Optical Property Measurement System and Method

In the prior art it is known to obtain an indication of color and brightness characteristics of a paper web during manufacture by an on-line measurement of reflectance value (Rg), but this measurement is decidedly different from that necessary for actual color and brightness characterizations. Accordingly, such a measurement must be accompanied by very frequent off-line testing, so as to enable an adequate empirical calibration of the measuring instrument. Further, a separate set of calibration parameters is required for each grade and weight of paper. Off-line instruments which adequately measure these characteristics require that a pad of several thicknesses of paper be exposed to the light source aperture so that a different reflectance value (Roo) is obtained. Obviously this is impossible with an on-line instrument unless the far more inaccessible reel itself is tested.

Only where the on-line measured reflectance value (Rg) approaches the off-line value (Roo), as in instances of paper of extremely high opacity such as heavily coated or heavily dyed paper, can the above problems be minimized to the point where accuracy becomes sufficient for control purposes.

There is some possibility that the prior art includes the sequential measurement of two separate reflectance parameters such as one with a backing of near zero absolute reflectance and one with a white backing, but there appears to be no concept of a multiple property measurement system and method utilizing corresponding multiple sets of parameter measurements based on respective common spectral response functions and the prior art concept may be largely limited to a direct determination of Tappi opacity and/or to an abstract investigation of the feasibility of determining absolute reflectance (Roo) at a given wavelength such as 457 nanometers.

This invention relates to an optical device and method for sensing optical properties of a substantially homogeneous sheet material, and particularly to an on-the-paper-machine device and method for simultaneously sensing both transmitted and reflected light so as to obtain measurements from which the optical properties of interest can be calculated substantially independently of grade and weight of paper involved.

Accordingly it is an object of the present invention to provide an optical monitoring device and method for sensing optical properties based on measurements made on a single thickness of partially translucent substantially homogeneous sheet material and which measurements sufficiently characterize the actual properties of interest that a minimum of empirical calibration is required regardless of changes in grade and weight of paper.

Another object of the invention is to provide such an optical monitoring device and method capable of accurately sensing two or more optical properties such as brightness, color, opacity and/or fluorescent contribution to brightness preferably essentially continuously, or at least substantially simultaneously, and especially adapted for operation with a continuously moving web.

With such an optical monitoring device is useful off-line for sensing optical properties of a single thickness sample, it is a further important object of the present invention to provide such an optical monitoring device which is of sufficiently light weight and compact construction so as to be adapted for on-line monitoring of the desired optical properties.

Another and further object of the invention is to provide an on-the-paper-machine optical monitoring device of sufficient flexability and accuracy to enable control of desired optical properties during the paper making process.

A unique feature of the on-line optical monitoring device is its ability to essentially simultaneously measure both reflected and transmitted light. By measuring two independent optical parameters it is possible to thoroughly characterize the paper optical properties of a partially translucent web with a minimum of empirical correction for factors such as paper grade and weight. Preferably a plurality of sets of optical parameters are sensed continuously or at least substantially simultaneously.

Other objects, features and advantages of the present invention will become apparent from the following detailed description taken in connection with the accompanying drawings.

FIG. 1 is a fragmentary somewhat diagrammatic longitudinal sectional view of a paper machine showing in outline a side view of an optical monitoring device in accordance with the present invention operatively mounted on line with the machine;

FIG. 2 is a fragmentary somewhat diagrammatic transverse sectional view of the paper machine of FIG. 1 and taken generally as indicated by the line II--II of FIG. 1 and looking in the direction of the arrows (toward the wet end of the paper machine), the view being taken so as to show in outline a direct front view of the optical monitoring device of FIG. 1;

FIG. 3 is a diagrammatic longitudinal sectional view of one embodiment of an on-the-paper-machine optical monitoring device in accordance with the present invention;

FIG. 4 is a partial diagrammatic plan view of the filter wheel assembly utilized in the monitoring device of FIG. 3;

FIG. 5 is a somewhat diagrammatic view illustrating an optical analyzer unit in electrical association with the optical monitoring device of FIGS. 1-4 and with a power supply unit;

FIG. 6 is an electric circuit diagram illustrating the electrical connections between the various components of FIGS. 1-5;

FIG. 7 is a flow chart illustrating an existing direct digital control analog point scan program which has been adapted to allow for the collection and temporary storage of the reflectance and transmittance data acquired from the system of FIGS. 1-6;

FIGS. 8-16 when arranged in a vertical series represent a program fourteen which is designed to read the reflectance and transmittance values stored pursuant to FIG. 7 and generally to control the operation of the system of FIGS. 1-6 and to apply correction factors to the raw reflectance and transmittance data;

FIGS. 17-20 when arranged in a vertical sequence represent a data reduction program forty-two whose purpose is to reduce the corrected reflectance and transmittance data as produced by the program of FIGS. 8-16 into terms with which papermakers are familiar and upon which paper optical specifications are based, e.g. brightness, opacity, color and fluorescence;

FIG. 21 is a diagrammatic view illustrating the embodiment described in the section entitled "Proposed Instrument Design" of the aforementioned copending applications;

FIG. 22 illustrates further details of the embodiment of FIG. 21 as described in the copending applications;

FIG. 23 shows an embodiment similar to that of FIGS. 21 and 22, but with the interior of the transmittance cavity providing the backing for the reflectance measurements; and

FIG. 24 shows an embodiment similar to that of FIGS. 21 and 22, but utilizing multiple sets of reflectance and transmittance optical light pipes.

FIGS. 1 and 2 will serve to illustrate the modifications of an existing paper machine which are required for carrying out a preferred embodiment of the present invention. Referring to FIGS. 1 and 2, an on-the-paper-machine optical monitoring device is diagrammatically indicated at 10 and comprises an upper sensing head 11 and a lower sensing head 12 which are maintained in precise relative alignment and disposed for operative association and transverse scanning movement relative to a paper web located as indicated at 14 in FIGS. 1 and 2. As will be described hereinafter with reference to FIGS. 3 and 4, in a particular design of the optical monitoring device, upper head 11 includes a light source for projecting light onto the web such that a portion of the light is reflected parallel to an optical axis indicated at 15, while a further portion of the light is transmitted through the paper web for collection and measurement by means of the lower sensing head 12.

For purposes of illustration, FIGS. 1 and 2 show portions of an existing web scanner construction which is utilized to scan the web 14 for conventional purposes. The conventional scanner construction includes fixed frame components such as 20, 21 and 22 forming what is known as an "O" type scanner frame. The conventional scanning structure further includes upper and lower slides 25 and 26 for joint horizontal movement along the horizontal beams 21 and 22. Associated with the slides 25 and 26 are movable assemblies 27 and 28 carried by the respective slides 25 and 26 and including vertically disposed plates 31 and 32 and angularly disposed flange members such as indicated at 33 and 34 in FIG. 1. These flange portions 33 and 34 have broad surfaces lying in planes generally parallel to the plane of the web 14 and are utilized for mounting of the monitoring device 10 of the present invention. In particular a top head mounting bracket is indicated at 41 in FIGS. 1 and 2 and is shown as being secured to the existing flange part 33 so as to mount the upper head 11 for scanning movement with the assembly 27. Similarly a lower head mounting bracket is indicated at 42 and is shown as being secured to flange part 34 of the lower movable assembly 28 so as to mount the lower sensing head 12 for scanning movement jointly with the upper sensing head 11.

For the purpose of electrical connection with the monitoring device 10 during its traverse of the web 14, electric cables are indicated at 51 and 52 for electrical connection with the components of the upper sensing head 11 and lower sensing head 12 of the monitoring device 10. The cable 51 is shown as being fastened by means of straps 53 and 54 to a top carrier slide bracket 55. The bracket is shown as being secured by means of fasteners 56 and 57 to the upper portion of vertical plate 31. As indicated in FIG. 2, successive loops of cable 51 are secured to swivel type ball bearing carriers such as indicated at 61. A trolley track 62 is supported from existing channels such as indicated at 63 and mounts the carriers 61 for horizontal movement as required to accommodate the scanning movement of the monitoring device 10 across the width of the web 14. Similarly, successive loops of the cable 52 are fastened to the eyes such as indicated at 71 of a lower series of carriers 72. As seen in FIG. 1 each of the carriers such as 72 includes a pair of rollers such as 73 and 74 riding in the trolley track 75 which is secured directly to the lower flange 22a of beam 22. A lower carrier slide bracket 81 is secured to vertical plate 32 by means of fasteners 82 and 83 and is provided with a horizontally extending flange 84 for engaging with the first of the series of lower carriers 72. In particular, carrier 72 is provided with a shank 85 which extends into a longitudinal slot 84a of flange 84. Thus, the first carrier 72 is interengaged with the bracket 81 and is caused to move with the lower assembly 28 and the lower sensing head 12. The remaining lower carriers such as that indicated at 83 move along the trolley track 75 as necessary to accommodate movement of the monitoring device 10 transversely of the web 14.

While FIGS. 1 and 2 have illustated the optical monitoring device of the present invention as being mounted on line with the paper machine and have further illustrated the case where the monitoring device is to be scanned transversely of the web, it is considered that the optical monitoring device of the present invention would also be of great value if redesigned for bench mounting. By placing a single sheet of paper in a sample mount of the device, a technician could simultaneously test the sample for color, brightness, fluorescence, and opacity in a matter of seconds.

In the illustrated embodiment, however, it is contemplated that the monitoring device 10 will be mounted on line with the paper machine and will be capable of movement to a position clear of the edge of the web as indicated in FIG. 2 at the end of each hour of operation, for example. When the end of a production run for a given web 14 has been reached, or when a web break occurs for any reason (such as accidental severance of the given web), the monitoring device 10 will be moved clear of the edge of the web path as indicated in FIG. 2. Each time the monitoring device 10 is moved to the off-web position shown in FIG. 2 it is preferred that readings be taken of the reflectance and transmittance values (without the web in the optical path) for the purpose of obtaining an updated calibration of the monitoring device. Thus, such updating of calibration may take place automatically (for example under the control of a process control computer controlling the paper manufacturing operation) at hourly intervals and also after web breaks. The monitoring device can, of course, be retracted manually any time desired by the operator for the purpose of checking calibration. By way of example, the monitoring device 10 may be capable of a normal scanning travel over a distance of 115 inches with provision for an additional travel of 16 inches to the position shown in FIG. 2. A flange is indicated at 87 which serves to insure proper re-engagement of the sensing head with the web at the operator's side of the illustrated paper machine (opposite the side indicated in FIG. 2).

The lower head 12 is designed to contact the web 14 during scanning thereof. The design spacing between the upper and lower heads 11 and 12 is 3/16 inch. The optical opening in the upper head 11 is aligned with the optical axis 15 and is to be maintained in alignment with the center of the window in the lower head 12. Four adjusting screws such as those indicated at 91 and 92 are provided for accurate positioning of the upper head 11. Similarly four position adjusting screws such as 93 and 94 serve for the accurate positioning of the lower head in conjunction with set screws such as indicated at 95 and 96. The adjusting screws are located at each corner of mounting brackets 41 and 42.

Where the web is not perfectly horizontal, but instead is curved across, its width, it is desirable to provide a web deflecting guide bar as indicated at 97 in FIG. 3 for insuring stable contact between the web 14 and the web engaging surface 98 of the lower sensing head 12. By way of example the guide bar may protrude from the lower surface of the upper sensing head a distance of 5/16 inch so as to overlap with respect to the vertical direction a distance of 1/8 inch relative to the lower sensing head web contacting surface 98. The guide bar 97 may have a width to force down at least about four inches of the width of the web at a section of web centered with respect to web engaging surface 98 of the lower sensing head relative to the machine direction. This insures a minimum of a 1/8 inch bellying of the sheet as it travels over the lower sensing head in all lateral positions of the sensing head.

In order to minimize changes in the 5/16 inch thickness dimension of the guide bar 97 due to wear, the guide bar is provided with a flat web engaging surface 97a which has a dimension in the direction of web movement of about one inch. By way of example, the guide bar may be made of Teflon.

Since the guide bar 97 is not necessary when the web is fed from the calender stack to the reel in a relatively planar configuration, it has not been shown in FIGS. 1 and 2.

Various modifications may of course be made to adapt the monitoring device of the present invention to various types of paper machinery, and to secure any desired degree of accuracy in the joint scanning movement of the upper and lower sensing heads relative to the paper.

Referring to FIG. 3, the upper sensing head 11 is shown as comprising a casing 110 having suitable connectors 111 and 112 for receiving suitable internally threaded fittings 114 and 115, FIG. 1 associated with the electric cable 51. The casing 110 receives a top head shoe 120 including an interior open rectangular frame 121 having a base flange 12a spot welded to shoe plate 122. The upstanding portion 121b engages the adjacent wall of casing 110 along all four sides thereof and is secured to the casing 110 by suitable fastening means such as indicated at 124 and 125 in FIG. 3. An edge 122a of shoe plate 122 is bent up at an angle of 45.degree. at the side of the sensing head 11 facing the wet end of the paper machine, and a similar inclined edge 122b, FIG. 1, is provided at each of the sides of the sensing head so as to present smooth faces to the paper web during scanning movement of the sensing head. The shoe plate 122 is provided with a circular aperture of less than one inch diameter as indicated at 130 centered on the optical axis 15 of the device. In a present embodiment aperture 130 has a diameter of about 7/8 inch. This aperture 130 is preferably of minimum diameter necessary to accommodate the light paths of the instrument. In the illustrated embodiment the light path for the incident light beam as indicated at 133 is directed at an angle of approximately 45.degree. and is focused to impinge on a window 135 at the optical axis 15. A reflected light path as indicated at 137 is normal to the web engaging surface 98 (which is the upper surface of window 135), and is coincident with the optical axis 15, while light transmitted through the web 14 and through the window 135 is directed as indicated by rays 141-143, for example, into an integrating cavity 145 of lower head 12.

The lower head 12 comprises a casing 150 having an annular dished plate 151 secured thereto and providing a generally segmental spherical web-contacting surface 151a surrounding window 135. The window 135 is preferably formed by a circular disk of translucent diffusing material. In the illustrated embodiment the window 135 is made of a polycrystalline ceramic material available under the trademark "Lucalux" from the General Electric Company. This material has physical properties similar to that of sapphire. The opposite faces of window 135 are flat and parallel and the thickness dimension is 1/16 inch. A lip is indicated at 153 for underlying an annular edge portion of window 135. This lip provides a circular aperture 154 having a diameter of about 15/16 inch so that the effective viewing area for the transmitted light is determined by the diameter of aperture 154. The casing 150 is shown as being provided with an electrical connector terminal 155 for receiving a suitable internally threaded fitting 156, FIG. 1, of cable 52.

As diagrammatically indicated in FIGS. 3 and 4, the upper sensing head 11 includes a light source 201, incident optical path means including lenses such as indicated at 202 and a photocell 203 for measuring reflected light returning along the reflected light path 137. A filter wheel 210 is shown diagrammatically as being mounted on a shaft 208 for rotation by means of a low torque motor indicated at 209. As best seen in FIG. 4, the filter wheel includes an outer series of apertures 211-217 for selective registry with the incident light beam path 133, and includes a series of inner apertures 221-227 for selective registry with the reflective light beam path 137. The various apertures may receive suitable filter elements as will hereinafter be explained in detail such that a series of measurements may be taken by successively indexing the filter wheel 210 to successive operating positions. In each operating position one aperture such as 211 is in alignment with the incident beam path 133 and a second aperture such as indicated at 221 is in alignment with the reflected light beam path 137.

By way of example, the motor 209 may be continuously energized during operation of the monitoring device, and the filter wheel may be retained in a selected angular position by engagement of a ratchet arm 230 with one of a series of cooperating lugs 231-237 arranged generally as indicated in FIG. 4 on the filter wheel 210. A solenoid is indicated at 240 as being mechanically coupled with ratchet arm 230 for momentarily lifting the ratchet arm 230 out of engagement with a cooperating lug such as 231 so as to permit the filter wheel to index one position. Immediately upon release of the energization of solenoid 240, the force of gravity returns the ratchet arm 230 to the position shown in FIG. 3 so as to be disposed in the path of the lugs and thus to engage the next lug in succession such as lug 232 as the motor 209 moves the filter wheel 210 into the next operating position.

As will hereafter be explained in greater detail, reed switches are mounted in circles on respective switching boards 241 and 242, FIG. 3, and the filter wheel shaft 208 carries a magnet 243 for actuating a respective pair of the reed switches in each operating position of the filter wheel 210. Thus the position of the filter wheel 210 determines which of the switches on the switching boards 241 and 242 are closed. As will be explained hereinafter, the reed switch on the upper switching board 241 which is closed determines the gain setting of an upper head amplifier at a level appropriate for the set of filters which are in the operating position. The reed switch on the lower switching board 242 which is closed activates a relay on a circuit board 245 in the lower head 12, and such relay in turn sets the lower head amplifier gain at the proper level. As will be explained in connection with the electric circuit diagram for the monitoring device, certain conductors of the cable 51 may be interconnected at a remote location so as to cause an indexing movement of the filter wheel 210. This external command serves to momentarily energize solenoid 240 and lift the ratchet arm 230 about is pivot point 250, allowing the motor 209 to rotate the filter wheel 210. The ratchet arm 230 returns to the position shown in FIG. 3 to catch the next lug on the filter wheel stalling the motor 209.

Four heaters such as indicated at 251 are mounted around photocell 204 so as to minimize the temperature variations of the photocell. A circuit board for mounting an amplifier for photocell 203 and for mounting the gain setting resistances associated with the reed switches is indicated at 255 in FIG. 3.

Referring to the lower head 12, FIG. 3 indicates a photocell 260 for receiving light from the intergrating cavity 145 and a series of heaters such as 261 mounted around the photocell 260 to minimize the temperature variations of the photocell. Circuit board 245 may mount a suitable amplifier for photocell 260, the gain of which being controlled by the relays previously mentioned.

The heaters 251 and 261 in the prototype unit were Pennsylvania Electronics Technology Type 12T55. (These are positive temperature coefficient thermistors with 55.degree. C. switching temperatures.) These heaters will tend to stabilize the temperature since their ability to provide heat decreases as the ambient temperature increases. Above 55.degree. C., they provide essentially no heat at all.

A basic feature of the illustrated embodiment resides in its ability to measure simultaneously both reflected and transmitted light. While in the illustrated embodiment, the reflected light path 137 and the transmitted light path intersect the web 14 essentially at a common point, reflected light could be obtained from a point on the sample or web offset from the point where light is transmitted through the sample. For example, a backing of some specified reflectance such as a black body of zero or near zero reflectance could be located on the lower sensing head just ahead of or behind the transmitted light receptor compartment (with respect to the machine direction of the sample or the direction of movement of the web). In this case the upper sensing head could contain the light source as well as a reflected light receptor for receiving light reflected from the sample or moving web at a point directly above the backing of specified reflectance. Both the reflected light receptor in the upper sensing head and the transmitted light receptor in the lower sensing head could then supply signals simultaneously and continuously during measurement operations. Many other variations in the arrangement of the optics for measuring both reflected and transmitted light will occur to those skilled in the art.

