SILICON BUTTING CONTACT IMAGE SENSOR CHIP WITH LINE TRANSFER AND PIXEL READOUT (LTPR) STRUCTURE
FIELD OF THE INVENTION
The invention relates generally to the field of contact type image sensing systems for scanning and digitizing documents. More particularly, the invention relates to a contact image sensor chip with a line transfer and pixel readout (LTPR) structure.
BACKGROUND OF THE INVENTION
The technology required to scan and digitize documents has existed for approximately twenty years. Earlier scanning devices consisted of charge-coupled devices or self-scanned photodiode arrays (or MOS). Contact image sensor (CIS) systems improved on these devices by replacing their optical system with a rod lens system. This arrangement permitted the distance between the image sensor and the document being scanned to be reduced to approximately 2 cm. A schematic diagram of a typical CIS system is shown in FIG. 1. The system contains three principal components: (1) an LED light source array 1 , (2) a rod lens array 3, and (3) a long image sensor array 4. LED light source array 1 illuminates a document 2, reflecting light off it through a cover glass 7 to the rod lens array 3. The rod lens array 3 is composed of a line of individual glass rods, each of which will cover approximately 16 photodetectors 4.6. The rod lens array 3 focuses the reflected image to the image sensor array 4. Image sensor array 4 is composed of a plurality of individual image sensing photoelements 4a, each composed of a number of photodetectors 4b. Image sensor array 4 is typically mounted on a hybrid image sensor board 5 capable of converting the image to an electronic signal.
All of the components and the optical path are integrated into a compact module called the CIS module 6, which is shown in cross section in FIG. 2. In addition to the components mentioned above, the CIS module 6 includes an output line 8, and a plastic case 9. The output line 8 transmits the signal generated by the hybrid image sensor board 5 to the outside circuitry. The plastic case 9 houses all of the above components in a compact module.
One well known embodiment of this system involves a paper feed image scanning system such as a facsimile machine. The CIS module 6, including the cover glass 7, is fixed in place relative to the roller 10. The document 2 is sandwiched between the roller
10 and cover glass 7 and is held in place by its contact with both the roller 10 and the cover glass 7. When a stepping motor 10a causes roller 10 to rotate, the roller 10 advances the document 2 until the first line of the document 2 comes into position to be read by the CIS module 6. The light source array 1 is turned on. The optical signal of the reflected document is then focused on a series of photoelements 4a. This series of photoelements 4a converts the optical signal into an electronic signal in either charge or voltage form and stores that signal in its associated holding capacitors.
The CIS module 6 then integrates the first line of the document. After the CIS module 6 has read out the first line of the document 2, the CIS module 6 sends a signal to the stepping motor 10a. In response to the signal, the stepping motor 10a starts driving roller 10, continuing until the second line of document 2 comes into position to be read by the CIS module 6. The CIS module 6 then reads the second line of the document. This process is repeated until each line of the document has been read. The stepping motor 10a then advances document 2 one final time to eject it from its position sandwiches between the roller 10 and the cover glass 7.
FIG. 3 shows the prior art circuit diagram of the hybrid image sensor board 5. The hybrid image sensor board 5 is a long printed circuit board containing a plurality of identical individual silicon contact image sensor chips 11 butted together end-to-end on a single substrate in one line with approximately equal space between adjacent detectors. The board 5 also includes with peripheral circuitry including signal processing means 12 to serially activate the individual chips and an operational amplifier 13 capable of combining and then amplifying all of the analog signals generated by said CIS chips. The number of individual image sensor chips 11 chosen depends on the desired width of scanning. If twenty-seven individual chips are selected, the resulting width will be roughly equivalent to standard paper width, which is about 8.5 inches for A4 size paper.
