procedure coupled with some analysis of
`
`blade stressing [ 61] [ 62].
`
`In the case of the scumming press trials, gauges were mounted in the centre of each
`
`surface finish band on the cylinder. This gave confirmation that the same load was
`
`being applied to each surface, with no significant variation caused by either the blade,
`
`cylinder or mounting system.
`
`3.3 Running the Press Trials
`
`3.3.1
`
`Ink Release From Image Areas
`
`Several inks are available to gravure printers, but in the UK at least, almost all
`
`gravure inks are of one of two standards, either NC (Nitrocellulose) based or PVB
`
`(Polyvinyl Butyral) based. NC inks are the most common, largely because of their
`
`lower costs, and (in the opinion of most printers) lower scumming attributes. As a
`
`result, NC inks were used for this trial.
`
`Several different substrates were considered for the trial, from newsprint to
`
`cartonboard. The gravure industry is approximately evenly split between the use of
`
`paper and film substrates, but the packaging industry primarily uses film substrates.
`
`Due to the need to measure ink transfer, it was decided to use a plastic film. This
`
`(
`
`would not absorb the ink (as most papers would) but would leave the ink on the
`
`surface of the substrate, where the volume transferred could be measured, and thus
`
`64
`
`FAST FELT 2010, pg. 79
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`quantified. It was decided to use a film commonly used in the industry, and as such it
`was agreed to use Mobil 247, white OPP (oriented poly-propylene) substrate due to its
`widespread usage, and thus direct commercial and practical implications for this
`research.
`
`3.3.1.1 Trial Methodology
`
`Prints were made on a standard OPP substrate using a small commercial press. The
`appropriate ink was loaded into the ink tray, and the press was run up to speed
`(I OOm/min). Prints were made for 60s, producing approximately 1OOm of print I 160
`repeats of the image. Prints were collected at the end of the press on the take-up spool,
`and all prints were retained for analysis. Unprinted samples of substrate were also
`collected to allow analysis of the substrate. The speed of the press was confirmed
`using a handheld optical tachometer.
`
`A series of thermocouples was used to monitor operating temperatures. They were
`placed on either side of the ink tray, two were attached to the frame of the press, and a
`further thermocouple was used to monitor ambient temperatures.
`
`Strain gauges were attached to the doctor blade to accurately measure the applied load
`in three places along the length of the doctor blade, or over the width of the cylinder.
`Blade load measurements and thermocouple measurements were logged at a rate of
`0.5Hz over the entire period of the measurements. These were used to ensure that the
`blade load was maintained, constant over the duration of the trial, it having been
`shown previously that doctor blade load I angle can make a significant difference to
`the printed colour I density (39] (40].
`
`65
`
`FAST FELT 2010 I pg. 80
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`0
`
`c
`
`c
`
`c
`
`3.3.2
`
`Ink Release From Non-Image Areas
`
`3.3.2.1
`
`Ink Selection
`
`A variety of inks were selected for use. For the surface finish trial, both NC and PVB
`
`inks were used, whilst at the ink trial, three NC inks were used. NC inks were used
`
`because these are the most commonly used in the packaging industry [10] [63] [64]
`
`[65] [66]. It was decided to use standard commercial inks to ensure correlation
`
`between experimental results and practical experience.
`
`Inks were standard,
`
`commercial inks, supplied as they would be supplied to any printer, and thinned with
`a standard solvent. In the case of the second trial, a special 'low scum' ink was tested,
`but again, this was a standard ink available from the manufacturer.
`
`The surface finish trial used black inks, while the ink trial used red inks. In each case,
`
`the inks used were produced to the same initial colour strength. They were then
`
`thinned with solvent, thus lowering the viscosity, and also lowering the colour
`
`strength. By reducing each ink to the same viscosities, colour strength was maintained
`
`at approximately similar levels.