Referring to the details of the illustrated embodiment, however, and to the case where it is desired to measure brightness, color, opacity and fluorescent contribution to brightness, light source 201, FIG. 3, may consist of a Model 1962 Quartzline lamp operated at 5.8 volts as measured at the lamp terminals. The 45.degree. incident beam path 133 and the normal reflected beam path 137 correspond to those of a standard brightness testor, and a casting (not shown) from a bench type standard brightness tester was used in constructing a prototype of the illustrated embodiment to give rigid support for the optical components such as indicated at 202 and 271-276 in FIG. 3. In the specific prototype unit, a stock thickness polished Corning type 4-69 glass filter 271 and a second type 4-69 filter 272 ground and polished to an appropriate thickness were used in the incident beam path to absorb most of the infrared as well as to give proper spectral response.

The reflected light path 137 included a pair of lenses 273 and 274 which focus the light on a 3/8-inch aperture in the plate 275 of the casting. A piece of diffusing glass 276 is located on the 3/8-inch aperture so that the light distribution over the surface of photocell 203 will be reasonably uniform. A Weston model 856 RR Photronic cell was employed.

The filter wheel 210 is designed and located in such a way that either the incident or the reflected beam or both can be filtered as desired. In the prototype, the wheel 210 was driven by a small motor 209 operated at reduced voltage so that it could operate continuously in a stalled condition.

Commerically available color and brightness meters are usually manufactured with the spectral response filters located in the reflected beam. In the prototype device, and in the later on-machine version here illustrated as well, however, the filters which determine the spectral response of the first six filter positions are located in the incident beam. There are two basic reasons for this choice of design.

(1) Both the reflected and transmitted light have the same incident intensity and spectral response against which each can be compared. The alternate would necessitate two sets of identical filters, one set located in the reflected beam and another in the transmitted beam--a difficult design to achieve in practice.

(2) Filters in the incident beam can be used to absorb all ultraviolet light and prevent it from striking the specimen. Thus, fluorescence, a phenomenon not accounted for by Kubelka-Munk theory is avoided.

For reasons explained shortly, the seventh filter position is an exception to the above in that substantial ultraviolet light is intentionally permitted to exist within the incident beam. Outside of the phenomenon of fluorescence the spectral response is independent of whether such filters are located in the incident or the reflected beams.

The spectral response provided by the respective positions of the filter wheel 210 were as follows: (1) papermaker's brightness (TAPPI brightness), (2) blue portion of the E.sub.c x function, (3) red portion of the E.sub.c x function, (4) E.sub.c z function without fluorescence (5) E.sub.c y function, (6) E.sub.a y function, and (7) E.sub.c z function, with fluorescence.

As is understood in the art, the symbols E.sub.c x, E.sub.c y, E.sub.a y, and E.sub.c z refer to tristimulus functions of wavelength as defined by the Commission Internationale c l'Eclairage which is identified by the abbreviation C.I.E. and is also known as the International Committee on Illumination. The subscript a in the function designation E.sub.a y indicates that the function is based on a standardized illumination designated as C.I.E. Illuminant A, while the subscript c in the other function designations refers to a somewhat different standardized illumination which is designated as C.I.E. Illuminant C.

Filters for providing the above spectral response characteristics in the respective operating positions of the filter wheel 210 have been indicated in FIG. 4 by reference numeral 281-288. In the specific example under discussion, apertures 221-226 are left open. Filter 281 is a standard filter for use in measuring TAPPI brightness, TAPPI referring to the Technical Association of the Pulp and Paper Industry. This filter transmits a narrow band of wavelengths in the vicinity of 457 nanometers.

Filters 282-285 are standard filters for a four-filter colorimeter and are conventionally designated X (blue), X (red), Z, and Y.sub.C. These filters provide the wavelength distributions required for the measurement of the C.I.E. X, Y, and Z tristimulus values under Illuminant C.

Filter 286 is conventionally designated as a Y.sub.A filter and is required by the TAPPI standard method for opacity measurements. This is a board band filter producing the C.I.E. Y wavelength distribution for Illuminant A, in conjunction with the source 201 previously described in this section. A discussion bearing on the feasibility of this type of measurement is found in a paper by L. R. Dearth, et al entitled "Study of Instruments for the Measurement of Opacity of Paper, V. Comparison of Printing Opacity Determined with Several Selected Instruments", Tappi, volume 53, No. 3 (March, 1970).

With respect to position No. 7 of the filter wheel 210, filters 287 and 288 are conventionally designated as Z (blue) and Z (yellow). As previously indicated, the purpose of the filters is to provide for a determination of the C.I.E. Z tristimulus value with the fluorescence component included. In filter position No. 4, filter 284 serves to remove the ultraviolet component from the incident beam so that a measure of the Z tristimulus value without fluorescence is obtained. In position No. 7 of the filter wheel, however, filter 287 in the incident beam is designed to transmit the ultraviolet component, so that the fluorescent component if any will be transmitted to photocell 203. The ultraviolet absorbing component of the Z type filter means is located in the reflected beam 137, whereas this component is in the incident beam for the No. 4 position. The fluorescent component is lineally related to the difference between the Z tristimulus values determined in the No. 4 and No. 7 positions of the filter wheel 210.

Filters 281-288 have been shown in FIG. 4 with different types of hatching which have been selected to represent generally the different light transmission properties of the filters. In particular, the hatching for filters 281-288 are those for representing white, blue, red, blue, green, orange, blue and yellow light transmission properties. The selection of hatching is primarily for purposes of graphical illustration and is not, of course, an exact representation of the light transmission properties of the respective filters.

FIG. 5 illustrates diagrammatically the optical monitoring device 10 of FIGS. 1-4, and illustrates by way of example an optical analyzer unit 300 which may be electrically associated with the monitoring device and serve as an operator's console to be disposed at a convenient location adjacent the paper machine. By way of example, the optical analyzer unit may be mounted near the dry end of the paper machine, and may receive conventional alternating current power from the paper machine dry end panel. The optical analyzer unit 300 is illustrated as being coupled with the monitoring device 10 via a power supply unit 301 which is mounted adjacent the vertical column 20, FIG. 2, of the "O" frame along which the monitoring device is to travel in scanning the width of the web. For purposes of diagrammatic illustration, power supply unit 301 is shown as being provided with a mounting plate 302 which is secured by means of a bracket 303 to an end of horizontal beam 22 which has been specifically designated by reference numeral 304 in FIGS. 2 and 5. Referring to FIG. 2, it will be observed that the ends 305 and 306 of cables 51 and 52 are adjacent the end 304 of beam 22 so that this is a convenient location for mounting of the power supply 301. The electrical interconnections between the power supply unit 301 and the optical analyzer unit 300 are indicated as extending via a signal conduit 311 and a control conduit 312. By way of example, the signal conduit 311 may contain shielded electric cables for transmitting millivolt signals from the analogue amplifiers of the upper and lower sensing heads 11 and 12. The control conduit 312 may contain conductors which are respectively energized to represent the angular position of filter wheel 210, and may also contain a conductor for controlling the indexing movement of the filter wheel as will be explained in detail in connection with FIG. 6.

Referring to the optical analyzer unit 300 of FIG. 5, the front panel of the unit has been diagrammatically indicated at 320 as being provided with a series of lamps 321-327 for indicating the angular position of the filter wheel 210 within the upper sensing head 11. The lamps 321-327 have been numbered 1 through 7 in correspondence with the seven positions of the filter wheel, and the color of the lamps, for example, may be selected so as to signify the characteristics of the filters located in the openings of the filter wheel such as those indicated at 211-217.

In order to provide a visual indication of the amplitude of the millivolt signals supplied from the sensing heads 11 and 12, a suitable meter is indicated at 330 and a selector switch is indicated at 331 for selectively supplying to the meter the analogue signal from the upper sensing head 11 or from the lower sensing head 12. A switch 332 is indicated for controlling the supply of conventional alternating current power to the meter, and a second switch 333 is indicated for controlling the supply of energizing power for the lamps 321-327. Another switch 334 may be momentarily actuated so as to index the filter wheel 210 to a desired station. The switches 331-334 may, of course, take any desired form, and have merely been indicated diagrammatically in FIG. 5.

Referring to FIG. 6, various of the components previously referred to have been indicated by electrical symbols, and for convenience of correlation of FIG. 6 with FIGS. 1 through 5, the same reference characters have been utilized. In particular, FIG. 6 shows symbolically a light source 201, associated photocells, 203 and 260, filter wheel drive motor 209, control solenoid 240, and permanent magnet 243 which rotates with the filter wheel 210 so as to represent the angular position of the filter wheel. Also shown in FIG. 6, are the four heaters 251 associated with photocell 203, and the four heaters 261 associated with the photocell 260. Further, lamps 321-327, millivoltmeter 330 and switches 331-334 of the optical analyzer unit 300 have been symbolically indicated in FIG. 6.

Referring first to the components associated with the upper sensing head 11, there is illustrated in the upper left part of FIG. 6 a diode 340 connected across solenoid 240. For diagrammatic purposes, permanent magnet 243 is shown arranged between two series of reed switches 341-347 and 351-357. A further reed switch 358 is indicated for actuation in the number 1 position of the filter wheel 210 along with switches 341 and 351. The conductors 359 and 360 associated with switch 358 may be connected with the optical analyzer unit 300, and may be connected via the optical analyzer unit 300 with a remote computer, where the illustrated apparatus forms part of a computer control system for controlling the associated paper machinery.

The reed switches 341-347 are shown as being associated with an operational amplifier 361, so that switches 341-347 serve to select the desired value of feed back resistance for the amplifier in each position of the filter wheel 210. Thus, switches 341-347 served to selectively connect in parallel with resistance 370, additional resistance values 371-377, respectively, for adjusting the total resistance between the input and output terminals of the amplifier 361. Thus, in the number 1 position of the filter wheel, permanent magnet 243 is in a position to actuate switch 341, and connect resistance value 371 in parallel with resistor 370. As will hereinafter be explained, resistance means 371-377 may include variable resistors for adjustment so as to provide the desired gain of amplifier 361 in the respective filter positions, or fixed resistance values may be inserted as indicated, once the desired values have been determined for a given filter wheel. As indicated in FIG. 6, the output of amplifier 361 may be transmitted by means of shielded cables 381 and 382. These cables form part of the overall cable indicated at 51 in FIG. 5 leading from the upper sensing head 11 to the power supply unit 301.

Also forming part of the cable 51 would be the conductors such as indicated at 383 from the respective reed switches 351-357. These conductors such as 383 would connect with respective conductors 391-397 of cable 52 leading from the power supply 301 to the lower sensing head 12.

Included as part of the power supply unit 301 would be components such as relay actuating coil 401, associated normally open contact 402, and resistors 403 and 404 shown at the upper left in FIG. 6. Further, the power supply would include an adjustable direct current lamp power supply component 410 for supplying a precisely adjusted or controlled electrical energization for light source 201. Further, of course, the power supply would supply the required direct current operating potentials for the upper sensing head as indicated in FIG. 6.

The lower left section of FIG. 6 illustrates the electrical components of the lower sensing head 12. In the lower sensing head, conductors 391-397 control energization of the operating coils of respective relays K1 through K7. With the permanent magnet 243 in the number 1 position, reed switch 351 is closed, and operating coil 420 of relay K1 is energized closing the associated relay contact 421. The remaining relays K2 through K7 are deenergized, so that the respective associated contacts 422-427 remain open. The contacts 421-427 serve to control the resistance in the feed back path of operational amplifier 429 in conjunction with resistor 430 and resistance means 431-437. As explained in reference to the upper sensing head, resistance means 431-437 may include adjustable resistors, or fixed resistors as shown selected to provide the desired gain of amplifier 429 for the respective positions of the filter wheel 210. The shielded cables 441 and 442 from the output of amplifier 429 connect with power supply unit 301 as part of cable 52. The outputs from the amplifiers 361 and 429 are conducted from the power supply unit 301 to the optical analyzer unit 300 via signal conduit 311, and within the optical analyzer unit connect with respective terminals of the selector switch 331 as indicated at the lower part of FIG. 6. Thus, in the upper position of the selector 331, the output of amplifier 361 is connected with the meter 330, while in the lower position of selector 331, the output of amplifier 429 is supplied to the meter 330. Of course, the optical analyzer 300 may further include analogue to digital converters for converting the outputs of the amplifiers 361 and 429 to digital form for transmission to a remote computer, for example. It will be apparent to those skilled in the art that the remote computer could be programmed to control the sequential actuation of relay 401 during each increment of scanning movement of the monitoring device 10 so as to obtain readings from each desired sampling region of the web 14 for each of the seven positions of the filter wheel 210. The remote computer would then be in a position to correspondingly determine the average optical characteristics of a given length section of the paper web 14, for example, and control suitable inputs to the paper machine so as to maintain desired optical characteristics of the paper being manufactured. Alternatively, of course, the arrangement of FIGS. 1-6 can be utilized simply to take readings from the meter 330 for each filter wheel position during scanning of the web, so as to obtain readings reflecting the optical characteristics of the length sections of the web so scanned. Still further, of course, the circuitry of FIGS. 5 and 6 can be utilized either with the monitoring device located in a fixed position relative to the width of the web (by means of a C-type frame), or with the device off-line from the paper machine, so as to obtain desired readings from the meter 330 for each position of the filter wheel 210 during optical excitation of a single sheet sample of the web held in a sample holder so as to be disposed essentially as indicated for the web 14 in FIG. 3.

Commercially available components which are included in the present design of FIGS. 1-6 are as follows.

Main power supply. Lambda Electronics Corporation Model LQS-DA-5124 providing a direct current (DC) output voltage of 24 volts and a maximum current at 40.degree. C. of 5 amperes.

Reed switches. For reflectance amplifier gain settings--Model MMRR-2, and for transmittance amplifier gain settings-Model MINI-2, manufactured by Hamlin, Inc. The relays in the lower sensing head of Type 821A of Grigsby-Barton, Inc.

Operational amplifiers, Model 233J chopper stabilized amplifiers of Analog Devices, Inc. Model 904 power supply supplying plus or minus 15 volts with a minimum full load output current of plus or minus 50 milliamperes.

Digital panel meter (used for off-line studies and for on-line operation before being interfaced with the computer). Weston Model 1290.

Filter wheel advance solenoid Type T 12.times.13-C-24 volt DC flat plug plunger of Guardian Electric Manufacturing Company. Antibottoming washer made of polyurethane rubber. Operation of the solenoid until interfaced with the computer has been with the use of a time adjusted relay, namely a Model CG 102A6 transistorized repeat cycle timer of G. & W. Eagle Signal Co.

Filter wheel drive motor. Type 1AD3001 Siemens brushless DC motor. The drive belt and pulleys for coupling the motor 209 with the shaft 208 are specified as positive drive belt FS-80 and positive drive pulleys FC5-20 and FC5-40 of PIC Design Corporation, a Benrus subsidiary. The belt has a stainless steel core and the pulleys have a 1/4 inch diameter bore.

In preparing the monitoring device for on-line operation on the paper machine, the zero to 140 millivolt DC signals from the sensing heads will be supplied to respective emf-to-current converters of component 501, FIG. 6. As an example, Rochester Instrument Systems Model SC-1304 emf-to-current converters may be used. Such a converter will provide an output of 10 to 50 milliamperes DC suitable for driving an analog to digital converter at the computer. The emf-to-current converters will provide an isolated input and output so that grounding will not be a problem.

The converters are component 501, will be housed with optical analyzer 300, FIG. 5, and will connect with respective points thirty one of Groups five hundred and six hundred (not shown) at the control computer analog signal input via conductors such as indicated at 502 and 503 in FIG. 6.

Conductors 505 and 506, FIG. 6, associated with filter wheel indexing solenoid 240, FIGS. 3 and 6, may extend within control conduit 312, FIG. 5, and connect with the control computer output terminals at a location designated Group forty two hundred and six, point nineteen (not shown). (Switch 334 should remain open (off) during computer operation of FIGS. 1-6.)

Conductors 359 and 360, FIG. 6, may connect with an input of the control computer at a location designated Group fourteen hundred, point twenty-three (not shown).

A prototype optical monitoring device was first constructed so as to test the feasibility of the concepts of the present invention. As a result of the experimental work with the prototype system, a preferred system has been designed and will hereinafter be described in greater detail. Since the operation of the prototype system is somewhat different from that of the later designed system, a description of the prototype system will serve to illustrate alternative features and an alternative method of operation in accordance with the present invention.

In the original setting up of the prototype system, the upper and lower sensing heads should be brought into proper alignment and spacing. The spacing should be just under 1/4-inch between the case 110 and the surface of the diffusing glass of window 135. (In the prototype unit, there were no additional parts between the case 110 and window 135 such as the shoe plate 122 shown in FIG. 3.) The lower sensing head should be moved laterally in all directions to locate the point where the maximum reading occurs from photocell 260 as well as the point of least sensitivity to relative movement of the upper and lower sensing heads. In an initial calibration of the prototype monitoring device, potentiometers are included as part of the resistance means 371-377 and 431-437 and are adjusted for the respective positions of the filter wheel 210 to give the correct readings for the reflectance and transmittance of the diffusing glass 135 (in the absence any paper sample between the upper and lower sensing heads). The values which were used in this initial calibration are indicative of percentage absolute reflectance and transmittance on a scale of 100, and are as follows:

The readings in millivolts can be converted to other desired units by comparing the readings in millivolts for a given paper specimen with the readings obtained with a standard laboratory instrument, measuring the reflectance of the specimen with the laboratory instrument while backing the paper sheet with a piece of Lucalux and a black body. By measuring the reflectance of the single sheet backed with a black body (no fluorescence), the value of transmittance for the specimen can be calculated and this calculated value utilized for calibrating the lower sensing head. If the fluorescent component is included in the laboratory instrument, and if fluorescence is involved, the fluorescence component can be determined by means of a standard reflection meter, and the fluorescent component can then be subtracted from the measured data before making the calculation of transmittance.

The laboratory testing of the prototype system confirmed that a monitoring device such as illustrated in FIGS. 1-4 should have a potential accuracy equal to that of comparable off-line testers provided certain web scanning requirements are met.

Laboratory tests were run on color standard samples of the grades and colors usually run on the paper machine shown in FIGS. 1 and 2. In addition, a variety of opaques, and a variety of colored 50 pound and 70 pound offsets were included in the tests. A four centimeter diameter circle was scribed on each sample to insure that all tests would be done within the same 12 square centimeter section of the sample. Values of R.sub.o, R.sub.oo, and TAPPI opacity measurements were made on the available standard laboratory instruments. All test were made on the felt side of the sample with the grain in the standard direction. For R.sub.oo measurements, the samples were backed by piles of tabs cut from the edge of the same sheet of paper. In addition to the TAPPI opacity measured on the standard opacimeter, TAPPI opacity was calculated via Kubelka-Munk theory from data obtained with a standard automatic color-brightness tester.

The sample paper samples were clamped to a holder which held the sample under tension with the lower head of the monitoring device bellying 1/8-inch to 1/4-inch into the sheet. The grain of the sheet was oriented parallel to the longitudinal axis of the upper sensing head (that is the machine direction of the sheet was in the same orientation as would occur on the paper machine as indicated in FIGS. 1 and 2). The felt side was always up. Care was taken to make sure that the tested area was within the twelve square centimeter circle scribed on the sample.