The hybrid image sensor board 5 generates an analog video signal output. Two input pulses generated by the signal processing means 12 trigger the operation of the CIS module 6. These two triggering input pulses are a start pulse φSP 14 and a clock pulse φCP 15. As shown in FIG. 3, the start pulse φSP 14 triggers the generation of an input pulse φ|p 16 by a first chip 11. The first chip 11 also generates an end pulse φEP 17 that triggers the generation of an input pulse φiP 16 by the second chip. The process repeats itself until all chips 11 have fired. On all of the chips 11 , the input sites for the
clock pulses φCP 15 are wired together and therefore all clock pulses φcP 15 occur simultaneously. All of the analog signals generated by the image sensor board 5 are combined and then amplified by the operational amplifier 13.
The timing diagram of the prior art hybrid image sensor board 5 is shown in FIG. 4, illustrating the temporal relationship between the activation of stepping motor 10a to advance the document 2 to the next line which is to be read out, the transmission of clock pulse φCP 15, the transmission of start pulse φSP 14, the time required for each chip 11 to read a given line of the document 2, the readout time required for one line of the document to be read, and the integration time, which is the time between two consecutive readouts for a given photodiode.
A block diagram illustrating the function of the prior art CIS module 5 using phototransistor sensing elements 4a is shown in FIG. 5. (The structure and function of this chip is described in detail in U.S. Pat. No. 5,299,013, issued Mar. 29, 1994.) The image sensor chip 11 comprises an array of phototransistor sensing elements 111 , an array of multiplexing switches 112, an n-stage digital scanning shift register 113 comprising n identical stages 113b, a built-in buffer 113a, and an active chip selector 114.
In operation, the image sensor chip 11 is triggered by a start pulse to the first in sequence active individual photoelement 1 1 1 which serially activates the photodetectors 4b on the first photoelement 111. After the signal from the last phototransistor sensing element 111 of the first image sensor chip 11 is read, an end of scan pulse is generated so that the next image sensor chip in sequence is triggered. Each of the multiplexing switches is coupled between one of the photoelements and the output line.
The input pulse φiP 16 triggers the reading of the current line of the document 2 by each phototransistor sensing element 111 in sequence. The input pulse φiP 16 also triggers the generation of the clock pulse φCP 15. The input pulse φ)P 16 is transmitted to the first stage 113b of the shift register 113, which then activates the active chip selector 114.
Each stage 113b of the shift register 113 is then turned on one by one in sequence.
The output of each shift register stage 113b is connected to a control input of one of the multiplexing switches 112. When one of the multiplexing switches 112 is activated, it
then transmits the signal to a video line 115. After all of the shift register stages 113b have been scanned, the shift register 113 generates the end pulse φEP 17. This end pulse φEP 17 disables the active chip selector 114 and indicates that the scanning by that chip is completed. In this device, the readout mechanism is transmitted by the current flowing into video line 115. The phototransistor 111 transfers an electronic signal to video line 115 and video line 115 simultaneously reads out an electronic signal from phototransistor 111. This transfer and readout scheme is known as a pixel transfer and pixel readout (PTPR) structure.
FIG. 6 shows a block diagram of the prior art image sensor chip 4 utilizing a photodiode sensing element 4a and a differential voltage pickoff. (The structure and function of this chip is described in detail in U.S. Pat. No. 5,724,094 issued Mar. 3, 1998.) The output of the nth stage 113b of the shift register 113 is transmitted to a corresponding nth multiplexing switch 117 and to an (n-1)the reset transistor 118. Turning on multiplexing switch 117 initiates the readout mechanism. Turning on a reset transistor 118 initiates the reset mechanism. The device simultaneously reads out the nth pixel signal and resets the (n-1)the pixel signal. The charge readout and charge transfer is thus separated by one clock cycle. This transfer and readout scheme constitutes another example of a PTPR structure.