`
`3.3.2.2 Substrate selection
`
`A variety of substrates were selected for use. For the first trial, three distinct
`
`substrates were chosen. A white OPP (Mobil 247 - as with the ink release from image
`
`areas trial), a clear OPP and a metallised foil. These were selected as they make up
`
`approximately 90% of the output of the printers. It was suggested that almost all
`
`scumming problems occur on plastic substrates [9] [10] [63) [64) [65] [66], so it was
`
`decided not to examine the effects on papers in this experiment.
`
`For the second trial only two substrates were examined, the same white OPP and the
`
`same clear OPP. These were selected for the same reasons, as well as to allow direct
`
`(
`
`comparison between the two trials.
`
`66
`
`FAST FELT 2010, pg. 81
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`c
`
`c
`
`3.3.2.3 Trial Methodology
`
`Prints were made on commercial presses8, under conditions as close to standard
`
`
`production conditions as possible, and conditions were replicated over the two trials.
`
`Print speed was 250m/min, and impression pressure was set at 2 bar, gauge. Blade
`
`load was set at a maximum and reduced (trial one) or at a minimum and increased
`
`(trial 2). In each case, the load was incremented by 0.25 Bar (gauge). Thus load was
`
`indicated by pressure setting on the press, and was monitored using strain gauges.
`
`Once the press was up to speed, it was run for 45s, the blade load was incremented,
`
`the press was run for a further 45s, and the process was repeated until either
`
`scumming appeared to be eliminated, or until 0 Bar gauge pressure was being applied.
`
`The press was only stopped when necessary for ink or cylinder changes
`
`3.4 Print Measurement Techniques
`
`As detailed, interferometric analysis was used to quantify the features of the cylinder.
`
`Further application of this technique was used for the analysis of prints, allowing the
`
`calculation of precise dot volumes. Spectrophotometric techniques were also used,
`
`allowing the calculation of traditional values of tone gain I CIEL·a·b• colour.
`
`Measurement methodologies for colour and density measurement have been
`
`standardised, and these standard methods are described in [ 15]. The following
`
`sections describe the key elements.
`
`3.4.1 Printed Density and colour
`
`Measurements were taken with both handheld Gretag SPM-50 spectodensitometers
`
`and automated Gretag Spectrolino/Spectroscans. All results were taken using a D50
`
`illuminant, 065 filter, and using the ANSI T standard. Transmission measurements on
`
`8 In each case, I 0 units were available, with a printing width of approximately 1.2m. Press footprint
`was approximately 4m x 30m, with a height of 5m
`
`67
`
`FAST FELT 2010 , pg. 82
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`c
`
`c
`
`c
`
`(
`
`clear
`
`substrates were
`
`taken usmg a Gretag Spetroscan-T
`
`transmission
`
`spectrophotometer.
`
`3.4.1.1 Density calculation
`
`To produce prints of differing colour, it is necessary to vary the thickness of the
`
`printed ink film. Since the rotogravure process is not a contone process, it is necessary
`
`to use an alternate method. By varying the size of the cells, and thus of the printed
`
`dots, it is possible to vary the amount of ink transferred over a given area. These
`
`different sizes of dots are referred to as halftones [ 19]. The density of this area
`measures the intensity of the printed colour. It is defined as:
`
`density = log10
`
`1
`reflectance
`
`Equation 3.5 - Density calculation
`
`3.4.1.2 Optical Tone gain calculations
`
`When ink is transferred from the printing cylinder to the substrate, the printed dot will
`
`not in most cases be the same size as the cell from which it is printed. This change in
`
`size is referred to as the tone gain (or dot gain). Tone gain is the effect caused by the
`
`ink spreading out on the substrate, and thus giving a higher optical density than would
`
`otherwise be expected. Tone gains were calculated using the Murray-Davies equation
`
`(see Equation 3.6, below)
`
`Where:
`a = printed area, 0 1 = tonal density, Ds = solid density
`
`Equation 3.6- Tone gain calculation
`
`68
`
`FAST FELT 2010, pg. 83
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`0
`
`In certain circumstances, the printed dot may be smaller than the cell opening. In this
`
`case, this is still referred to as tone gain, (with a negative value) and generally only
`
`occurs in the highlight regions of t