The transmittance and reflectance readings were taken from a digital volt meter attached to the output terminals of amplifiers 361 and 429. Calibration data was taken off the Lucalux with no sheet present. Test values were taken on all filters with the sheet in place. The transmittance and reflectance values were keyed into a standard calculator with the calibration data. The calculator was programmed to calculate the color (in C.I.E. X, Y, Z, for example), fluorescent component, brightness, TAPPI opacity and printing opacity (based on Y.sub.c). By supplying the basis weight, the computer could also be requested to calculate s, the scattering coefficient (an index of the effect of pigment efficiency and fiber surface area), and k, the absorption coefficient (an index of the effectiveness of dyes in the sheet). The coefficients s and k are essentially independent of basis weight. Kubelka-Munk theory is the basis of the calculations used.

All of the samples were tested without changing the relative position of the two sensing heads. One set of data was obtained with the heads in a variety of positions to determine the effect of geometric variations.

Since fluorescence is not compatible with Kubelka-Munk theory, the prototype system was carefully designed so that all data used for Kubelka-Munk analyses have excluded fluorescence. The prototype system measures fluorescence separately. A fluorescent contribution is determined from the prototype data by subtracting the Z distribution reflectance without fluorescence (filter wheel position No. 4) from the Z distribution reflectance with fluorescence (filter wheel position No. 7), and multiplying by the appropriate factor.

An independent check on fluorescence measurements, a modified brightness tester was utilized which had a filter wheel allowing for standard brightness and Z distribution filters to be put in the reflected beam. In addition, the filter wheel contained brightness and Z distribution filters which had been modified by removing the ultraviolet absorbing component of these filters. A special mount allows the operator to put the appropriate ultraviolet absorbing filter in the incident beam. Thus, measurements of brightness and C.I.E. Z tristimulus, with and without fluorescence, could be made. Fluorescent contributions were calculated by difference. Some measurements were made on signal sheets with a standard backing. Most of the samples were measured with an infinite pack of tabs. The incident beam filter of the prototype's No. 7 position was such that it permitted about twice the standard quantity of ultraviolet light to strike the specimen. Consequently, measurements of the fluorescent contribution measured on the modified brightness tester and the prototype system correlated well (correlation coefficient of 0.992) but the modified brightness tester value is only 0.528 as large as that measured by the prototype system. Calculations of prototype data now involve calculation of the fluorescent component by multiplying the difference of filter positions No. 7 and No. 4 by 0.528.

Because only one fluorescent dye (Tinopal) in all of the paper specimens was used, the fluorescent contribution needed to be measured only once. The prototype data provides a basis for measuring the fluorescent component Z. Measurements by an independent laboratory showed that the paper specimens do not fluoresce significantly in the X (red) or Y distributions; therefore, fluorescent contributions need only be determined for the blue colored distributions. A linear regression was run on the independent laboratory data which demonstrated that the fluorescent component for X (blue) can be predicted by multiplying the fluorescent component for Z by 1.204. A regression run on fluorescent data from the modified brightness tester shows that the fluorescent contribution for brightness can be calculated by multiplying the fluorescent contribution for Z by 0.864. In summary, fluorescent contributions are calculated by the following formulas:

F.sub.Z =0.528 (Z reflectance with fluorescence minus Z reflectance without fluorescence.)

F.sub.X(blue) =1.204 F.sub.Z

F.sub.Brightness =0.864 F.sub.Z

These fluorescent contributions are added to the respective calculated R.sub.oo values when calculating optical properties from prototype data. The test results for fluorescent and non-fluorescent papers agree with values measured on the standard automatic color-brightness tester.

The following details concerning the results of life testing of the prototype system are considered to reflect minor problems of construction and operation which considered individually are readily corrected for by those skilled in the art. In order to minimize the burden of the total number of such minor problems, and thus to expedite practice of the prototype system, solutions to the various problems which were encountered are briefly referred to.

The filter wheel is advanced by a low torque stallable motor. A timing belt links sprockets on the motor and the filter wheel shaft. The original timing belt had a dacron core. The core of the original belt broke in two places resulting in stretching and eventual loss of teeth. Uneven rate of rotation of the filter wheel occurred due to binding of the belt. Eventually, the plastic drive sprocket broke. Both sprockets were replaced with stainless steel sprockets and the timing belt was replaced with a belt containing a steel core. Installation of the steel sprockets and steel core belt revealed that excessive belt tension could stall the motor. The motor mount holes were slotted allowing the motor to pivot slightly around one mounting screw. Belt tension was adjusted by pivoting the motor. It is concluded that future models should include an idler wheel or some other means of adjusting the tension of the timing belt.

Some problems were experienced with respect to indexing of the filter wheel with the ratchet arm sticking on the tooth so that the ratchet arm does not clear the tooth when a command is given to index the filter wheel. The remedy has been to reduce the roughness of the mating surfaces by filing on the tooth, or smoothing the tooth with a stone. In future models, the shapes and/or smoothness of the ratchet arm and the teeth should be altered to minimize sticking. One solution would be to provide the ratchet arm and the teeth with highly polished mating surfaces.

The ratchet arm is lifted by a 24 volt direct current solenoid. After some time, the plunger of the solenoid became magnetized and would stick to the inside of the coil. This "hanging up" would prevent the ratchet arm from catching the next tooth. A resistor was installed in series with the solenoid coil to reduce the strength of the magnetic field. The plunger of the solenoid was coated with a special material. The coated plunger worked well for about three months before it, too, magnetized enough to hang up. The solution adopted was to provide the solenoid with a flat topped plunger which is stopped at the end of its stroke by a bumper of rubber-like material.

The response of a photocell is somewhat temperature sensitive. For this reason, it is necessary to keep the photocells at a constant temperature. Ambient temperatures on the O-frame of the No. 6 paper machine indicated in FIGS. 1 and 2 have been measured as high as 118.degree. F.(48.degree. C.) in the summer. The photocells in both heads are mounted in massive metal blocks. Each metal block has four thermistor heaters mounted in close proximity to the photocell. These thermistors have switching temperatures of 55.degree. C., (that is about 130.degree. F.). The intention of this design was to add enough heat to the instrument to hold the temperature steady at about 55.degree. C. During bench studies, this temperature was never reached due to the low capacity of the heaters. At machine room temperatures, however, the instrument temperature may reach 55.degree. C.

During the bench studies, it was found that the heaters did minimize temperature variations. The few degrees of temperature variation that were observed during normal operation usually occurred slowly. Changes in instrument temperature affected the output signal less than acticipated. Based on this experience in the laboratory, the maximum variation in head temperature should be less than 3.degree. F. per hour. Temperature variations of this magnitude will not have a significant effect on the output signal. Long term temperature changes would be corrected for by the calibrations each time the head goes off web.

In the laboratory, there was a minimum of dirt problems. On the machine, however, the hole could allow dirt to enter the upper head. Up to a point, dirt on the lenses and filters will be corrected for by the periodic calibration routine. Excessive dirt, however, will reduce the sensitivity of the instrument and may even affect its accuracy. Periodic cleaning of the lexses and filters will be required. If dirt accumulates too rapidly, it may be necessary to attach an air purge to the upper head.

The lower head of the prototype system is completely sealed so that no dirt problem is anticipated inside the lower head. Because the Lucolux window is in contact with the sheet, friction will keep it clean.

Most of the filters consisted of two or three component parts. There have been some problems with dirt getting between the components of the filters.

The case on the lower head as well as the case on the upper head should allow most general maintenance and trouble-shooting to be done without dismounting the head. A completely removable case would be desirable. At a minimum access should be provided for the following: (1) convenient light bulb change, (available on the prototype), (2) cleaning of lenses, (available on the prototype), (3) cleaning of the filters. (Access is presently available to one side of each filter. The side which is most likely to collect dirt is not accessible in the prototype.) (4) The amplifier. The amplifier is a standard plug-in module. In the event of a breakdown it could be replaced in seconds if it is accessible. Furthermore, it is necessary to remove the amplifier to do any trouble-shooting on the gain circuitry. (5) The circuit board holding all of the gain control resistors. The choice of gain circuitry is controlled by reed switches which are not accessible on the prototype without a partial disassembly of the instrument. Malfunctions of the reed switches, however, can easily be diagnosed by removing the amplifier and taking resistance measurements on the gain control circuits. There is also the possibility of mechanical or electrical damage to a resistor or a potentiometer mounted on this circuit board. With proper access a damaged part could be replaced in five to twenty minutes. (6) The photocell. With proper access, the photocell could be replaced quickly and easily. (7) The heater. The heater are adjacent to the photocell and are generally just as easily serviced. (8) Indexing mechanism. The present accessibility to the ratchet teeth, rachet arm and solenoid is adequate but not very convenient on the prototype. A certain amount of access to these parts is needed to correct chronic indexing problems such as sticking and "hanging up".

The filters are presently mounted in the filter wheel of the prototype by spring clips. Most of the filters are compound filters containing as many as four component pieces of glass. During laboratory trials, increases in the optical density of a filter were frequently observed which could not be corrected by cleaning the surfaces of the filter. Upon removing one of the filters, it was discovered that foreign material was collecting between the components of the compound filter. The use of lens cleaning solution on the filters may have accelerated the problem if capillary action drew foreign material between the components. A set of gaskets and some type of threaded mount should be used to mount the filters in such a way as to minimize foreign material (including cleaning solutions) from getting between the components of compound filters.

In mounting the prototype sensing heads on an O-frame, it is necessary to bring the geometric alignment of the heads as close to their optimum relationship as possible. The original intention was to set the gap between the heads with the aid of a spacer; however, flexibility of the sheet metal case of the prototype upper sensing head prevented the use of a spacer for setting the gap. Accordingly, the shoe plate 122 of the new upper sensing head shown in FIG. 3 has been made of a thickness and consequent rigidity so as to enable the use of a spacer gauge to set the gap between the upper and lower heads. (The gap is reduced by 1/16 inch to 3/16 inch because of the thickness of shoe plate 122.)

The gap between the heads is a most critical dimension as far as calibration and reproducability is concerned. In the prototype it was intended to calibrate relative to an average gap, thus correcting the readings for variations in the gap from the average gap.

One of the criteria used in designing the prototype was minimum head length in the machine direction. Unfortunately, the upper head was turned 90.degree. in order to give the prototype unit the same geometry as the General Electric Brightness Meter, Automatic Color-Brightness Tester, and Hunterlab Color Meter. In this new position, the prototype head is 121/4 inches long in the machine direction plus 21/2 inches for cable connectors. Redesign should be possible to reduce the machine direction dimension to about 8 inches and to relocate the position of the cable connections.

The lining of the case for the upper head should be matte as well as black to prevent reflection of ambient light within the case and a possible spurious effect on the photocell reading.

Following the correction of miscellaneous start up problems the prototype system was found to function well mechanically. As a test of its durability, the prototype system was placed in continuous operation for a period of over ten months and no serious mechanical problems resulted except the failure of the solenoid. The solenoid failure was expected and the replacement solenoid is of a design which is expected to give a long service life. The light application of silicone lubricant spray to the indexing control ratchet arm and cooperating teeth corrected a problem of malfunctioning of the filter wheel indexing mechanism (which occurred on two occasions during the ten months). The prototype system was not intended to be a low maintenance instrument; however, the experience during the durability test with the prototype in continuous operation indicates that the prototype system should operate on a paper machine with an acceptably small amount of down time.

In the prototype system, potentiometers are included as part of the resistance means 371-377 and 431-437 and are adjusted for the respective positions of the filter wheel 210 to give desired values such as given in the foregoing Table 1. In the preferred system of FIGS. 3-6, these potentiometers for adjusting amplifier gain are omitted and are replaced with fixed resistors 371-377 and 431-437 selected to give scale readings from meter 330 in the respective filter wheel positions which are well above the values given in the preceding Table 1. This is intended to improve the stability and increase the sensitivity of measurement.

In calculating optical parameters from measurements relative to various samples, values were first established for the reflectance RD of the diffuser 135, FIG. 3, in the absence of a paper specimen, for each filter wheel position. Initially calculated values for RD were used in a first computation of optical values, and then the values of RD were adjusted slightly to give the best agreement with the corresponding optical measurements by means of the standard automatic color-brightness tester. The following table shows the reflectance values which were established for certain laboratory testing of the system of FIGS. 3-6.

The transmittance of the diffusing glass 135 need not be known since the ratio of the transmittance of the diffusing glass and paper (in series) to the transmittance of the diffusing glass is employed in calculating the desired optical parameters.

A computer program was developed to process the data collected during laboratory operation of the monitoring device 10 as well as to compare the calculated reflectance value R.sub.oo and the calculated fluorescent components with the data collected with the standard automatic color-brightness tester. A listing of the symbols employed in a symbolic statement of the computer program in the Fortran computer language utilized in this laboratory study is set forth in Table 3 on the following pages.

In the foregoing listing of symbols, the letters of the symbol OMOD are taken from the phrase on-machine optical device; however, this particular section of the specification refers to a system essentially conforming to the system of FIGS. 3-6 operated to measure optical properties of individual paper sheets under laboratory conditions. (The laboratory work here reported was with an earlier version of the monitoring device designed for on-machine operation, prior to adoption of a thickened shoe plate 122. The standard spacing between the upper and lower sensing heads for the earlier version was 1/4 inch, rather than 3/16 inch as with the final version of on-machine device as specifically shown in FIG. 3.) The OMOD scale readings are obtained from the meter 330, FIGS. 5 and 6, with the filter wheel 210, FIGS. 3 and 4, in the respective positions to activate the respective filters 281-286 (indicated as "Filters 1 through 6" in the preceding listing) and to activate filters 287 and 288 (indicated as "No. 7 filter" in the listing), and with switch 331, FIG. 5, in its upper position to measure reflectance, and in its lower position to measure transmittance. As to reflectance measurements, the cavity 145 is considered to form essentially a black body backing for the diffusing glass 135.

The symbol "ACBT" in the foregoing listing of symbols is used to designate a measurement made on the standard commercially available automatic color-brightness tester. The brightness measurement obtained from the ACBT represents a value accepted as standard in the U.S. Paper industry. A further appreciation of the importance of the fact that the OMOD measurements can closely conform to this industry standard is gained from a consideration of the article by L. R. Dearth et al "A Study of Photoelectric Instruments for the Measurement of Color Reflectance, and Transmittance, XVI. Automatic Color-Brightness Tester", Tappi, The Journal of the Technical Association of the Pulp and Paper Industry, Vol. 50, No. 2, February 1967, pages 51A through 58A. As explained in this article, the ACBT is photometrically accurate, and the spectral response is correct for the measurement of both color and standard brightness. The spectral response of the ACBT very nearly matches the theoretical CIE functions as indicated by the special technique for determining spectral response. This involves the determination of the tristimulus values for deeply saturated colored glass filters a very rigorous check on the spectral response, especially when it is noted that colored papers are less saturated.

The symbols in the foregoing Listing of Symbols which as shown include lower case characters may also be written exclusively with capital letters. This form of the symbols is convenient for computer printout. The alternate forms of these symbols are as follows: AR.sub.oo FC or AROOFC; R.sub.o or RO; R.sub.oo or ROO and R.sub.oo FC or ROOFC.

In the foregoing Table 4, the symbols representing basic mathematical operations were as follows:

To indicate more concretely the calculations which are performed, the following Table 5 will illustrate exemplary input and output data for a given sample. The meaning of the various symbols will be apparent from the listing of the symbols of Table 3:

In the foregoing table showing exemplary input and output data, the input and output data symbols have been shown as they are actually printed out by the computer with all letters capitalized. In the text, certain of the input and output data symbols are shown in a more conventional manner with subscripts since the symbols are more familiar in such form.

The data such as exemplified in Table 5 are based on a single determination for each specimen. The "grade correction" GC is based on the average difference between R.sub.oo FC and AR.sub.oo FC for two specimens, specimens A and B.

The data as exemplified in Table 5 show that there is generally good agreement between the calculated R.sub.oo FC and AR.sub.oo FC values. The spread in values for the duplicate specimens (A and B) is good with the exception of several samples. Some difficulty was experienced in positioning the specimen on the monitoring device 10 to give reproducible results. The difficulty should be minimized when the unit is placed "on-machine". The grade correction GC takes this discrepancy into consideration so the correction should be established "on-machine".

The RD values shown in Table 5 were punched into the first data card along with the values for RK and TK for input to the computer in advance of a desired computation. The factors RK and TK were included as factors in the computations so that the transmittance and reflectance values could be adjusted independently, if desired. In this evaluation, RK and TK were left at 1.000. (Calculated values for RD were used in a first computer run and then the values were adjusted slightly to give the best agreement with the standard automatic color-brightness tester. The values for RD shown in Table 5 are the slightly adjusted values utilized in obtaining the data discussed in this section of the specification.)

A second set of data for the same fourteen samples was collected using the monitoring device in the same condition as for the collection of the data previously given. All of the variables were left the same to see how closely the data could be reproduced for the identical specimens. The agreement was quite good except for samples 8 and 14. It appears that the paper may not have been lying flat in one or the other tests. The grade correction GC on some of the grades was changed and the second set of data was again calculated for samples 1, 2, 4, 5, 6, 8 and 14. This improved the agreement between the monitoring device and the standard automatic color-brightness tester.

The reflectance head of the monitoring device was then lowered 0.025 inch and another set of data was collected for the same seven samples. The same ACBT data was used. The data show that lowering the reflectance head reduces the reflectance while transmittance remains essentially unchanged. The effects are not as large as was expected and could be corrected through adjustment of RK; however, the variables RK, TK and GC were again held constant.

The reflectance head was then raised to a spacing of 0.050 inch (0.025 inch above the normal position for these tests), and another set of data was collected for the same seven samples. The effects were larger than when the reflectance head 11 was lowered. Again, an adjustment of RK would improve the agreement.

It was concluded from these test results that a change of plus or minus 0.025 inch from "normal position" is larger than can be tolerated. An estimate of a reasonable tolerance, based on this and earlier work, would be plus or minus 0.010 inch from "normal position".

All of the variables used in calculating the data for samples 1, 2, 4, 5, 6, 8 and 14, after the initial change in the grade correction GC, were held the same to determine the effects of changing the reflectance head position. The same input data for the case of the reflectance head being raised 0.025 inch were processed again but with RK equal to 0.975 instead of 1.000. This reduces the reflectance value to the proper level. The data obtained in this way show good agreement between the monitoring device and the standard automatic color-brightness tester. Apparently the factor RK can be used quite effectively in adjusting for some variation in the geometric relationship of the upper and lower sensing heads. It would be preferred, of course, to maintain proper alignment and spacing.

A second set of samples were evaluated after returning the reflectance head to its normal spacing from the transmittance head. Before calculating new output data, the computer program of Table 4 was corrected in statements S.0022 and S.0028 by changing RD to RD4. The corrected computer program has been shown herein since the error in the previously referred to data was insignificant in most cases. Thus with the corrected computer program, the input data for the second set of samples were processed. The values RK and TK were set to 1.000 and the same grade corrections were used as for samples 1, 2, 4, 5, 6, 8 and 14 previously referred to.