One difficulty of the prior art has involved distortion which arises during pixel readout in the video signal output of a CIS module 6 employing a PTPR image sensor board 5 and a constant light source (as is used, for example, with a facsimile machine). The distortion arises due to different integration times for each pixel in the scanned document. After readout, each pixel of the video signal combines data from the present and immediately previous lines. In addition to this undesirable distortion, this effect also reduces the vertical resolution to approximately half of its optimum value.
Consider the simplified, digitized document as shown in Fig. 7(a) in which four rows of the original digitized document are shown. When the first row of the document is located on the glass surface of the CIS module, the first sensing element reads a first pixel on the document. At this time, no signal for the first pixel 18 is detected because the light source did not illuminate the first pixel 18 yet. After the first pixel 18 is readout, a first sensing element integrates the first pixel 18. During this time, a second sensing element reads a second pixel 19 on the document. The output of the second sensing element reads out 10% of the second pixel 19 on the document. After readout, the second
sensing element integrates the second pixel 19. This process is repeated again until end of the row. The CIS module is then moved to second row and reads the image signal on second row of document. The CIS module read first pixel 20 out from second row. In this condition, the first sensing element on the CIS module carries about 100% of first pixel 18 signal on first row and 0% of first pixel 20 signal on second row. Similarly, the second sensing element on CIS module carries about 90% of the second pixel 19 signal on the first row and about 10% of the second pixel 21 signal on the second row. The video signal output of the CIS module is distorted because the integration time of each pixel is different, varying from 0% to 100% of the integration time. The sensing element on the CIS module already carries a lot of the information from previous row. After readout, each pixel of the video signal consists in part of the present row and in part of the previous row. FIG. 7b shows the image data read out from an image sensor chip having a PTPR structure. The image data comprises part of the image signal on the nth row and part of the image data on the (n-l)the row.
While strobe light illumination of the CIS module 6 is known in the prior art to prevent this problem, as with a color scanner, it does reduce the scan speed to half of what it would otherwise be. FIG. 8 shows a prior art timing diagram using strobe light illumination. The strobe is timed to turn on and off at a frequency which enables the CIS module 6 to avoid the distortion discussed above by only illuminating the pixels it is desired to sample within a given line. However, the scan time to read one line is now the sum of the integration time 22 during which the light source array 1 is turned on plus the pixel readout time 23 during which the light source array 1 is turned off. In this case, the image data is not distorted, but the scan speed is reduced by half.
Thus a need exists for an accurate, sensitive CIS chip with high signal-to-noise ratio capable of avoiding distortion, vertical resolution reduction, and reduction of scan speed. It is desirable to create a device in which all necessary clock pulses and operational amplifiers can be built on a single chip, reducing necessary peripheral circuitry.
SUMMARY OF THE INVENTION
The invention provides an image sensor chip for a CIS module capable of accurate scanning a document at optimal speed while avoiding distortion effects and maintaining optimal vertical resolution. The circuit comprises a plurality of sensing photoelements and a like number of dummy elements capable of providing a baseline signal. The
signal from the dummy elements is combined with the signal from the active photoelements by an operational amplifier which then outputs a signal scrubbed of all dark fixed pattern noise and all direct current (DC) offset voltage.
The invention uses a line transfer and pixel readout (LTPR) structure. A series of photodetectors scan each line of a document and generate a signal. One by one, each line of data is isolated and then parallel transferred via a line of transfer gates to holding capacitors associated with the corresponding photodetectors. A digital scanning shift register then reads out a line of the video signal. In this case, the integration and readout are therefore separated. This structure not only provides more accurate data and higher vertical resolution but also increases scan speed. To build a CIS board, the chips may be butted together end-to-end on a single substrate in one line with approximately equal space between adjacent detectors. Such a CIS module offers higher sensitivity, low reset noise, and high signal-to-noise ratio.
