‘Connections between the cathode followers 21 and as sociated ampli?ers may take the form of shielded cable members 22 and the ampli?ers may be positioned any convenient distance from the photoelectric means. The shielded cables carry the cathode follower signal output lead as well as leads for the D.C. supply to the cathode follower and the photoelectric cells. As previously men tioned, the photoelectric pickup means herein consists of a row of photoelectric cells which Will reliably detect all blemishes present on paper sheets which mave substan tially uniform re?ection characteristics. ' Upon investigation of various types of high grade bond paper, it has been shown that if such paper is scanned by the illumination of an elementary area 0.25 inch x 0.25 inch equals 1,66 of a square inch, the percentage modu-' lation of the re?ected, diffused light can be as high as three parts in one-hundred. When the illuminated ele mentary area is reduced, the modulation due to the un blemished paper texture usually increases and the light intensity is reduced. Thus, for purposes of the following consideration, it is reasonable to assume a practical ele mentary spot size of 1/16 of a square inch area. When it is required to detect a small black spot of say ‘1/32 inch x 1/32 inch dimensions, that represents an area of approximately 1AM) of a square inch. Therefore, the percentage modulation, that is the defect signal, obtained when a blemish of 1/1000 of a square inch area is‘scanned _by an illuminated elementary area of 1/16 of a square inch is 1.6 parts‘ of one~hundred so that the defect signal is below the noise level of the paper (three percent) and can not be detected. For reliable detection, the defect signal should be at least twice the value of the noise signal. Further investigations on a wide selection of bond type ‘papers ihave shown that when a narrow section, say 1&2 of an inch wide by 1.6 inches long is illuminated, the noise modulation is a maximum of one part in one-hundred or less. When the length of such section of paper is in creased, the maximum noise modulation is reduced to less than one part in one-hundred. The reduction of noise modulation is due to the averaging out of the modu lation caused by the small size of the distinct areas of the texture having different shades or di?erent re?ective prop erties. When a black spot having a 0.001 square inch area en ters an inspection section of 1X42 inch by 1.6 inches, which equals 0.05 square inch area, the. resulting percentage modulation is 0.001/ 0.05, that is to say, two parts in one hundred.
Therefore, such blemish is detected over the one percent paper background noise modulation. When the outputs of several photoelectric cells are con nected to the control grid of a cathode or anode follower stage, the modulation due to the paper background noise is reduced while the modulation due to the blemishes is added. For example, it is practicable to connect ten photoelectric cells of the IP39 type to a common anode follower. In what follows, it is shown that the inherent noise of the photoelectric cells due to the current passing from the photocathodes to the anodes caused by the re?ected il lumination of the unblemished paper and collected by the photocathodes is by order of magnitudes smaller than the paper background noise. Inherent noise in high vacuum photoelectric cells, such as for example in type IP39 cells used for investigation, is caused mainly by the “shot” ef feet which is the generic term given to current ?uctuations in a beam of electrons arising from the randomness of emission of the photocathode. The mean square ?uctu ation of the photoelectric current diz=leld? where di=instantaneous ?uctuation of current from its mean value I, e=electron charge=l.6 10~19 coulombs, df=fre quency interval in which E2 is observed. Therefore, the mean noise voltage %=\/2eIdf-R, where R=the load resistance of the photoelectric cell. A 1/32 inch by 1.6 inch illuminated aperture gave, in each photoelectric cell, a current 1:3.3-10-5 amperes. The load resistance R was 1 megohm. The frequency re sponse range of the photoelectric cells, anode follower and ampli?er system d)‘=3-103 c./s. for a paper velocity of two-hundred inches per second. Therefore, in ten cells 'd_12=\/2- 1.6- 10-19-33 -lO-8-3- 103- 106\/T0=18 microvolts. Measurements have con?rmed such theoreti cal noise levels. The highest paper background noise peak voltage v corresponding to a modulation of one per cent of I is v=1-l0-2-R=33O microvoltsy This is the paper background noise voltage pickup by one cell only, so that the inherent noise of ten type 1P39 photoelectric cells can be neglected against the basic paper noise. In FIG. 3, the plan view illustrates (the housing 9 with aperture 7 extending lengthwise from end to end thereof and the relative position of the multiple photoelectric cells 6 with respect to the aperture 7. It will also be observed that the housing 9 extends the entire width of the sheet paper 2 so that in operation the entire Width of the sheet 2 is subject to inspection. As illustrated in FIG. 4, photoelectric cells 23, 24, 25 and 26 ‘are paralleled by connecting their cathodes to the grid 34 of triode 32. For purposes of simplicity only four photoelectric cells are shown. The number of photo electric cells thus paralleled may be ten or more ‘as long as the inherent noise of such a group is below the paper background ‘noise and as long as the paper background noise is half or less than half of the signal produced by the smallest defect to be detected. Additionally, instead of the triode 32, a pentode or other type ‘of amplifying devices may be used. The anode ‘follower stage consists of tube 32, anode-grid feed-back resistor 29, grid-leak re sistor 30, cathode bias resistor 31, coupling capacitor 28 and anode resistor .27, and is the preferred form of cou pling, but could be replaced by a cathode follower stage which is well known in the art. When the paper under inspection has a uniform tex ture, without blemishes, the amount of light re?ected from the illuminated paper sheet falling upon the photocathodes of cells 23, 24, 25 and 26 is constant and the current in the photoelectric cells 6 is also constant. When however a paper defect such as a dark area is viewed by one or 10 15 20 25 30 40 45 55 60 65 70 75 8 more photoca-thodes, the steady current of the cells is less. Without teed-back resistor 29 such current decrease would produce a voltage decrease at the grid 34 as de?ned by the current change and the value of resistor 30. Due to the ampli?cation of tube 32 and feed-back impedance 28 and 29, substantially all the change of the photoelectric cell currents is ?owing to the anode 33 and the potential of grid 34 remains substantially constant so that the e?ec rtive load resistance of cells 23, 24, 25 and 26 is resis tor 29. 光电检测与分选机器英文文献和中文翻译(4):http://www.youerw.com/fanyi/lunwen_35843.html