Conclusions drawn from all of the data are that the grade correction GC will handle errors resulting from less than ideal characteristics of the monitoring device 10 such as the relatively wide bandwidth of light transmitted in the various filter positions in comparison to the requirements of Kubelka-Munk theory and the fact that this theory applies strictly only to diffuse light rather than collimated light as actually employed in the illustrated monitoring device 10. This correction must be established "on-machine". Use of the diffusing glass 135 to calibrate the monitoring device 10 will handle changes in light level, photocell sensitivity and amplifier gain. The reflectances RD of the diffusing glass 135 for the various filters as established in the present work are set forth in the previous Table 2 entitled "Table Showing Reflectance of the Diffusing Glass With No Paper Specimen Present in a Laboratory Test of the System of FIGS. 1-6".

As previously mentioned, the transmittance of the diffusing glass 135 need not be known as the ratio of the transmittance of the diffusing glass and paper (in series), identified by the symbol TSP, to the transmittance of the diffusing glass 135, identified by the symbol TSD, is employed as will be apparent from the explanation of the calculations employed set forth hereinafter.

The fluorescent component is handled through the difference in reflectance as measured with the number 4 and the number 7 filters (RPD7 minus RPD4). The factors used in the subject computations, for filters number 1, 2 and 4, are 0.500, 0.600 and 0.550 respectively. This means of determining the fluorescent contribution FC appears to be successful.

The factor RK whereby the reflectance can be adjusted to account for misalignment or incorrect spacing seems to function better than was expected.

The following examples will serve to explain the calculations of the output data for the different filter positions in greater detail.

On the basis of further experimental data, the factors relating the fluorescent component, as measured on the monitoring device, to the fluorescent component as measured with the standard automatic color-brightness tester, have the following presently preferred values for filter wheel position numbers 1, 2 and 4: 0.528, 0.636, and 0.456, respectively.

In the prototype system, potentiometers were included as part of the gain control resistance means and were adjusted for the respective positions of the filter wheel 210 to give values correlated directly with absolute reflectance and transmittance of the diffusing glass, such as given in the foregoing Table 1. In the preferred system of FIGS. 1-6, however, these potentiometers for adjusting amplifier gain are omitted and are replaced with fixed resistors 371-377 and 431-437 selected to give scale readings from meter 330 in the respective filter wheel positions which are well above the values given in Table 1. The higher gain values selected for the amplifiers 361 and 429 in the preferred system are intended to provide improved stability and increased sensitivity of measurement.

The upper and lower sensing heads are placed at a spacing of 3/16 inch by means of a gauging plate made of 3/16 inch Teflon. The incident beam 133 forms a light spot of elliptical configuration on the planar upper and surface 98 of the diffusing window 135. The major axis of the elliptical light spot has a length of about 5/8 inch and is parallel to the direction of web movement, i.e. the machine direction, while the minor axis has a length of about 3/8 inch and is at right angles to the machine direction. The reflected beam 137 consists of the total light reflected from a circular spot of approximately 3/8 inch diameter. This viewed area lies substantially within the elliptical illuminated area on surface 98; however, the two essentially coincide in the direction of the minor axis of the illuminated spot.

Since the effective optical aperture 154, FIG. 3, of the lower sensing head is of a diameter of about 15/16 inch, the system will be insensitive to a certain amount of lateral offset between the optical axis 15 of the upper sensing head and the optical axis 515 of the lower sensing head.

In setting up the system, the position of the lower sensing head may be adjusted laterally so that the spot formed by the incident beam 133 is essentially centered on the surface 98 of window 135.

The optimum relationship between the upper and lower sensing heads can be precisely detected by observing the reflectance output from the upper sensing head (in any position of the filter wheel 210) as the heads are moved relative to one another while maintaining the spacing of 3/16 inch between the heads. When the correct geometrical relationship is attained between the incident beam 133, the reflected beam path 137 and the plane of the surface 98 of the window 135, the reflectance signal will have a maximum value.

With the upper and lower sensing heads in the optimum geometric relationship, and with the incident beam impinging on the central part of surface 98, it is considered that relative shifting between the upper and lower heads in the plane of surface 98 over a range of plus or minus 1/8 inch in any lateral direction should have an insignificant effect because of the flat planar configuration of surface 98.

The program subroutine of FIG. 7 accepts digital information from the analog to digital converters of component 501, FIG. 6, at one second intervals. Referring to FIG. 7 where the blocks containing the flow chart steps are individually numbered in their operational sequence, the step 701 represents the entry into the subroutine at one second intervals. Step 702 shows the acceptance of an analog input and conversion to engineering units. Step 703 indicates saving such converted input as a process variable in a scan only file of the digital control computer. A type of control computer which has utilized such a scan program for a number of years for collecting data in an overall paper machine direct digital control system is the General Electric Company PAC 4020 Process Control Computer. Minor additions to the existing program routine will allow for the collection of the reflectance and transmittance data by means of the existing computer system. A suitable computer interface between monitoring device 10 and such a control computer has been described previously. Block 702 suggests that valid reflectance and transmittance values might be limited to a range from 0 to 1.0 units, for example. In this event the program could include provision for checking that the collected reflectance and transmittance values were within the range and for printing out a message or the like if invalid data is received.

Block 704 indicates the sequential reading of process data input points in a predetermined order until the last data input point has been scanned, whereupon the computer exits from the subroutine.

The arrangement of FILE X which is utilized during acquisition of data from the system of FIGS. 1-6 and conversion thereof to desired output paper optical quantities can be visualized from the following Table 7. In the first column of Table 7, sequential memory locations of the process file have been assigned sequential numbers beginning with zero. A convenient label has been assigned to certain groups of sequential memory locations, and this is also given in the first column. (The term FILE X is used to designate all of the locations zero through one hundred forty while subsequent labels refer to only the subadjacent group of sixteen locations or less.)

In general the significance of the various stored data will be apparent from the descriptions given in the righthand column of Table 7 and from the use of the stored data as indicated by the flow charts of FIGS. 8-20.

In referring to locations of FILE X in the program flow charts of FIGS. 8-20, the relative location of FILE X is indicated by the number in parenthesis. Thus FILE X(4) refers to relative location number four of FILE X as given in Table 7. File X(138) corresponds to MPAR(4) in Table 7, and both refer to relative location number one hundred and thirty eight. FILE X(four) is an alternative to FILE X(4), and is used in the text to avoid any possible confusion with drawing reference numerals.

The following general discussion of successive locations or groups of locations of FILE X, taken in numerical order, will serve as an introduction of the description of the program routines of FIGS. 8-20. Contemplated modifications of the programs and of FILE X will be discussed in a later section.

In location O of FILE X, the Status word includes a bit number 23 which is set to a logical one when conditions are met for making standardizing calculations. For example, the OMOD should be off sheet and the beta gauge with which the OMOD is mounted for scanning movement should be in its standardizing mode. The set condition of bit 23 is responded to by the program to bypass data smoothing and to store data in a special table STTABLE at locations 28-34 and 36-42 of Table 7. The condition of bit 23 is reset to logical zero when the filter wheel has indexed through seven positions or if the beta gauge completes standardization before the complete set of OMOD standardization data is collected.

Locations 1 and 2 of FILE X may store the PCW (process control word) addresses for Loops P and Q. Loop P is a subroutine for controlling the processing of reflectance data and Loop Q is concerned with the processing of transmittance data. These loops begin at FIG. 13 of Program Fourteen.

Location three of FILE X, contains the address of the DDC scanner file containing the position of the sensing head 10 along its path of traverse of the web. This position is monitored by the Scan-Only (DDC) routine of FIG. 7 and is stored in the scan system file identified by TAG one hundred twenty nine. A current value of sensing head position is transferred from the referenced DDC file into location three of FILE X periodically.

Location four of FILE X stores a SLOWDOWN COUNT index value which is used to cause a specified number of dummy readings at each filter wheel position to be made after each advance of the filter wheel to allow time for the OMOD electronics to reach steady state, and to allow for any transient error in synchronization between the DDC Scan-Only routine of FIG. 7 and the Data Reduction routine (Program Fourteen) of FIGS. 8-16 which runs at one second intervals under the control of the RTMOS Scheduler (a computer real time operating system of the General Electric Co.).

Location five of FILE X stores a SLOWDOWN INITIAL VALUE COUNT which is used when a processing cycle is being initiated.

Location six of FILE X of Table 7 stores an index value I which represents the estimated filter wheel position, based on the number of actuations of the filter wheel indexing solenoid, since the initial filter wheel position wherein reed switch 358, FIG. 6, is closed in response to the proximity of permanent magnet 243, FIGS. 3 and 6. In the computer program, the successive filter wheel positions are designated 0, 1, 2, 3, 4, 5 and 6, and result in spectral response distributions designated B.sub.R (brightness), X.sub.B (blue portion of the E.sub.c x function), X.sub.R (red portion of the E.sub.c x function), Z (E.sub.c z function without fluorescence), Y.sub.C (E.sub.c y function), Y.sub.a (E.sub.a y function), and Z.sub.FL (E.sub.c z function, with fluorescence).

Location seven of FILE X serves to store the address of the DDC file for basis weight average. TAG BOO, which is used by Program Fourteen to access the basis weight data and store it in location MPAR (4) in floating point format. This will be used by Program Forty-Two during the reduction of data.

Location eight of FILE X stores a value for the minimum on-sheet head position. When the head position is less than such minimum value, a standardization cycle may be set in motion by Program Fourteen.

Location nine of FILE X contains the value of an initialization index K which is used to determine when all seven smoothed input values have been initialized to equal the latest unsmoothed input for each of the reflectance and transmittance channels.

Location ten of FILE X stores a filter wheel cycle completion index designated CYCLE that can be used to determine when a specified number of filter wheel cycles have been completed through the last filter wheel position (position six in the programming notation). This prevents the data reduction program (Program Forty-Two) of FIGS. 17-20 from running until a specified number of data sets have been collected since the last time it ran or the unit was standardized.

Location eleven of FILE X stores a smoothing constant ALPHA which is used in smoothing the input data from FIGS. 1-6 by Program Fourteen, as indicated in FIG. 14.

A temporary location index J is used to point to either the reflectance vector tables or the transmittance vector tables. J is equal to zero to indicate reflectance, and is equal to one to designate transmittance. Each of the tables such as CTABL of table 7 has a first set of locations (e.g. 12 through 18) which can become active while J is equal to zero and a second set of locations (e.g. 20 through 26) which can be selected when J is equal to one. Thus the sets of two numerals in parenthesis at locations 12 through 133 represent respectively the value of the filter wheel position index I and the J index value corresponding to the location.

The table CTABL at locations 12-18 and 20-26 of the process file of Table 7 is used to store a standardization correction factor C. The reflection values of the factor C for the respective filter wheel positions are stored in locations 12-18 and are active when J=0, and I=0, 1, 2, 3, 4, 5 and 6, respectively. Similarly the transmission values for the lower sensing head and the respective filter wheel positions are stored in locations 20-26, which are selected when J=1 and I=0, 1, 2, 3, 4, 5 and 6, respectively.

The table ST TABL, stores data from the OMOD system of FIGS. 1-6 in the standardization mode with the heads in the off-sheet position.

The table RG TABL stores the value RG, the nominal reflectance of the backing for the web in the off-sheet position, for the respective filter wheel positions, and also the value TD, the nominal transmittance of the diffusing window 135 in the off-sheet position. These values may be experimentally determined as previously explained and inserted into table RG TABL at start up of the system.

Table VTABL serves to store input data after it has been processed by Program Fourteen of FIGS. 8-16. The raw data is corrected on the basis of the most recent standardization values from table ST TABL and, multiplied by the correction factors C from table CTABL, and exponentially smoothed by means of the subroutine of FIG. 14 before being stored in the VTABL locations.

The SG TABL table of the process file of Table 7 stores the specific grade correction factors SGCF.

The OUTABL locations store the output quantities as computed under the control of the Data Reduction Program Forty Two of FIGS. 17-20.

The tables STABL and KTABL store the scatter coefficient S and the absorption coefficient K which together serve to characterize the paper web being monitored.

The subroutine of FIGS. 8-16 is referred to as Program Fourteen (or Program 14) and is designed to perform various data acquisition functions as indicated in the flow chart.

It is believed that the flow chart of FIGS. 8-16 will be self-explanatory given the foregoing comments concerning Table 7. The following Tables 8-16 are a tabulation of the blocks of the subroutine with supplementary comments to indicate the meaning of any abbreviations, or to paraphrase any possibly cryptic statements. (Arabic numerals within the blocks in FIGS. 8-20 do not refer to reference numerals of FIGS. 1-6. This is indicated in the following tabulation by spelling of such numerals so far as feasible.)

1. The values stored under RSTABL (118-133) will be identical to the values stored under RGTABL(44-59) and, therefore, the former table will be eliminated when the necessary program changes are also made. The RK and TK correction factor approach allows this simplification of the calculations.

2. Program Fourteen can be further revised to eliminate the need for the nominal diffusor transmittance, T.sub.d, completely. Manipulation of the terms of the equations involved permits this elimination. Note referring to Table 4, S.0023, that TPDOTD, the ratio of TPD/TD, is equal to (TSP* TK)/TSD, eliminating the need to know the absolute value of TD. TD is the computer symbol representing T.sub.d, the transmittance of the diffuser. Refer also to the paragraph following Table 2.

3. Program Fourteen also calls for the re-initialization of the algorithm which smooths the "raw" reflectance and transmittance after each standardization. It is presently considered that this re-initialization will not only be unnecessary, but would add to control problems. Consequently, this will likely be changed so that this smoothing goes on indefinitely after a run start-up. (Note: Do not confuse the smoothing of the "raw" data from the paper with the correction factors acquired during standardization--The latter will not and likely should not be smoothed at all.)

4. Program Fourteen has not as yet been debugged. Debugging can only be accomplished after connecting the computer to the system of FIGS. 1-6 via an A to D converter. It is considered that such debugging is a routine matter well within the skill of the art. It may be noted that the OMOD is now on line as shown in FIGS. 1-6 and data collection has begun.

Program Fourteen is designed to perform various functions described as follows:

1. Sequentially read the reflectance and transmittance values stored in the Process File until both values for each of the seven OMOD filter positions are obtained.

It takes about two seconds from the time the filter wheel is advanced until the photocell readings reach a near equilibrium condition. Program Fourteen is, however, linked timewise to the DDC scan program and is programmed to run every second also. Consequently, any data acquired before the photocells reach a near equilibrium condition, will be liable to intolerable error. Program Fourteen solves this problem by processing data on a multisecond interval basis only, e.g., every 2, 3, 4, etc., seconds depending upon the choice of the value of the term SLOWDOWN which inserted in File X(five).

2. Check the OMOD to see if it is operating properly and issue alarms if it is not. "OMOD Filter Stuck" and "Skipped Filter" alarm messages were made available.

The upper OMOD head is designed with an extra reed switch 358, FIG. 6, which closes when the brightness filter is in the optical train pathway. (Previous description herein refers to the brightness filter as the first position; however, Program Fourteen refers to it as the zero position.) The computer program checks the status of this switch as being open or closed by means of Point 22-Group 1400. The filter index is initialized back to zero each time the status of Point 22-Group 1400 is closed. Discrepancies, should they occur between the expected filter index based on the incremented count and the actual filter position can be readily recognized by this program. This serves as the basis for the alarms previously mentioned.

3. Determine when and how often the optical property Data Reduction Program, No. Forty Two, (see FIGS. 17-20) is to be run. This is controlled by the value chosen for the term "CYCLE".

4. Read the OMOD head position and the average basis weight of the paper being produced and store for use in subsequent calculations or program logic tests. This information is readily available from a basis weight control program which has been in use for several years.

5. Correct the "raw" reflectance and transmittance data by multiplying each of the fourteen values by the appropriate correction factor. The values of these correction factors are updated by the last standardization sequence which occurred prior to their actual use (see 7 below).

6. Exponentially smooth each of the corrected reflectance and transmittance values and store for subsequent calculations.

Exponential smoothing requires a previous value to act upon; however, such previous value is not available for run startup, etc. An initialization technique involving an initialization index, k, is employed to solve this problem. The degree of smoothing is determined by the value chosen for .alpha. (ALPHA).

7. Initialize and control the automatic standardization of the OMOD.

The OMOD heads are mounted next to an Electronic Automation Inc. (EA) basis weight gauge in a piggyback fashion. This EA system utilizes an "O" frame to permit scanning the full web width. It is also designed to automatically retract the carriage 25, 26, FIG. 2, of the scanning mechanism upon which the basis weight gauge and OMOD are mounted, to an offsheet position (FIG. 2) at 1-hour intervals. Program Fourteen takes advantage of this schedule to standardize the OMOD at the same time that the basis weight gauge is being standardized. When offsheet, the very durable Lucalux backing 135, FIG. 3, is always in position to permit checks of its reflectance and transmittance. Due to its durability and inertness, the latter should remain unchanged for long periods of time. In addition, the moving web will insure its cleanliness prior to each standardization occurrence. Consequently, this standardization procedure will allow for accurate updating of the correction factors for each filter position. In so doing, it compensates for any changes which may inadvertently occur in the light source, filters, photocells, lenses, electronic amplification, etc.

Two overall geometrical correction factors are also employed at this point of the program to adjust for any relative head spacing or alignment change that may also inadvertently occur. Experimental data has shown that the same geometrical correction can be used for each reflectance measurement. The values of these two factors are, however, not determined automatically, but must be determined by external means involving offline audit testing by comparing OMOD readings with those of off-line standard laboratory instruments before being fed into the proper computer storage. Initial values of these two factors will be unity; in which case, the relative head geometry will be assumed to be in standard condition and no geometrical correction factor required.

The alternative to using these geometrical correction factors is to realign and/or respace the heads when needed. In the case of minor adjustments, the former approach is clearly the more desirable where the heads are in an inaccessible location and functioning on a high-speed paper machine with little downtime available for such mechanical readjustments.

Program fourteen as presently devised, does not call for the exponential smoothing of the correction factors updated upon each standardization. This could be easily changed should on-line experience indicate that such smoothing is desirable.

8. Control the advance of the OMOD filter wheel 210, FIG. 4, to the next filter position at the desired time interval. This is accomplished by the computer 996, FIG. 6, directing the closure of a loop 505, 506, FIG. 6, which energizes the solenoid. The energized solenoid lifts the rachet arm 230, FIG. 3, clear of the lug against which it was previously braced. The filter wheel shaft is under a continuous torque, tending to rotate it at all times. Thus, it begins to rotate when freed of the holding ratchet arm; but it is stopped again at the next lug, since by then the solenoid attached to the ratchet arm is once again de-energized by computer command. The low torque motor 209, FIG. 6, designed to be stalled indefinitely without harm provides the necessary filter wheel torque.

9. Program fourteen reschedules itself to run again in approximately 1 second.

The purpose of this program is to reduce the corrected reflectance and transmittance data into terms with which papermakers are familiar and upon which paper optical specifications are based; e.g., brightness, opacity, color and fluorescence. A description of this program follows.

The following Tables will serve to supplement the labels applied to the blocks of the flow chart illustrating this program.

1. Although not indicated as yet, the output of the fluorescent contribution to TAPPI brightness will be part of the computer output when the programs are finalized.

2. Program Forty-Two has been checked out against a currently operating program used on a research Hewlett-Packard computer and both give the same results.

3. It is planned to study means of determining and using the Specific Grade Correction Factor other than that described in Program Forty-Two. It may be decided to apply such correction directly to R.sub.oo rather than to the smoothed values of T.sub.pd and R.sub.g. The transmittance of the paper, T, may not need any Specific Grade Correction and could then be used along with the corrected R.sub.oo to compute the scattering and absorption coefficients s and k. The latter will be very useful and may represent preferred parameters for closed loop control.