The dummy elements function in the same manner as the active photoelements. However, they are shielded by some blocking means so that they receive no light. The output of the dummy sensors will therefore not vary with the images on the subject document. The dummy sensors therefore provide a baseline signal that is used to cancel any dark fixed pattern noise or DC offset voltage from the signal from the sensors. This scrubbing of the output signal is accomplished in the operational amplifiers. The optical black level reference also significantly assists the process of color reconstruction. The image sensor chip internally generates the two external input clock pulses needed to drive its operation, clock pulse φCP 15 and input pulse φ)P 16.
One advantage of the invention is that the transfer gates permit the separation of the sensing element from the holding capacitor, so that the sensing element can integrate a line of signal and the previous signal on the holding capacitor can be simultaneously read out by the shift register. Because all of the charge is parallel transferred, the accuracy and vertical resolution are both excellent. This structure improves the scan speed.
A further advantage of this invention is that a small value for the capacitance CH of the holding capacitor increases the device's sensitivity, reduce reset noise, and improves signal-to-noise ratio.
A further advantage of this invention is that all necessary clock pulses and operational amplifiers can be built on a single chip, reducing the necessary peripheral circuitry.
A still further advantage of this invention is that the linearity of the photoresponse is greatly assisted by the optical black level reference, strongly promoting accurate color reconstruction.
A still further advantage of this invention is that the dummy element array is operated independently from the active photoelement array, promoting design flexibility for different configurations of the CIS modules.
Other and further advantages, embodiments, variations and the like will be apparent to those skilled in the art from the present specification taken with the accompanying drawings and appended Claims.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a schematic diagram of a prior art contact image sensor (CIS) system;
FIG. 2 shows a cross section view of a prior art CIS module;
FIG. 3 shows a prior art circuit diagram of a hybrid image sensor board;
FIG. 4 is a timing diagram for the prior art hybrid image sensor board;
FIG. 5 is a block diagram of the prior art silicon butting CIS module with phototransistor sensing elements;
FIG. 6 is a block diagram of the prior art silicon butting CIS module with photodiode sensing elements;
FIGS. 7a and 7b provide a schematic representation of a simplified, digitized original document (FIG. 7a) and a detected range signal using a prior art, silicon butting image sensor chip having a pixel transfer and pixel readout (PTPR) structure (FIG. 7b);
FIG. 8 is a prior art timing diagram with strobe light illumination;
FIG. 9 is a timing diagram for a silicon butting CIS chip with line transfer and pixel readout (LTPR) structure;
FIGS. 10a and 10b provide a schematic representation of a simplified, digitized original document (FIG. 10a), and a detected range signal according to the invention (FIG. 10b);
FIG. 11 is a block diagram of the silicon butting CIS chip with LTPR structure;
FIG. 12(a) shows a cross section view of the image sensing photoelement, transfer gate, and holding capacitors for the image sensor chip with LTPR structure;
FIG. 12(b) is a potential diagram during one integration period for the image sensor chip with LTPR structure;
FIG. 12(c) is a potential diagram during one transfer period for the silicon butting CIS chip with LTPR structure;
FIG. 12(d) is a potential diagram following a transfer for the silicon butting CIS chip with LTPR structure;
FIG. 13 is a timing diagram for the CIS chip with LTPR structure;
FIG. 14 is a chip layout plan for the silicon butting CIS chip with LTPR structure;
FIG. 15 is a block diagram of a hybrid, silicon butting CIS board with LTPR structure;
FIG. 16 is a timing diagram for a hybrid, silicon butting CIS board with LTPR structure; FIG. 17 is a simplified schematic diagram of a phototransistor APS image-sensing photoelement;
FIG. 18 is a block diagram of the image sensor chip and an Active Pixel Sensor (APS) for an image sensing photoelement;
FIG. 19 is a block diagram of the CIS chip with LTPR structure, an APS image sensing photoelement and Correlated Double Sampling (CDS) circuitry; and
FIG. 20 is a timing diagram for the CIS chip with LTPR structure, an APS image sensing photoelement and CDS circuitry.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
One advantage of the silicon butting contact image sensor (CIS) chip with line transfer and pixel readout (LTPR) structure are that it has the capability to scan a document accurately and quickly while avoiding distortion and while maintaining optimum vertical resolution.