The purpose of this program is to reduce the corrected reflectance and transmittance data into terms with which papermakers are familiar and upon which paper optical specifications are based; e.g., brightness, opacity, color and fluorescence. A description of this program follows.

1. The data reduction steps of this program are performed only if the term "CYCLE" which is decremented in Program Fourteen, is zero or negative. Otherwise, this data reduction routine is by-passed completely.

2. The exponentially smoothed reflectance and transmittance data acquired by Program Fourteen are first corrected by multiplying each of the fourteen values by an appropriate Specific Grade Correction Factor (SGCF). With a few exceptions the SGCF'S provide only small corrections, if any at all. The SGCF'S serve two purposes.

a. They compensate for the small errors resulting from the use of the Kubelka-Munk or energy balance equations when the latter do not apply exactly.

b. They allow the reduced data values to precisely correspond to any one of several possible off-line instruments. Existing off-line instruments do not agree among themselves. Thus, the choice of the off-line instrument for performing the audit testing will affect the values of the SGCF'S.

3. The next part of Program Forty-Two computes and stores the values of R.sub.o, T, Roo, S and K of the paper being measured for each of the first six of the seven OMOD filter positions. Kubelka-Munk and IPC derived energy balance equations are used for this purpose. Note that prior to these calculations, the reflectance values were those of the single ply of paper when backed by the Lucalux and were symbolized by R.sub.g. Similarly, the former transmittance values were those of the single ply of paper in series with the Lucalux, now serving as a diffusing window. This was symbolized by T.sub.pd.

The phenomenon of fluorescence is not accounted for by Kubelka-Munk theory. For this reason the OMOD optical geometry was chosen to exclude fluorescence by eliminating ultra violet (U.V.) light from the incident light beam of the first six filter positions. For reasons explained later, the seventh filter position permits the reflectance of the Z function with ultra violet energy present in the incident beam.

4. The degree of fluorescence as measured by the "Fluroescent Contribution" is determined next. Fluorescence occurs as a result of excitation of special dyes (optical brightness or fluorescent dyes) by ultra violet energy contained within the incident light beam(s). The "Fluorescent Contribution" is defined here to be the increase of the reflected light flux that occurs as a result of the existence of some standard quantity of ultra violet energy in the incident light beam.

Such U.V. energy is rapidly absorbed by the outer layers of most conventional papers. Consequently, fluorescence is primarily a characteristic of the surface of the paper being viewed. Thus, for practical purposes, the value of [R.sub.g (with fluor.)-R.sub.g (without fluor)]=[R.sub.oo (with fluor.)-R.sub.oo (without fluor.)] when the same incident light beam containing the same U.V. energy is used in both cases. The right side of this equation is by the definition above, the Fluorescent Contribution provided the standard quantity of U.V. light is employed in the incident beam. The left side of this equation is a quantity measureable on a single ply of the moving paper web. In the case of the OMOD a measure of the fluorescent contribution to the Z function reflectance is obtained from the term, [R.sub.g (Filter No. 7)R.sub.g (Filter No. 4)]. The filter arrangement existing in the No. 7 OMOD filter position permits about twice the standard quantity of U.V. energy to strike the paper. (The U.V. energy in the incident beam of the Standard TAPPI Brightness Tester is considered to be the standard quantity here). This increases the sensitivity of this measurement by two-fold. It also necessitates the use of a proportionality constant of approximately one-half to compute the value of the standard Fluorescent Contribution to the Z function reflectance.

The Fluorescent Contribution to the reflectance of the X.sub.R and Brightness functions can be computed directly from the Z function Fluorescent Contribution. The multiplication factors involved are constant for a given optical brightner and need to be changed only if the type of optical brightner is changed. The Fluorescent Contributor to the Y.sub.C, Y.sub.A and X.sub.R functions can be ignored as being inconsequential to the typical optical brightner used in the paper industry today.

5. Defining equations are used to compute and store for accessible putout values of the following:

a. Standard TAPPI Brightness

b. Printing opacity based on illuminant C

c. R.sub.89 based on illuminant A

d. TAPPI opacity based on illuminant A

e. X Tristimulus value

f. Y Tristimulus value

g. Z Tristimulus value

h. Hunter Coordinate, L

i. Hunter Coordinate, a

j. Hunter Coordinate, b

k. Fluorescent Contribution to TAPPI Brightness

The following pages are excerpts based on a paper prepared for the American Paper Institute, which paper is dated Jan. 10, 1974 (and is understood to have been actually published some time after Jan. 23, 1974.) The paper was prepared by an author who is a joint contributor to certain improvements described but not claimed herein. The paper exists in printed form and is incorporated herein by reference. The incorporated paper is identified as "REPORT NO. 58, TO: American Paper Institute Instrumentation Program", "SUBJECT: An Analysis of On-Machine Optical Instrumentation", "DATE: Jan. 10, 1974" and is submitted by the Institute of Paper Chemistry, Appleton, Wis.

The Institute of Paper Chemistry was retained to evaluate an early conception of an on-the-paper-machine optical monitoring device for simultaneous measurement of reflected and transmitted light, and to assist in the optimum implementation of such conception. Accordingly, a substantial portion of the work reported in the following excerpts appears to inure as part of the original conception.

The following discussion is presented with respect to each of the illustrated embodiments as constituting a description bearing on the background of the invention and as clarifying and amplifying on the nature of such invention, even through the paper may also include subject matter which is based on work entirely independent of the project sponsored by the assignee of the present invention. Further, the paper will indicate the range of equivalents to each of the illustrated embodiments with respect to matters such as spectrum and geometry of illumination.

Various aspects of the on-machine measurement of the optical properties of paper for the control of opacity, standard brightness, and color have been examined. Whereas many optical property specifications are based on reflectances determined on opaque pads of paper, on-machine measurements are limited to the various optical values which can be determined on single thicknesses of a moving web. Thus, one must either control to the optical property which can be measured on-machine or strive to develop reliable correlations between the on-machine and off-machine measurements. The latter approach is more desirable. For this purpose, it is advantageous to adopt the design features of the off-machine testing apparatus to the fullest extent possible for on-machine use. Fortunately, the important factors in optical instrument design related to spectral characteristics, geometry, and photometric linearity can be translated to on-machine use with considerable exactness.

A large number of different approaches are possible in the measurement of optical properties of single sheets for purposes of control. Of these, however, distinctions can be made between single measurements, of reflectance, for example, and the measurement of two optical parameters for the same sampled area. The latter approach permits calculation of thick pad reflectivity values using appropriate theory with an essential independence of basis weight, whereas single reflectance data are functions of basis weight requiring empirical compensation. Although various possible pairs of optical measurements can be made on the same specimen area, among the most satisfactory for the use of theory are reflectance with black body backing and transmittance. The Kubelka-Munk theory, though not rigorously applicable in practice, has been shown to be rather successful in predicting thick pad reflectivity from such data in laboratory tests for white papers of reasonable homogeneity. It is less successful for deeply colored papers and for sheets of very low basis weight.

The sensitivity of spacing of white backings from paper specimens was measured with respect to the color of various commercial papers. At a central spacing of 0.32 cm, it was determined that a variation in spacing of about 0.020 inches (0.05 cm) is possible for most papers without noticeable color difference.

Optical measurements would be made on-machine on webs of varying moisture content and at elevated temperatures compared to a controlled laboratory testing environment. An experiment study of the effect of changing moisture content (R.H. range of 5 to 84%) on color showed small effects for most white papers, which were attributed to differences in surface structure. Somewhat larger effects were noted using directional illumination compared to diffuse illumination. The largest effects, for some colored papers, were attributed to changing spectral absorption characteristics of the dyestuff. Sheet moisture content effects on optical properties were judged to be within tolerable limits over a range corresponding to relative humidities between 5 and 50%.

The effect of temperature on the color of various commercial papers was studied over a range of 23.degree. to 62.degree. C. Important effects were found only for two colored papers. It was noted that significant elevations of specimen surface temperature can occur in optical apparatus employing high-intensity illumination.

Most other variables involving paper properties, machine operation, and mill environmental conditions are unlikely to be eliminated through instrument design or through the development of appropriate compensating factors. For these remaining factors, empirical correlations would be required to establish agreement between optical data or obtained on and off machine.

Specifications for the various optical properties of paper are presently based on measurements made with laboratory instruments. Considerable standardization of optical instrumentation and testing methods have been developed, but new instruments and analytical methods are often introduced to the industry as well. Changes are welcome when they provide advantages in such areas as the utility of the measurement, improved accuracy, better agreement between laboratories, and in overall testing costs. The well-established advantages of industry-wide standardization in the measurement of the optical properties, however, must always receive serious consideration.

In recent years, the control of paper quality on the paper machine has grown in importance and, for optical properties, on-line control is a very practical objective. Good control strategy requires that the properties of the sheet be determined rapidly on the moving web, but it is usually not possible to duplicate laboratory instrumentation for on-machine use. Whereas on-machine measurements are limited to such data as can be acquired on a single thickness of paper, optical properties such as brightness and color are determined on multiple thicknesses. Too often, optical instrumentation is developed for on-machine use with emphasis on the control function, but without serious consideration given to the further problem of conforming to off-machine optical property standards. The implied assumption is that a good reliable correlation will exist between the on-machine and off-machine optical measurements. Perhaps the most logical approach to the development of on-machine optical instrumentation is to design for maximum conformance with laboratory instrumentation is such factors as instrument geometry and spectral characteristics, to employ existing theory as far as possible to inter-relate single-sheet versus multiple-sheet optical measurements and to employ empirical correlations to the minimum extent required.

As on-machine optical instrumentation becomes more widely adopted, the advantages of continuously monitoring the optical properties of paper in real-time compared to the intermittent and few data which can be acquired by off-machine testing could lead to the use of specifications in the buying and selling of paper which are based on the on-machine instrumentation. Should this ever occur, it is particularly desirable that such on-machine specifications bear the highest possible degree of correlation with the off-machine specifications in current use.

In this report, many of the factors involved in the optical characterization of a moving web in a paper machine environment are discussed relative to the off-machine properties of brightness, color and opacity.

Brightness

Papermaker's brightness, sometimes called G.E. brightness and now "standard brightness" was first established in the early 1930's as the particular reflectivity (R.sub.oo) of paper determined with an instrument having a specified spectral response, specified geometry, and good photometric accuracy (1). At the same time, a system of calibration was developed whereby opal glass and paper standards are furnished periodically for each instrument.

In the early days, the scale was based on "smoked" magnesium oxide but, because of difficulty in arriving at a reproducible reflecting surface, a technique for measuring the absolute reflectance of magnesium oxide was developed (2). Thus, a total system (3) was made available so that this particular refectivity could be measured industrywide with an accuracy of about .+-.0.3 reflectivity units. TAPPI standards T217 and T452 give the detailed specifications for the measurement of standard brightness for pulp and for paper and paperboard respectively. The following specifications are involved.

Spectral Response

The effective wavelength of an instrument for the measurement of standard brightness is 457 nm. Although the effective wavelength is the most important parameter describing the spectral response, wavelength bandwidth and shape of the function also influence the result and are specified. The standardized overall spectral response of the brightness instrument which includes the spectral power distribution of the light source, the spectral transmittance of the glass lenses and filters, and the spectral response of the phototube is given in Table I.

The prescribed spectral response precludes use of a spectrophotometer employing a narrow bandwidth at 457 nm for the accurate measurement of standard brightness.

The spectral response function was chosen in the blue region of the spectrum for maximum sensitivity to changes in bleaching and the fading of paper with time. Once specified, however, the spectral response of different instruments must be maintained to close tolerances for reproducibility in measurement on an industry-wide basis.

When papers exhibit fluorescence, whether naturally or because of the addition of fluorescent dyes, the spectral power distribution of the light incident on the specimen must be specified. For standard brightness, the specified spectral power distribution of the light incident on the specimen is given in Table II. Thus, adherence to the spectral specifications of Tables I and II permits the accurate determination of standard brightness of fluorescent as well as nonfluorescent papers.

In addition, the spectral transmittance of the filters and the phototube response are selected such that the instrument has negligible response to near-infrared radiant energy whether reflected from the specimen or as a result of speciment infrared fluorescence (4). It is important to note that some colored glass filters with essentially no transmittance in the red region of the spectrum will transmit substantially in the near infrared. Geometry

The geometry employed for the measurement of brightness is illumination at 45.degree. and normal viewing with the incident and reflected beam cone half-angles specified at 11.5.degree. and 22.5.degree. respectively. The angles of illumination and viewing are critical as paper surfaces are not ideal diffusers, and the numerical values obtained are a function of the particular geometry employed. Paper surfaces also exhibit directional effects. The light reflected when the specimen is illuminated in the "machine direction" is generally less than if the specimen is illuminated in the "across-machine direction". The brightness measurement is usually performed with the specimen illuminated in the "machine direction" and on the felt or top side. A sufficient number of sheets are required to form an opaque pad.

In the more translucent papers, an appreciable penetration of light into the sample occurs. As a result of internal light scattering, the illuminated area may differ significantly from the area of reflectance or light emergence. When this condition exists, the relative dimensions of the areas illuminated and viewed, the distribution of light on the illuminated spot, the alignment of the illuminated and viewed areas and their shapes can influence the result. In the instrument employed for the brightness measurement, the viewed area and the size and position of the illuminated spot are adjusted to prescribed standards. Conformance with the standard is ensured through use of the calibration standards. Properly adjusted, the instrument can be used to measure standard brightness of strongly translucent as well as opaque material. Photometry.

The photometric accuracy of an instrument for the measurement of brightness should be better than 0.1 point on a 0-100 scale (5). The overall error introduced through discrepancies in spectral response, geometry and the basis of standardization must total less than 0.3 point.

Opacity has long been defined in the paper industry as 100 times the ratio of the diffuse reflectance of a single sheet backed with a black body to its diffuse reflectance backed by a white body having an effective absolute reflectance of 0.89. An instrument designed and built in the early 1930's and has formed the basis of a system for determining TAPPI Opacity (6) .

Spectral Response

The overall spectral response of the instrument including the spectral power distribution of the light source, spectral transmittance of the glass lenses and filter, spectral reflectivity of the integrating cavity lining and spectral response of the photocell is that the E.sub.a y function (visibility function, Illuminant A) of the CIE system (7). The effective wavelength is 572 nm and the function extends over the entire visible spectrum. The specified broad-band spectral function makes the use of narrow-band instruments inappropriate for the measurement of opacity even though the effective wavelength is proper.

Fluorescent dyes, known in industry as optical brighteners, have a rather small, if not negligible, influence on opacity as the spectral response of the instrument in the usual fluorescent region (blue) of the spectrum is quite low. Also, the fluorescent radiation from a single sheet probably would not be too different when backed by a black or white body.

The spectral reflectivity of the integrating cavity lining does influence the overall spectral response of the instrument (3) and, because it is difficult to maintain a constant lining reflectance, a system for checking and maintaining the lining reflectance is essential to good accuracy. Geometry

The geometry employed for the measurement of opacity is illumination at 20.degree. and diffuse viewing. The photodetector receives light that is both diffusely and specularly reflected from the specimen and, because there is no baffle, the photodetector also views the light directly reflected from the specimen. The ratio of diffusely to directly reflected light depends upon the level of reflectance of the integrating cavity lining (physical dimensions also are influencing factors but remain constant) and, as this ratio changes, significant changes in measured opacity can occur (8).

The illuminated area is about 10 mm in diameter with a specimen aperture of about 14.3 mm in diameter. If translucent papers or standards are to be evaluated or used, the ratio of the viewed to the illuminated area is important (8). The state of focus and alignment of the optical system particularly influences the values obtained for translucent materials.

Stray light caused by a dirty or misaligned optical system can be a source of error. The otpical system should be cleaned and aligned such that the difference in scale reading with the black body over the specimen opening when the light is blocked off before entering the cavity and with the light passing through the cavity into the black body should not be over 0.5 (0-100 scale). The reflectance of the black body should not be more than about 0.1%. The instrument scale is adjusted to read zero when the stray light is included.

Photometry

The photometric accuracy of different original instruments, employing a photocell-galvanometer system, varied from near perfect linearity to deviations as much as several points, depending upon the components. More recently, with the addition of solid state amplifiers and digital readouts, the photometric accuracy can be better than 0.1 (0-100 scale).

Calibration Standardization

In the measurement of the ratio R.sub.o /R.sub.0.89, it is necessary that the white body have an effective reflectance of 0.89. The instrument is equipped with a rotatable tube, one end of which contains a black cavity and the other the white body. A sheet of paper is placed over the specimen opening and, alternately, the black and white bodies are brought into position. The usual white body consists of a plug of appropriately surfaced magnesium carbonate within a protective glass cover. Changing the spacing between the surface of the magnesium carbonate and the specimen permits adjustment of the effective reflectance of the white body. There are two generally accepted means for arriving at the proper white body effective reflectance. One is to employ properly calibrated opal glass standards. While convenient, unless the instrument is properly adjusted with respect to translucency effects, substantial error can result. As constructed originally, the specimen supporting surface often departed from the intended plane. Thus, while the paper could follow a particular contour, the rigid glass standards will not, resulting in further error. After correcting these potential defects, it is possible to use opal glass standards and take advantage of their great convenience.

A second more basic method consists of determining R.sub.0 and R.sub.oo for a particular paper specimen on the absolute scale and, through use of the relationship sometimes known as the "balance of energy" equation (9) or the Kubekla-Munk theory (10, 11, 12), to calcuate R.sub.0.89 for that specimen. The white body can then be adjusted so that this value is obtained instrumentally. Care should be exercised so that all reflectances are obtained on an identical area of the specimen. Charts are available (13) relating the reflectances R.sub.o, R.sub.oo and R.sub.0.89 or an appropriately programmed computer can be used to calculate the R.sub.0.89 value.

Paper opacity standards calibrated for use with the opacimeter are now also available. These are convenient to use and will eliminate some of the difficulties associated with the opal glass calibration standards.

Magnesium oxide powder with an assigned absolute reflectance value is also available for use in calibrating the opacimeter for the measurement of relfectance on the absolute scale.

While the choice of spectral response for the measurement of opacity was excellent, the choice of the ratio R.sub.o /R.sub.0.89 as opposed to R.sub.o /R.sub.oo was not. Printing opacity (R.sub.o /R.sub.oo) more nearly relates to the end use of the product and would eliminate the problem of adjusting the white body (14). The fact that a single sheet is required for the measurement of TAPPI Opacity whereas an opaque pad is required for printing opacity appears to be a factor in the reluctance of the industry to change. It is more convenient to determine the opacity of the single sheet using the white body.

Spectrophotometers and filter colorimeters are the two main classes of instruments employed in the measurement of color. The spectrophotometer provides basic reflectivity information as a function of wavelength over the entire visible spectrum. The reflectivity (R.sub.oo), obtained on the thick pads of paper, with the values based on the absolute scale is basic to color measurement. The reflectivity curve contains the essential information regarding the color of the object, but considerable computation is required to derive the desired colorimetric specifications. Spectral

In the numerical specification of color, it is necessary to specify the spectral characterisitics of the illuminant and the spectral response of the observer. The CIE system (7) gives the spectral power distribution for various illuminants and the spectral response of the standard observer. Illuminant C has been used almost exclusively in the past in the specification of color, however, the use of illuminant D.sub.6500 (15) is now being considered. The specifications for Illuminant D.sub.6500 include the ultraviolet region of the spectrum. The ultraviolet region for Illuminant C was not specified.

For color definition in the CIE system, the psychophysical response of the "standard observer" to the spectral distribution of light reflected from a specimen (as provided by the spectral power distribution of the illuminant and the spectral reflectivity curve) is matched by a combination of three standard stimuli, each of appropriate power. The relative levels for the three separate stimuli are the tristimulus values which together constitute the chromaticity of a color. It is more useful to compute the fraction each stimulus has to their sum since only two of the three fractions need by specified for chromaticity definition. It then becomes possible to restate the chromaticity of the measured color, for a given illuminant, in terms of "dominant wavelength" and "purity". To complete this specification of color, the luminous reflectance of the specimen is provided directly in the CIE system by the tristimulus value "Y".

Geometry

Four illumination and viewing conditions are recommended for use in the CIE system. These include illumination at 45.degree. and viewing normal to the surface (0.degree.), normal illumination with 45.degree. viewing, diffuse illumination with normal viewing and normal illumination with diffuse viewing. Various advantages and disadvantages relate to each of these geometries from the viewpoint of best representing visual estimates of color. Generally, the geometry employed for visual inspection is more nearly 45.degree.-0.degree. or 0.degree.-45.degree.. Thus, an instrument equipped with this geometry would be expected to agree more closely with visual estimates than an instrument equipped with diffuse--normal geometry. It can be clearly demonstrated that a colorimetric evaluation using an instrument equipped with diffuse-normal geometry does not correlate closely with visual estimates for certain surfaces. Also, it is difficult to maintain a constant integrating cavity lining reflectance for long periods. The diffuse-normal geometry, however, is less sensitive to surface roughness and will give more reproducible results when specimens having an irregular surface are evaluated.

Control of the sizes, shapes and relative positions of the illuminated and viewed areas is also required for proper accounting of specimen translucency effects. Photometry

Photometric accuracy of better than 0.1 point (0-100 scale) is desired.

Filter Colorimeters

Though the spectrophotometric approach to color measurement is the most basic and rigorous, its greater cost and computational demands have led to the development of filter colorimetry. One approach involves the use of suitable lamp, filters and photodetector combinations chosen to match the spectral functions of the CIE system (x, y, z). Thus, the instrument output may be in the form of the tristimulus values of the CIE system. Although the y and z functions can be matched quite well, the double peak of the x function precludes the use of a single filter-photocell combination. Recourse is made either to the computation of the blue contribution to the x function from the z function (three-filter colorimeter) (16) or to the use of two filters with properly weighted combined output (four-filter colorimeter) for the x function. The latter gives a more accurate measure of the X tristimulus value particularly for specimens having spectral reflectivity curves with a steep slope through the blue region of the Spectrum. For color matching, particularly in control applications, the three or four-filter colorimeter may prove useful for many colors of commercial interest. However, it is subject to many limitations such as basic accuracy and the fact that colorimetric data are obtained for a single illuminant. For instance, the match may be metameric and under another illuminant there could be a serious mismatch.

Another form of colorimeter involves the use of a larger number of narrow-band filters with transmittance peaks distributed across the visible spectrum. If the filter transmittances are confined to sufficiently narrow ranges of wavelength and an adequate number are used, one may approach the utility of an abridged spectrophotometer. For many purposes of control, the abridged spectrophotometer can have important advantages over the three or four-filter colorimeter.

If a specimen exhibits fluorescence, the best spectrophotometer or filter colorimeter design utilizes illuminants with broad spectral power distribution, including appropriate intensities in the ultraviolet, with viewing through a monochromator for the spectrophotometer and through appropriate filters for the colorimeter. Thus, the fluorescent radiation will be excited in accordance with the spectral power distribution of the illuminant and the photodetector will view the reflected light and fluorescent radiation properly.

The optical information which can be acquired on the moving web of a paper machine is limited essentially to that which can be obtained using a single sheet. Reflectances can be obtained for various conditions of illumination of the single sheet and for different backings. The backing can be black body or established at various reflectance levels. Ordinarily the backing would consist of ceramic or glass placed either at a specific distance from the sheet surface or in contact with the sheet. In addition to such reflectivity measurements as can be obtained, it is often possible to obtain useful transmittance information (except for very opaque sheets). Of course, where the transmittance is very low, the reflectance of a single sheet will approach the true R.sub.oo value.

Optical specifications which properly apply to thick pad reflectances, R.sub.oo, are not readily abandoned in favor of specifications based on single sheet reflectances. Hence, the question of correlation of such on-machine data as can be obtained with actual experimentally determined R.sub.oo data is of interest. The most useful approach is to utilize to the fullest possible extent the existing theory which permits calculation of R.sub.oo from on-machine optical data. To the extent that such calculated values are not in agreement with the experimental data, empirical correlations could then be applied to bridge the remaining gap. Such an approach is more desirable than is dependence on empirical correlations alone especially if the calculated result is in close agreement with off-machine determinations.

the equations, based on the Kubelka-Munk theory, which interrelate various reflection and transmittance measurements are of principal interest in obtaining estimates of the reflectivity, R.sub.oo, from on-machine measurements. It is always necessary to obtain two different optical parameters preferably on the same areas of single sheets for the calculation of R.sub.oo using these equations. The two measurements can take many forms. For example, the reflectance of paper with black body backing (R.sub.o) along with transmittance (T) is both appropriate and experimentally desirable. It is also possible to employ any two reflectances, obtained with different backings, but this introduces problems, particularly with the backing reflectance color. Through the appropriate measurement of two optical parameters, it is also possible to characterize papers in terms of their scattering and absorption powers--not possible with single reflectance measurements

The theoretical relationship between R.sub.oo, R.sub.o and T is given in equations 1 and 2. This relationship would be applied as far as possible various desired spectral power distributions, such as are employed in standard brightness, TAPPI Opacity, and the various spectral functions associated with color measurement. ##EQU1##

Where the R.sub.oo values are determined with the appropriate filters, the tristimulus values (Illuminant C) can be calculated as shown.

TAPPI Opacity can be calculated using equations 8 and 9 where R' is equal to 0.89.

Although it has often been demonstrated that the "balance of energy" equations and the Kubelka-Munk theory are very useful in interrelating the optical properties of paper determined under many different conditions of geometry and spectral power distributions, it is important to recall that some of the conditions required by theory are not met in practice. Among these, the specimen should be illuminated and viewed with diffuse light, monochromatic light should be employed and the optical properties of the material should conform to the requirement that the absorption and scattering of light be independent of each other and occur at numerous discrete sites spaced randomly throughout the substance. All reflectance and transmittance values should be determined on the absolute basis. The fact that these conditions are seldom met requires experimental testing of the theory for each intended use.

Experimental data were acquired to test the validity of this use of the theory for a number of "white" as well as more strongly colored papers. The extent of agreement which might be expected between the calculated reflectivity, R.sub.oo, using equations 1 and 2 and experimentally determined R.sub.o and T values, and actually measured R.sub.oo values was examined for two different optical systems. Neither system would likely be used in making optical measurements on moving webs, but both serve the purpose of testing the relationships in actual use situations.

In the first set of experiments, handsheets were prepared from bleached hardwood pulp, refined to 450 ml C.S.F., at basis weights of 32, 64, 96 and 127 g/m.sup.2. The optical properties of these samples were determined using the General Electric Recording Spectrophotometer with "reversed" optics (GERS-RF). The specimen was illuminated diffusely with the spectral power distribution of a tungsten filament source modified by the integrating cavity lining. Viewing of the specimen was at 6.degree. to the normal. Four filters were interposed separately in the reflected beam to give the spectral response for the overall system of the E.sub.c x, E.sub.c y and the E.sub.c z functions of the CIE system. Two filters were utilized to obtain the E.sub.c x function. The R.sub.oo values, calculated from the measured R.sub.o and T values using equations 1 and 2 , are compared with the R.sub.oo values measured directly with the GERS-RF. The data given in Table I are averages for five different specimens at each basis weight . . .

The calculated and measured tristimulus values are in good agreement for the white and blue bond paper with rather poor ageement for the pink bond paper. The color differences based on the differences in the calculated and measured tristimulus values are given in Table IV.

Where the color differences are very large, it is probably attributable to the broad bandwidth of the spectral functions used to determine the tristimulus values and the substantial changes in reflectance with wavelength for the more highly colored papers. This can lead to error in the calculation of R.sub.oo from R.sub.o and T. Such error would likely be eliminated if a reflectivity (R.sub.oo) curve were first calculated from the curves for R.sub.o and T (appropriate number of points should be used to give an accurate R.sub.oo curve) before the integration leading to the tristimulus values is performed. But, of course, this is not the means by which data are likely to be obtained and treated in an on-machine color measurement system, at least at present. The results indicate that, for many papers, the theoretical relationships will give excellent estimates of R.sub.oo from R.sub.o T acquired for single sheets. Where the discrepancies are greater than desirable, it is probable that useful empirical relationships may be established.

The estimation of R.sub.oo from measurements of black body backed reflectance and transmittance of single sheets using theoretical relationships is subject to less error than are estimates obtained using two reflectances obtained with different backing reflectances, for example. A further approach to the design of on-machine color measuring instrumentation involves reflectances determined on single sheets backed by a body having a selected reflectance. Obviously, such reflectances will be equal to R.sub.oo for any paper only if the effective reflectance of the backing is also equal to R.sub.oo for that paper. Also, the backing will not ordinarily have the color of the paper. Hence, recourse must be made to empirical relationships between the measured reflectance values and the color of the samples as would be determined directly using opaque pads. Further, since it is often desirable to employ a spacing of some magnitude between the moving web and the backing surface, variations in the spacing which would likely occur in practice would be another source of discrepancy.

The following experiments were conducted to explore the differences in the color of paper when backed by a translucent opal glass and an opaque, enameled plaque. The spectral reflectivity curves for both backings was determined with the GERS and are given in FIG. 1. It should be noted that these reflectivity curves cannot be use to determine the effective reflectances of the backings when employed against paper and, particularly for the translucent opal glass would be at different levels if determined with different geometry. Reflectance data on the paper specimens were obtained with both GERS-RF and with the Automatic Color-Brightness Tester (ACBT). The latter employs 45.degree. illumination and normal viewing. Both instruments were equipped with appropriate filters so that the tristimulus values could be determined from four reflectance measurements (Illuminant C). Six commercial papers were evaluated with the white body backings at different spacings from the sheet . . .

Basis weight variability, of which sheet formation represents a rapidly varying form, is a matter of interest in the on-machine measurement of optical properties. All optical properties are basis weight dependent in some degree. The dependence may arise because of changes in sheet structure with basis weight or may be a consequence of the simple change in mass per unit area for constant sheet structure. Thus, whereas the reflectance of a thick pad of paper may prove to be relatively independent of basis weight, the reflectance of a single sheet with black body or other designated backing and transmittance are expected to show basis weight effects. If the basis weight is known, it is possible to apply first-approximation corrections for departures in basis weight from a target value. However, such corrections would be different for different papers, would need to be developed experimentally and would best be applied to the longer-range basis weight variations.

Rapid changes in basis weight on the scale involved in sheet formation effects will result in rapidly changing optical properties as the moving sheet is scanned by an instrument in fixed position. The true time-varying signal might well be averaged by the on-machine instrument unlike the arithmetic averaging of the same optical property values determined statically off-machine. Whether the two averages are significantly different would depend both on the nature of the time-varying signal and the time-response characteristics of the on-machine instrument.

Where two optical measurements are made simultaneously at one position on a moving web, each would be averaged instrumentally. Values of R.sub.oo calculated from such averages may differ from an average of R.sub.oo values calculated from various pairs of optical values (for example, R.sub.o and T). Though such an error would be small for small basis weight variations, it could be of importance for some papers.

If sets of data are acquired on moving webs by interposing different filters in time sequence, for example, the particular values within a set would be obtained on different areas of the web and each could relate to a slightly different basis weight. Obviously if such values are affected by basis weight, the optical property described by the set (color, for example) would be in error if the basis weight were not constant. One could in such an instance, resort to the repetitive collection of sets of values with an averaging of the art results over a longer time period. It would be desirable to avoid the collection of data such that any particular value within a set is always obtained at the same unique web position or time cycle.

Fiber Orientation

Machine-made papers usually have some degree of fiber orientation which causes a difference in reflectance if the sheet is illuminated in the "in-machine" or "across-machine" direction. Generally, the reflectance is lower when the specimen is illuminated in the "in-machine" direction. Fiber orientation is usually less pronounced on the felt side; hence, optical data are usually obtained on that side. Standard brightness is measured on the felt side and the "in-machine" direction. On-machine measurements can of course, be performed in the same way.

Polarization of light occurs to some extent when a paper surface is illuminated at an angle such as 45.degree. and the extent of polarization depends upon the kind of surface and to some degree upon fiber orientation. For this reason, the on-machine instrumentation should have the same response to polarized light as the off-machine instrument.

Two-Sidedness

Most papers have different spectral reflectivities for the felt and wire sides with the effect being more pronounced for very light basis weights and for coated papers. This affects the relationship between R.sub.o, T and R.sub.oo causing an error in the calculation of R.sub.oo. This effect is not large if the measurements of R.sub.o, T, and R.sub.oo are all made with the same side of the sheet facing the light beam on the on-machine as well as the off-machine instrument.

Moisture Content

In on-machine testing of paper, the moisture content may be at a level different from that employed in off-machine testing. Also, the intensity of the light incident on the specimen in some off-machine colorimeters is of a sufficiently high level to cause an appreciable change in temperature moisture content of the specimen during the course of performing a reflectance measurement.

Reflectance data have been obtained for "white" and dyed paper samples using the GERS and the ACBT, as these instruments employ a very low level of illumination thus minimizing departure from established laboratory environmental conditions. The GERS employs 6.degree.-diffuse geometry with the specular component partially included and the ACBT employs 45.degree.-0.degree. geometry with the specular component excluded. Using both systems, one should be able to deduce if the change in reflectance of the specimen is due to changes in absorption, scattering, or surface structure. Changes in absorption and scattering would influence the data from both instruments in about the same way whereas changes in the specimen surface would influence the data differently. Changes in absorption could be more pronounced in selected portions of the spectrum whereas changes in scattering or surface should have a minor dependence on wavelength.

In the case of the GERS, air at different levels of relative humidity was passed throught the integrating cavity. Thus, the area of the specimen measured by the instrument was exposed to the conditions air while the measurement was being performed. The same was true for the ACBT except that the air was passed through the cylindrical opening in the instrument directly beneath the specimen opening. . . .

The data show small changes for the "white" papers while the dyed papers and the newsprint show more significant changes. The effects were generally greater with the ACBT than with the GERS suggesting that changes in surface characteristics with changing relative humidity is principally involved. It is interesting to note that the reflectance of the red paper increased at 450 nm with increasing moisture content and decreased at 550 and 500 nm. This effect was noted with both instrument and is probably attributable to changes in light absorption.

Colorimetric data obtained with the ACBT at the several levels of relative humidity are given in Table XIII. The E value represents the color difference between the first determination at 5% relative humidity and the subsequent results. Several samples show a E value greater than one with sample "H" over two.

Sample A (fluorescent) has a reflectance of 85.0% for the GERS at 400 nm and 55.8 for the ACBT. This large difference is related to the erroneous evaluation of the fluorescent component by the GERS.

It appears that reflectance of paper, especially dyed papers, is significantly affected by changes in moisture content. Indications for the samples tested are that changes in moisture content resulting from exposure to levels of relative humidity from 5 to 50% represent a reasonable limiting range for good accuracy.

Temperature

The web temperature would be higher for on-machine than off-machine testing. A study was performed to determine the effects of changing temperature on the reflectance of paper. The same paper samples (different specimens) evaluated in the moisture study were evaluated at four different temperatures. The GERS and the ACBT were employed because of their low level of illumination. Temperature at the surface of the specimen in the area exposed to the incident beam was determined with a 0.004-inch diameter wire chromel-alumel thermocouple. The junction was placed in contact with the paper surface. It is understood that differences in the absorption characteristics of the thermocouple and paper preclude the assumption that the paper surface and the junction temperature are the same when exposed to the incident radiation. However, when the temperature measurements were made, paper sample B was placed over the specimen opening in every case so that the relationship between junction and paper temperature should be fairly consistent for the different instruments. . . .

A reasonable upper limit on temperature, as indicated by these data would be about 40.degree. C. If on-machine measurements are made at higher temperatures, the potential effects of temperature may need to be considered for comparison with off-machine optical data.

Fluorescence

Widespread use of fluorescent dyes has made the matter of fluorescence an important factor in the measurement of optical properties of paper. The fluorescent "whitening" agents used in the paper industry generally absorb strongly in the violet and ultraviolet regions of the spectrum and emit light at somewhat longer wavelengths in the violet and in the blue regions of the spectrum. For fluorescent dyes, in general, the region of absorption may extend from the short wavelengths (ultraviolet) to the region where light is emitted by the dye. Actually, there may be some overlapping of the absorption and emittance regions.

In the case of the fluorescent "whitening" agents, the ultraviolet light needed to excite dye is largely absorbed in the surface layers of the sheet. Thus, with fluorescence present, reflectance would be most influenced whereas transmittance would be only minimally affected. This has a pronounced effect on the calculation of R.sub.oo from R.sub.o and T.

Properly designed instrumentation should be employed where fluorescence is a factor (19).

Web Position

In all optical instruments, the position of the web must be fixed at the appropriate design point. In the calibration of an instrument with paper or other material, a web position will be indicated. The moving web should, of course, be at the calibration position. This is best accomplished by ensuring that the web is in contact with a reference surface. Through establishing such contact, it is possible to have the optical instrumentation on one side of the web properly placed with respect to web position. The other side, however, must be maintained at the proper spacing. Changes in instrument to web distance can introduce errors of significant magnitude. Two options are available; the apparatus to web spacing may be fixed, or the spacing may be measured and corrections of the results made for changes from the desired spacing. The former method is preferred whenever possible.

Web flutter is obviously undesirable. If web flutter, exists web position is not known. Similarly, vibration of the optical apparatus may influence the results.

Web Speed

Potential effects due to web speed depend on the nature of the time constants of the optical instruments. For a time varying signal, with linear photometric response of the instrument, and with slow response, an appropriate arithmetic average value might be expected. However, if the time varying signal is not symmetrical about the mean value, the instrument may not indicate the mean correctly whereas the off-machine instrument could. Thus, the reading could be speed dependent under some conditions of sheet variability and instrument design.

Calendering

All optical properties of paper are affected by calendering of the sheet. Hence, on-line measurements of final paper properties must be made after calendering. In the usual application of optical apparatus between the calender and the reel, the measurements would be obtained only a fraction of a second after the sheet leaves the calender. It seems likely that the sheet would be undergoing compression recovery during this period and for some time after calendering with the result that changes in the sheet thickness and surface smoothness would occur between the time the on-line optical measurements are made and some later time when off-machine optical measurements are made. The possible importance of such effects is not known. The fact that they may occur is recognized as one of the possible factors leading to lack of agreement between on-line and off-machine measured optical properties.

Stray Light

It is usually possible to design optical instrumentation with proper shielding from stray light. Obviously, such shielding is required, since appreciable error may occur if stray light is permitted to enter the measurement zone.

Dust and Dirt

All on-machine optical instrumentation should be designed to eliminate or minimize dust or dirt accumulations. Some contamination cannot be avoided and compensation for its effect must be developed through frequent calibration of the on-machine apparatus.

Instrument Temperature

The optical as well as electronic components of optical devices are temperature sensitive. Best design involves control of instrument temperature to values above the ambient temperature of the machine room with the web in running position. Compensation for temperature is also possible, but less desirable.

.[.The section entitled "Captions for the Figures" and FIGS. 1-7 of drawings are omitted.].

A basic conception of the present disclosure is crucially concerned with the art of paper manufacture wherein numerous grades and weights of paper are to be manufactured, and wherein access to the paper web for measurement of paper optical properties during the manufacturing process is restricted to a section between the calendering stack and the reel. The environment at this location has been detailed in the preceding section. By measuring two essentially independent optical parameters, for example measuring both the reflectance and transmittance with respect to incident light of the necessary spectral distribution, it is possible to calculate paper optical properties on the basis of existing theory with an essential independence of basis weight. The feasibility and effectiveness of this approach is confirmed in the preceding section.

Closely related to the foregoing is the conception of utilizing as nearly as practicable the optical response characteristics and geometry of existing instruments used in the paper industry, so as to achieve as close a correlation as possible with present off-line measurements of color and brightness, for example. Also of substantial significance is the conception of providing a rugged and compact temperature-stabilized instrument capable of reliable and accurate on-machine measurement of color, brightness and opacity.

The reduction to practice of these basic conceptions has included several sponsored projects at the Institute of Paper Chemistry as reflected by the preceding section and has included laboratory testing of a prototype device for accuracy and reliability on a wide range of paper samples, with careful comparison being made with corresponding measurements using standard laboratory instruments. Details of the life testing of the prototype unit over a ten-month period and the adaption of the device to reliable and stable operation on the paper machine have been included herein to document the practical implementation of the present invention. Because of the critical need for rapid calculation of paper optical properties in an on-machine device, the necessary computer programming has been developed and is fully disclosed herein. In spite of the substantial investments which have been made to secure expert assistance in implementing the conceptions, a period of time of over two years has been required to reach the stage of reduction to practice disclosed herein.

An important aspect of the disclosure relates to the measurement of the basis weight of the moving paper web concurrently with the simultaneous measurement of reflectance and transmittance values for essentially a common region of the web. Using the calculated value of infinite reflectance R.sub.oo (including the grade correction factor) and the value of transmittance T, for example, for the same sample region, along with a concurrently obtained, average value for basis weight, essentially accurate values of scattering coefficient s and the absorption coefficient k are obtained. Such coefficients will exhibit essential independence of any variations in the basis weight of the paper sheet material under these circumstances.

The measurement of both a reflectance and a transmittance value for a common sample region has an advantage over the measurement of two reflectance parameters under conditions such as found in the paper manufacturing process since the transmittance measurement is relatively insensitive to misalignment or tilting of the optical axis 515 of the backing assembly or lower sensing head 12, FIG. 3, relative to the optical axis 15 of the sensing head 11. This advantage is especially important for sheet material of relatively high opacity where two reflectance parameters would tend to be relatively close in value.

Generally the results of laboratory tests discussed herein are expected to be applicable to the on-line system. Thus the spread between values of R.sub.oo FC (See Table 3) obtained by the illustrated on-line system and the corresponding values of AR.sub.oo FC taken as standard should not differ by more than about plus or minus two points on a scale of zero to one hundred, prior to any grade correction, for a wide range of paper sheet materials of different color and basis weight.

The samples for which such accuracy was obtained in the laboratory included a range of basis weights of from 60 grams per square meter to 178 grams per square meter for white paper. Without the use of a correction factor, calculated R.sub.oo values which fell within two points of the measured value included samples of paper colored white (several tints), green, blue, canary, russett, ivory, gray and buff. Colors including pink, gold, salmon, and cherry required a significant correction factor for the X.sub.R, Y.sub.C, and Y.sub.A functions. All of the calculated R.sub.oo values involving the X.sub.B and Z functions fell within 0.77 units of the measured value on a scale of zero to 100, again without the use of any correction factor and regardless of color or basis weight.

The term quantitative measure of paper optical properties as used in the claims refers to output quantities of a numerical nature such as supplied by the on-line digital computer system 996, FIG. 6, programmed as explained herein with reference to FIGS. 7-20. Examples of such quantities are those indicated in block 990, FIG. 20; these quantities are identified with the corresponding conventional paper optical properties in Table 21.

The term on-machine optical monitoring device is intended generically and refers to the device 10, FIGS. 1 and 2, and other comparable devices for sensing two essentially independent optical response parameters such that a paper optical property is characterized prior to use of any correction factors with substantially improved accuracy in comparison to any characterization (prior to correction factors) of such paper optical property from either of such optical response parameters taken by itself. Such a monitoring device may be used as an aid to manual control of the paper making process or may be used as part of a closed loop automatic control system. Thus "monitoring" does not exclude active control in response to the output signals from the monitoring device.

Within the scope of the present subject matter, one or more of the following paper optical properties may be sensed; brightness, color, fluorescence, and/or opacity. Control of brightness and fluorescence offers a very substantial potential for cost reduction in the production of a significant range of paper types. Color control, on the other hand, may have important consequences regarding flexibility of manufacture, product uniformity, and grade change flexibility.

The value of on-line opacity control has already been demonstrated to a large degree in a prior closed loop analog opacity controller. In this installation, the average opacity across the web is controlled almost exactly at any given desired value. In previous manually controlled operations, the PKT (Pigmentary Potassium Titamate, K.sub.2 O--6T.sub.i O.sub.2 by du Pont) flow was set to some value chosen by the beater engineer and usually held to such value for the duration of the run of a given grade and weight. In the meantime, the paper opacity varied up and down, depending on process conditions at the time. Since the installation of the analog opacity controller, the opacity set point is adjusted rather than the PKT flow, thus holding opacity constant at the desired level. Instead of opacity, the PKT flow now varies up and down to compensate for other presently unavoidable process upsets resulting from variations in broke richness, PKT solids, dye usage, save-all efficiency, and other machine retention conditions. For a complete discussion of the installation of the analog opacity controller, reference is made to F. P. Lodzinski article "Experience With a Transmittance-Type On-Line Opacimeter for Monitoring and Controlling Opacity", Tappi, The Journal of The Technical Association of The Pulp and Paper Industry, Vol. 56, No. 2, February, 1973. This article of February, 1973 is incorporated herein by reference.

Existing on-line color meters have two serious disadvantages as follows:

1. Each measures a reflectance value (R.sub.g) which is decidely different from that necessary for actual color and brightness characterizations. Off-line instruments, which adequately measure these properties, require that a pad of several thicknesses (R.sub.oo) of the same paper be exposed to the light source aperture. Obviously, this is impossible with an on-line instrument, unless the far more inaccessible reel itself is tested. The use of R.sub.g instead of R.sub.oo requires very frequent off-line testing, and constant updating of an empirical calibration procedure to maintain adequate accuracy. A separate set of calibration parameters for each grade and weight is also required. Only in instances of extremely high opacity such as heavily coated, or heavily dyed colors where R.sub.g approaches R.sub.oo, can the above problems be minimized to the point where accuracy becomes sufficient for control purposes.

2. Existing color instruments are not equipped to measure transmitted light which is much more sensitive to differences and, so far, the only commercially proven method for the continuous monitoring of opacity.

To assist in indicating the scope of the present disclosure, the substance of excerpts from an early conception record with respect to the present subject matter are set forth in the following paragraphs, headed "Proposal" and "Proposed Instrument Design" having reference to the defects of existing on-line color meters just discussed:

Proposal:

An instrument built to the general specifications disclosed in the following section headed "Proposed Instrument Design" avoids the above described defects and, at the same time, provides for a concise, but extremely versatile, nearly total optical property monitor and controller. Highly trained specialists in all fields required here, including paper optics, color theory, photometry, computers, and others, if needed are available. As an example, exact specifications for the filters, photocells, and light sources are essentially ready for manufacture now. Such specialists are also aware of factors important to optical characterization frequently ignored by commercial producers of optical instruments.

Proposed Instruments Design:

An instrument made up of two scanning sensing heads, one above and one below the moving paper web, and a dedicated computer with appropriate couplers for input and output, is envisioned. The bottom head would receive light transmitted through the sheet and subsequently analyzed for its X, Y, and Z tristimulus components. It would also contain a backing of some specified effective reflectance (possibly a black body of zero, or near zero, reflectance) located just ahead or behind (machine direction) of the transmitted light receptor compartment(s).

The upper head could contain the light source, as well as a reflected light receptor. The latter occurs after reflection from the moving web at a point just above the backing, on the bottom head and would also be analyzed for its X, Y, and Z tristimulus components. Both light receivers and, for that matter, the light source itself could be integrating cavities of a type. This would be one way to insure the uniform distribution of emitted, transmitted, and reflected light in the X-direction in addition to providing identical samples of light going to each photoelectric cell installed with filters within the cavities themselves. Thermostatically controlled heaters or coolers would likely be desirable for temperature control. The flux of the light source could be monitored or controlled by a third partial, or full, set of filter-photocell combinations. The availability of both the transmitted (T) and reflected (R.sub.g) light signals described above allows for precise computation of the reflectance with an infinite backing (R.sub.oo). It is the latter, R.sub.oo value, which is required to characterize color, brightness, and an index of fluorescence. In addition, it would eliminate the need for any grade corrections in measuring either printing or TAPPI opacity, both of which could be made available if desired.

A small, rather low cost, dedicated computer with appropriate interface equipment, could be used to receive all signals, compute all pertinent optical properties, and determine the signal for direct, closed loop control of:

a. 2-5 separate conventional dye additions;

b. fluroescent dye feed to the size press; and

c. PKT, TiO.sub.2, or other slurry flow;

so that brightness, opacity, color (L, a,b) and fluorescence could be maintained almost exactly as chosen by, perhaps even a master computer, if desired.

Kubelka-Munk equations, quantitative color descriptions, and their inter-relationships, recently acquired wet end mathematical models, along with existing control theory, are all presently available in some form or other to convert the input signals from the scanning heads to optical measurements and flow feeds with which paper manufacturers are familiar. The combined mathematical technology above is also sufficient for adequate decoupling of this otherwise complicated information so that overlapped control is avoided.

Use of a dedicated computer would eliminate most of the electronics now associatd with optical measuring equipment. It could also be used to integrate results across the web and simplify and/or maintain calibration. The package would lend itself to rather universal application and minimize the time and effort on the part of the purchaser.

The key feature of this proposed instrument, which distinguishes it from existing on-line optical testers, is that it calls for the measurement of both transmitted and reflected light without undue complications. This, in turn, can cause a great deal of improvements regarding sensitivity, accuracy, flexibility, and thoroughness of a continuous optical property measuring device.

The following Table will serve to identify the computer symbols used in FIGS. 17-20 with the corresponding conventional symbols and terminology used in the text.

Given the foregoing conception, it is considered that many modifications and variations will be apparent to those skilled in the art. The basic conception claimed herein is the sensing of two essentially independent optical response parameters of a single thickness relatively homogeneous sheet material such that any other desired parameter or paper optical property can be accurately calculated.

For the case of opacity measurement, for example, the present invention is particularly applicable to an optical system wherein system spectral response essentially simulates the C.I.E. tristimulus Y filter with either illuminant A or C and to near-white papers as explained in the Lodzinski Article of February, 1973 incorporated herein by reference.

1. Light source lamp terminals were connected to test tip jacks so that lamp voltage (at the lamp) can now be quickly measured without opening the case.

2. An easily removable doorway was cut from the top of the upper head case so that the photocell and amplifier gain circuits are much more accessible now. The photocell can now be easily removed and its 3/16" diameter aperture viewed directly from above without removing the case. This permits a quick check to see whether the two heads are properly aligned. The diffuser 276, FIG. 3, in the photocell aperture will be uniformly lighted when alignment is correct. Non-uniform illumination of this diffuser is quite apparent when the heads are improperly aligned.

3. A temperature sensitive resistor is located in the upper head adjacent to the photocell. Conductors are connected to lugs on the power supply panel so that such resistance measurements are easy to acquire. An empirically prepared chart is used to convert the resistance to temperature. Thus, an upper head temperature can be monitored from the remote power supply panel. (This temperature measuring device has been on the OMOD since it was first constructed.)

4. The upper head weighs 20 lbs. and the lower head 93/4 lbs. The weight of the mounting brackets are 7 and 5 lbs., respectively. (The compact size and light weight of these heads is an important advantage when it comes to providing means of installing and traversing across the web).

5. The reason we choose the 45.degree.-0.degree. geometry is because this geometry is used in the standard TAPPI brightness measurement. There is no standard TAPPI color test geometry at this time. It is considered that the 45.degree.-0.degree. geometry with the light plane in the machine direction is the proper geometry for color measurement as well. The reason for this is that most of our paper products where brightness and color are important are eventually used for written communications purposes. Consequently, they are viewed on a table top or desk with the human eye and light source approximating the 45.degree.-0.degree. geometry as employed in the OMOD. Moreover, the grain direction of the paper (grain long 81/2.times.11 letterhead for example) is such that this directional effect is also simulated by the OMOD. Diffuse viewing is impossible by the human eye and diffuse illumination is quite unlikely in most offices or places of paper use.

1. We plan to test each individual reading of transmittance and reflectance (T and R) and compare it to the previously smoothed values. The latest individual readings will not be used to update the smoothed average whenever a difference between the two is greater than X%. The value of X will remain flexible, but likely in the neighborhood of 5-10%. This subprogram will reject and flag bad data since the paper optical properties could hardly change faster than this between readings. An exception is the very beginning of a run; however, the heads are not put on sheet until the operation has settled down somewhat anyway. Only the startup of a run will need to be manual or feed forward as far as color, brightness, and opacity control is concerned.

2. Initialization of the smoothing algorithm of R and T will be made to occur only when a grade change occurs; i.e., whenever a new set of specific grade correction factors are entered into the computer memory. There should not be any need for re-initialization for any other reason. Even basis weight changes occur gradually enough to permit the use of the previously stored smoothed averages without serious difficulty. We may, however, consider the use of an operator command to re-initialize such algorithms if found desirable.

3. For the No. 6 paper machine (shown in FIGS. 1 and 2), the OMOD heads will be pushed completely off the web on the front side to allow the basis weight gauge (mounted side of the OMOD) to measure right up to the front edge. The program will, therefore, need to be modified to reject data acquired whenever this occurs. Since the head position will be known, from the basic weight profile monitoring system, such data can be left unused whenever the position "Y" or greater is reached.

It turns out that this particular situation provides a convenient means of servicing the OMOD. The traversing mechanism can be stopped when the OMOD heads are pushed beyond the web edge where they are quite accessible for examination, checking of standardization, etc.

The following program listings are considered to be in conformity with the flow charts of FIGS. 8-20. The listing is provided by the General Electric PAC 4020 process control computer, and the following general discussion explains the GE-PAC 4020 program listing characteristics.

I. The first page contains the following unique information:

A. Line 1--control statement, with time of day and activity.

B. Lines 2 and 3 contain program identification numbers.

II. The remaining information is broken into three parts as follows:

A. The three left-most columns of numbers consist of assembler-generated machine coding:

1. The first column of numbers consists of the octal location, relative to the beginning of the program.

2. The second column of numbers consists of the contents of the octal location in absolute form--the first two numbers signify the operation code; the third number signifies the index register, if any; and the remaining five numbers signify the absolute operand of the instruction.

3. The third column of numbers consists of the contents of the octal location in relative form--it is identical to (2), except the five right-most numbers signify the relative operand from the location of the instruction.

B. The next eighty columns correspond to the symbolic instructions from which the assembler generated the machine coding.

1. Statements beginning with an asterick or "C" are comments only, with respectively a 6 or 7 in column seventy.

2. Other statements are divided as follows: label, if any; symbolic operation code; symbolic operand, if any; index register preceded by a comma, if any; comment, if any; and 6, 7 or 1 in column seventy.

3. The 6 in column seventy indicates a programmer-defined assembly language statement. The 7 in column seventy indicates a programmer-defined Fortran language statement. The 1 in column seventy indicates an assembler-defined assembly language statement, such as symbolic coding generated from a programmer-defined Fortran statement.

C. The right-most column contains the line sequence number of the printout.

The following listing is presented to illustrate the extent of the programming effort to implement the flow charts of FIGS. 8-16. It will be observed that the implementation of these flow charts together with the changes previously indicated herein and any necessary debugging is within the routine skill of the art. ##SPC1## ##SPC2##

The following listing is presented to illustrate the extent of the programming effort to implement the flow charts of FIGS. 17-20. It will be observed that the implementation of these flow charts together with the changes previously indicated herein and any necessary debugging is within the routine skill of the art. ##SPC3## ##SPC4## ##SPC5##

While the on-line automatic control of paper optical properties is an ultimate objective of the work reported herein, the claimed subject matter relates to on-machine monitoring of paper optical properties whether used as an aid to conventional manual control or for other purposes. Nevertheless, in order to provide a disclosure of the best mode presently contemplated for automatic control as a separate but related area of endeavor, the following discussion is presented.

The optical properties of a sheet of paper are dependent upon all of the materials of which it is made but primarily upon the furnished pulp, fillers, pigments, dyes, and some additives. If is often very difficult to maintain the optical attributes of the pulp, fillers and additives constant within a given production run. Such variation is even greater between runs. The optical properties of the finished paper may, however, be reasonably controlled to specified standards by varying the additions of dyes and fillers and pigments until the desired compensations are achieved. The problem is that each furnished ingredient affects each of the resulting paper optical properties in a rather complicated manner. Indeed the intuition of experienced papermakers has essentially been the sole method of optical property control. Unfortunately, this approach is inefficient, resulting in consideable off-standard paper and/or waste of costly materials. Accordingly, a dire need exists within the paper industry for a highly reliable and continuous optical property monitor coupled with a closed loop computer control system.

The value of such closed loop control, based on a feedback color detector, has already been demonstrated for the continuous addition of two and sometimes three dyes. (1) (2) Target dye concentrations changes of up to three dyes can be determined by solving three simultaneous equations containing three unknowns. (1) One disadvantage of such control is that accurate color monitoring is not presently available unless large and frequent empirically determined correction factors are applied to the original output results. A second disadvantage arises when opacity and the fluorescence must also be simultaneously controlled. In this case the number of independently controlled continuous additions increases from three to five. An optical brightener and an opacifying pigment constitute the two additional factors.

An object of this invention is to demonstrate a method by which fluorescence can be continuously monitored. A means by which the optical brightener addition can be separately and independently controlled is inherently implied. The paper color is also analyzed without the fluorescent contribution. It is, of course, this latter characterization (without fluorescence) which should be, but which has not in the past been, used to determine the required addition of the conventional dyes. In other words, the effect of the optical brightener is decoupled from the three conventional dyes making possible the simultaneous control of all four dyes.

Another portion of this invention demonstrates a means of continuously determining the scattering coefficient of the moving web for each of the six available light spectrums. It is possible to determine the scattering coefficient required to achieve a given opacity specification whenever the basis weight and absorption coefficient are known. When the latter are set equal to a given set of product specifications, then the calculated scattering coefficient becomes the target scattering coefficient. (The absorption coefficient can be acquired by off-line testing of a sample of the standard color to be matched. In reality, this becomes a target absorption coefficient as well.) The dyes have little, if any, effect on the scattering coefficient but the effect of the slurry pigment is very large. Thus the target scattering coefficient is used as the sole feedback variable to control the slurry pigment feed. This will insure that the opacity is at or near the specification as long, as the absorption coefficient and basis weight are also on target. The absorption coefficient should, of course, be on target by virtue of the independent color control. A completely independent system controls the basis weight.

A method by which the decoupling of three conventional dyes, one optical brightener and one opacifying pigment has hereby been explained. Heretofore, such decoupling as revealed in the prior art has been limited to three absorptive dyes and thereby neglecting the need to also achieve a specified degree of fluorescence and opacity.

The present invention is for the purpose of obtaining a quantitative measure of an optical property such as brightness, color, opacity or fluorescence of single thickness sheet material.

The sheet material is substantially homogeneous in its thickness dimension such that the optical property of interest can be reliably calculated from reflectance and transmittance measurements on the basis of existing theory. Thus the present invention is not applicable to the sensing of localized surface effects (such as due to surface migration of light absorbing powder particles, for example). To the contrary the present invention is concerned with the average or bulk optical characteristics of the sheet material considered as a whole, and especially is concerned with the characteristics of paper sheet material as it is delivered from a paper machine after completion of the paper manufacturing process.

The present invention in its broader aspect does not require strictly homogeneous material since empirical correction factors can be applied for cases where theory is less effective. For example, the paper optical properties of calendered and coated papers may be effectively measured by the system of the present invention using grade correction factors to correlate on-machine results with the measurements obtained by standard off-line instruments.

The optical system of the monitoring device includes components such as those shown in FIG. 3 which define or optically affect the incident, reflected and transmitted light paths such as indicated at 133, 137 and 141-143 in FIG. 3. For the case of a filter wheel as indicated in FIG. 4, each filter wheel position may be considered to define a separate light energy path with its own predetermined spectral response characteristics.

In each filter wheel position, there are two distinct light energy paths for measuring a reflectance value and a transmittance value, respectively. In the illustrated embodiment each such light energy path includes a common incident light path 133, but the paths diverge, one coinciding with the reflectance sensing light path 137 and the other including the transmittance sensing light path. The photometric sensors 203 and 260 thus provide simultaneous reflectance and transmittance output signals with respect to essentially a common region of the web. The reflectance sensing light path collects light from a circular region with a diameter of about 3/16 inch, and the transmittance sensing light path collects light from a total elliptical region which includes substantially the same circular region as mentioned above. Because of sheet formation effects and other localized variations in web characteristics it is considered valuable that the reflectance and transmittance output signals are based on readings from essentially a common region of the web.

By taking at least one reading in each traverse of the web, and taking such readings at different points along the width in successive traverses, it is considered that accuracies equal to or superior to those of an off-line sampling of a finished reel can be achieved, while at the same time the readings are available immediately instead of after completion of a manufacturing run.

By way of example, in the illustrated system a traversal of the web by the sensing head takes about forty-five seconds, so that the sensing head operates at a rate of at least one traversal of the width of the web per minute in the time intervals between the hourly off-sheet standardizing operations.

In accordance with the teachings of an improvement to the present invention, the optical window 135 is itself selected as to its optical characteristics so as to provide the basis for off-sheet standardization. To this end it is advantageous that the optical window exhibit an absolute reflectance value as measured by the standard automatic color-brightness tester of at least about thirty-five percent (35%). The corresponding absolute transmittance value as measured on the G.E. Recording Spectrophotometer with conventional optics is about fifty-six percent (56%). With the illustrated embodiment, once the system is properly adjusted with respect to the zero reflectance readings (as by the use of a black sheet of known minimal reflectance) the system maintains such zero adjustment quite stably; accordingly the higher the reflectance value of the optical window, the more effective is the reflectance standardization by means of the optical window. On the other hand a transmittance value which is of a reasonable magnitude is also desirable, so that the provision of an optical window with substantial values of absolute reflectance and transmittance is advantageous.

With the illustrated embodiment, the transmittance readings for the moving web are relatively more nearly independent of misalignment of the upper and lower sensing heads than the reflectance readings. Further it is considered that tilting of the lower sensing head relative to the optical axis of the upper sensing head has less effect on transmittance readings than on reflectance readings. Thus it is considered that it would be advantageous to have an optical window such as 135 with an absolute reflectance value of seventy percent (70%) or more. A value of reflectance as high as ninety percent (90%) would not be unreasonable and would generally still permit a transmittance value of a subtantial magnitude to give reasonably comparable accuracy of reflectance and transmittance readings for on-line operation as herein described.

While separate photometric sensing means for the reflectance and transmittance readings have been shown, it is possible with the use of fiber optics, for example, to use a common photometric sensor and alternately supply light energy from the reflectance and transmittance light paths thereto, providing the response time of the sensor enables reflectance and transmittance readings to be obtained for essentially the same region of the moving web. Generally the possibility of such time multiplexing of reflectance and transmittance readings will depend on the speed of movement of the web and the degree of uniformity of sheet formation and the like.

It is very desirable that the system of the present invention be applicable to sheet materials having a wide range of characteristics such as basis weight and sheet formation, and operable at high speeds of movement such as 100 to 3000 feet per minute. Further, for maximum accuracy, it is necessary that a region of the sheet material being sampled have substantially uniform opacity. Accordingly, especially for sheet material of relatively low basis weight and relatively poor sheet formation, greater accuracy can be expected when the response of the photometric sensor is relatively fast, and when reflectance and transmittance readings are taken simultaneously and are a measure of the characteristics of a common sampling region of minimum area (consistent with adequate signal to noise ratios). Thus multiplexing of reflectance and transmittance readings is not preferred for the case of high speed paper machinery and comparable environments, nor is it desirable to use reflectance and transmittance light paths which intersect the web at spacially offset regions.

With respect to speed of response of the photometric sensing means substantial improvements over the previously described components are deemed presently available. If the spectral response and other necessary characteristics are suitable, a sensor with such a higher speed of response is preferred for the illustrated embodiment. Good experience has been had with a silicon photocell presently considered as having an appropriate spectral response characteristics for color and other measurements in accordance with the present invention. The specific silicon cell referred to is identified as a Schottky Planar Diffuse Silicon Pin 10 DP photodiode of a standard series supplied by United Detector Technology Incorporated, Santa Monica, Ca.

In place of a rotatable filter wheel arrangement as shown in FIGS. 3 and 4, a set of twelve fiber optic light paths may define six simultaneously operative reflectance light paths in upper sensing head 11 and six simultaneously operative transmittance light paths in lower sensing head 12. The six reflectance fiber optic paths would include respective filters corresponding to filters 281-286 and respective individual photocells and would be located to receive respective portions of the reflected light which is reflected generally along path 137 in FIG. 3. The six transmittance fiber optic paths would also include respective filters corresponding to filters 281-286 and respective individual photocells, and would be located to receive respective portions of the transmitted light which is transmitted generally along paths such as 141-143 in FIG. 3. The filter means in the incident light path such as indicated at 133 in FIG. 3 might include a filter in series with filters 271 and 272 for filtering out the ultraviolet component from the incident beam, so that the twelve simultaneous photocell readings corresponding to those designated RSD1 through RSD6, and TSD1 through TSD6 (when the device is off-sheet), and corresponding to those designated RSP1 through RSP6, and TSP1 through TSP6 (when the device is on-sheet) will exclude a fluorescent contribution. (See Table 3 where this notation is introduced.)

If a reflectance reading corresponding to RSD7 (when the device is off-sheet) and corresponding to RSP7 (when the device is on-sheet) is desired so as to enable computation of fluorescent contribution to brightness, it would be necessary to mechanically remove the ultraviolet filter from the incident light path, or otherwise introduce an ultraviolet component of proper magnitude, and obtain another brightness (Z) reading, for example from the number four reflectance photocell.

As an alternative to the above fiber optic system with a common incident light path, seven fiber optical tubes incorporating filters corresponding to 281-287 of FIGS. 3 and 4, respectively, at say the light exit points of the tubes, could be used to supply the incident light to seven different points on the paper web. The reflected light from each of these seven points could be monitored by seven different systems, each involving lenses and a photocell, and the number seven reflected light path including also a filter corresponding to filter 288, FIG. 4. The transmitted light from the first six points would also need to be kept separately, and this could be accomplished by six integrating cavities and six photocells.

As a further alternative the seven fiber optical tubes defining the seven incident light paths could have a second set of seven fiber optical tubes and photocells respectively disposed to receive reflected light from the respective illuminated points. Another set of six fiber optical tubes and photocells could be associated with the first six illuminated points for receiving transmitted light. This could eliminate the need for the light collecting lenses in the upper sensing head and the integrating cavities in the lower sensing head.

The last two mentioned alternatives with seven fiber optical tubes defining the incident light paths appear to be rather complicated systems, but they do offer means of eliminating both the mechanical filter wheel as well as any mechanical device to control the presence of ultraviolet light in the incident beam.

Still another alternative is to use "screens" in addition to the filters in the embodiment of FIGS. 1-5. The new photodiodes are considered sensitive enough to measure reduced light intensites so that screens with different transmittance values could be used with six of the incident beam filters so that the net photocell output for each reflectance light path, and for each transmittance light path, would be similar enough so that separate and individual pre-amplification for the respective reflectance outputs would not be necessary, and so that separate and individual pre-amplification for each transmittance output would not be necessary. This means that reed switches 341-347 and 351-357, and relays K.sub.1 through K.sub.7 in FIG. 6 could be eliminated, and that the feedback paths for amplifiers 361 and 429 could have the same resistance value in each filter wheel position. A means of sensing filter wheel position would still be necessary, but this could be done in a number of simple ways, one of which would be a single reed switch such as reed switch 358 shown in FIG. 6. The number of necessary conductors in the cables 51 and 52, FIG. 5, would, of course, be reduced in this modification.

The term "screen" is understood in the art as referring to a network of completely opaque regions and intervening openings or completely translucent regions, such that light energy is uniformly attenuated over the entire spectrum by an amount dependent on the proportion of opaque to transmitting area.

The device of FIGS. 1 and 2 has been tested on a machine operating at about 1000 feet per minute, and no problems have appeared in maintaining the necessary uniform and stable contact geometry between the head and the moving web.

It will be apparent that many further modifications and variations may be effected without departing from the scope of the novel concepts of the present invention.

An instrument 1010 is shown in FIG. 21 made up of two scanning sensing heads 1011 and 1012, one above and one below the moving paper web 14, and a dedicated computer with appropriate couplers for input and output forms part of digital computer system 1996, FIG. 22. The bottom head 1012 receives light transmitted through the sheet and subsequently analyzed for its X, Y, and Z tristimulus components. It also contains a backing 1135 of some specified effective reflectance (for example a black body of zero, or near zero, reflectance) located just ahead or behind (machine direction) relative to entrance 1154 of the transmitted light receptor compartment indicated at 1145-T, FIG. 22.

The upper head 1011 contains light source means 1201 as well as a reflected light receptor indicated at 1145-R, FIG. 22. The latter receives light energy after reflection from the moving web at a point just above the backing 1135, on the bottom head 1012 and the reflected light energy would also be analyzed for its X, Y, and Z tristimulus components. Both light receivers 1145-R and 1145-T and, for that matter, the light source means 1201 itself may be integrating cavities of a type. This would be one way to insure the uniform distribution of emitted, transmitted, and reflected light in the X-direction in addition to providing identical samples of light going to each photoelectric cell installed with filters within the cavities themselves, this arrangement being represented in FIG. 22. The flux of the light source means 1201 could be monitored or controlled by a third partial, or full, set of filter-photocell combinations. The availability of both the transmitted (T) and reflected (R.sub.g) light signals described above and as indicated in FIGS. 21 and 22 allows for precise computation of the reflectance with an infinite backing (R.sub.oo). It is the latter, R.sub.oo value, which is required to characterize color, brightness, and an index of fluorescence. In addition, it would eliminate the need for any grade corrections in measuring either printing or TAPPI opacity, both of which could be made available if desired.

A small, rather low-cost, dedicated computer such as that forming part of system 1996 with appropriate interface equipment such as indicated at 1997 could be used to receive all signals, compute all pertinent optical properties, and determine the signal for direct, closed loop control of:

a. 2-5 separate conventional dye additions;

b. fluorescent dye feed to the size press; and

c. PKT, TiO.sub.2, or other slurry flow;

so that brightness, opacity, color (L, a,b) and fluorescene could be maintained almost exactly as chosen by, perhaps even a master computer, if desired.

Kubelka-Munk equations, quantitative color descriptions, and their inter-relationships, recently acquired wet end mathematical models, along with existing control theory, are all presently available in some form or other to convert the input signals from the scanning heads 1011 and 1012 to optical measurements and flow feeds with which paper manufacturers are familiar. The combined mathematical technology above is also sufficient for adequate decoupling of this otherwise complicated information so that overlapped control is avoided.

Use of a dedicated computer would eliminate most of the electronics now associated with optical measuring equipment. It could also be used to integrate results across the web and simplify and/or maintain calibration. The package would lend itself to rather universal application and minimize the time and effort on the part of the purchaser.

The key feature of the instrument 1010 which distinguishes it from existing on-line optical testers, is that it provides for the measurement of both transmitted and reflected light without undue complications. This, in turn, can cause a great deal of improvements regarding sensitivity, accuracy, flexibility, and thoroughness of a continuous optical property measuring device.

FIGS. 23 and 24 show further means by which the incorporated patent applications can be employed without the use of a mechanically operated filter wheel.

In FIG. 23, an ultra violet light absorbing filter is located at 2271, and the U.V. filtered incident light beam 2133 is directed to impinge upon the moving web 14.

Transmitted light is collected within the integrating cavity 2145-T below the web, just as in the embodiment of FIGS. 1-6. The resulting intensity and color of this light is analyzed in a manner comparable to such embodiment by employing, for example, six separate ports containing six different filter-photocell combinations including filters such as 2281-T, 2282-T and 2283-T (and as shown in FIG. 22) representing Brightness, X.sub.R, X.sub.B, Z, Y.sub.C, and Y.sub.A, respectively. The output of each photocell could be read continuously or intermittently by means of a computer equipped with appropriate interface electronics as in the previous embodiments.

The reflected light component is directed through an appropriate lens system 2273 into a second integrating cavity 2145-R located above the moving web. Its intensity and color is, likewise, analyzed by the use of six filter-photocell combinations including filters such as 2281-R, 2282-R and 2283-R (and as shown in FIG. 22) and connected eventually to the same computer.

FIG. 24 is similar to FIG. 23 except that fiber optics (light conveying tubes) are utilized in place of the integrating cavities. A representative portion of the transmitted light is directed into each of six such tubes (five of the tubes being indicated at 3001-3005 in FIG. 24) which are ended by means of specific spectral response filters (Br, X.sub.R, X.sub.B, Z, Y.sub.C, Y.sub.A,) and corresponding photocells, five of the filter-photocell combinations being indicated at 3011-3015. The same is done for the reflected light by means of filter-photocell combinations five being indicated at 3021-3025. All twelve outputs of this example are again read either continuously or intermittently by a computer.

The fluorescent component of light resulting from the use of "optical brightners" is not included in any of the above measurements. It can be measured, of course, by mechanically removing the U.V. incident beam filter and synchronizing such change with a status change to the computer. A short time later the U.V. filter is reinstated back into position. A rotating U.V. filter-chopper could be employed at 3271 providing proper synchronization of the computer data storage is also accomplished.

The set of twelve fiber optic light paths as shown in FIG. 24 may define six simultaneously operative reflectance light paths in upper sensing head 3011 and six simultaneously operative transmittance light paths in lower sensing head 3012. The six reflectance fiber optic paths include respective filters corresponding to filters 281-286 and respective individual photocells located to receive respective portions of the reflected light which is reflected generally along a path such as that indicated at 137 in FIG. 3. The six transmittance fiber optic paths would also include respective filters corresponding to filters 281-286 and respective individual photocells located to receive respective portions of the transmitted light which is transmitted generally along paths corresponding to paths 141-143 in FIG. 3. The filter means in the incident light path includes a filter 3271 similar to filters 271 and 272 for filtering out the ultraviolet component from the incident beam, so that the twelve simultaneous photocell readings corresponding to those designated RSD1 through RSD6, and TSD1 through TSD6 (when the device is off-sheet), and corresponding to those designated RSP1 through RSP6, and TSP1 through TSP6 (when the device is on-sheet) will exclude a fluorescent contribution. (See Table 3 where this notation is introduced.).

If a reflectance reading corresponding to RSD7 (when the device is off-sheet) and corresponding to RSP7 (when the device is on-sheet) is desired so as to enable computation of fluorescent contribution to brightness, it is necessary to mechanically remove the ultraviolet filter 3271 from the incident light path, or otherwise introduce an ultraviolet component of proper magnitude, and obtain another brightness (Z) reading, for example from the number four reflectance photocell.

As an alternative to the above filter optic system with a common incident light path as shown in FIG. 24, seven fiber optical tubes incorporating filters corresponding to 281-287 of FIGS. 3 and 4, respectively, at say the light exit points of the tubes, could be used to supply the incident light to seven different points on the paper web. The reflected light from each of these seven points could be monitored by seven different systems, each involving lenses and a photocell such as shown in FIG. 24, and the number seven reflected light path including also a filter corresponding to filter 288, FIG. 4. The transmitted light from the first six points would also need to be kept separately, and this could be accomplished by six integrating cavities and six photocells.

As a further alternative the seven fiber optical tubes defining the seven incident light paths could have a second set of seven filter optical tubes and photocells respectively disposed to receive reflected light from the respective illuminated points as in FIG. 24. Another set of six fiber optical tubes and photocells could be associated with the first six illuminated points for receiving transmitted light as in FIG. 24. This could eliminate the need for the light collecting lenses in the upper sensing head and the integrating cavities in the lower sensing head.

The last two mentioned alternatives with seven fiber optical tubes defining the incident light paths appear to be rather complicated systems, but they do offer means of eliminating both the mechanical filter wheel as well as any mechanical device to control the presence of ultraviolet light in the incident beam.

Still another alternative is to use "screens" in addition to the filters in the embodiments of FIGS. 21-24. The new photodiodes are considered sensitive enough to measure reduced light intensities so that screens with different transmittance values could be used with six of the incident beam filters or with the reflectance and transmittance filters so that the net photocell output for each reflectance light path, and for each transmittance light path, would be similar enough so that separate and individual pre-amplification for the respective reflectance outputs would not be necessary, and so that separate and individual preamplification for each transmittance output would not be necessary. This means that the feedback paths for the twelve amplifiers of components 1361-R and 1361-T in FIG. 22, for example, could have the same resistance values.

The term "screen" is understood in the art as referring to a network of completely opaque regions and intervening openings or completely translucent regions, such that light energy is uniformly attenuated over the entire spectrum by an amount dependent on the proportion of opaque to transmitting area.