Referring first to FIG. 9, there is shown a timing diagram for a silicon butting CIS board 5 with LTPR structure. A series of photodetectors 4b scan the first line of the document 2 and generate an electronic signal. This electronic signal is isolated and then parallel transferred to the holding capacitors associated with the photodetectors 4b when a transfer gate is turned on, triggering the transmission of a transfer pulse φτ 24. After the transfer, all of the photodetectors 4b are reset to the reset voltage by a reset pulse φR
25. The input pulse φ)P 16 triggers the shift register to read out a line of the video signal.
The input pulse φ|P 16 also turns on the stepping motor 10a which activates the roller 10 to advance the document 2 to the next line. The CIS board 5 then begins to read out the next line of document 2. As can be seen, integration times for all pixels are identical. The invention thus avoids the distortion problem experienced in the prior art due to different integration times for each pixel in the scanned document. Optimum vertical resolution is attained, providing accuracy which is superior to prior art conventional PTPR image sensors. The scan period for one line equals readout time 23. Unlike the case with the prior art system which used strobe light illumination to eliminate distortion, no additional integration time 22 is necessary to complete a scan period (See FIGS 10a and 10b). Relative to the prior art, the scan speed is improved significantly.
FIG. 11 shows a simplified block diagram of the silicon butting CIS board 5 with LTPR structure. Its principal components are (1) buffer 26, (2) a timing generator 27, (3) an active photoelement array 28, and (4) a dummy element array 29. The buffer 26 is used to isolate the device from outside circuitry and to provide enough power to drive the timing generator 27, the active photoelement array 128, and the dummy element array 29. Upon receipt of two pulses, clock pulse φCP 15 and input pulse φ)P 16, the
timing generator 27 generates the transfer pulse φτ 24, reset pulse φR 25, and frame pulse 25b.
The active photoelement array 28 consists of several components: (1) a plurality of active photoelements 30, (2) a plurality of transfer gates 31 , (3) a plurality of holding capacitors 32, (4) a plurality of reset transistors 33, (5) a plurality of emitter-followers 33a, (6) the plurality of multiplexing switches 37, (7) the n-stage digital scanning shift register 38, (8) the active chip selector 39, and (9) a video buffer 40. Each emitter-follower 33a is in turn composed of one MOS transistor 134, one depletion transistor 35 and one energy-saving transistor 36.
The dummy element array 29 contains a lesser number of the same components as the active photoelement array 28, replacing the plurality of active photoelements 30 by a plurality of dummy elements 41. The dummy elements 41 have the same size as the active photoelements 30. Each dummy element 41 is placed in close proximity to a corresponding said active photoelement 30 but is shielded by some blocking means, generally aluminum, to block out the light. Only dark fixed pattern noise is generated on the dummy elements 41. The dark fixed pattern noise associated with the photoelements 30 is the same as those associated with the dummy elements 41. Comparison of the outputs from the active photoelements 30 and the dummy elements 41 permits the scrubbing from the output signal of dark fixed pattern noise and DC offset voltage. Use of dummy elements 41 cancels out the dark fixed pattern noise of the photoelements 30.
The active element array may be operated either before or after the dummy element array. The active element array and the dummy element array may each be either operated or disabled without reference to whether the other array is being operated or is disabled.
FIG. 12(a) shows a cross section view of an image sensing photoelement 30, transfer gate 31 , and holding capacitor 32. The photoelements 30 are composed of a photodiode 301 and a V0G gate 302. Voltage of the V0G gate 302 is adjusted between the high and low levels of the transfer pulse φτ 24, setting a threshold level for said photodiode. The V0G gate 302 is used to set threshold levels for the photodiode 301. In one preferred embodiment, the V0G gate 302 will be made of polysilicon. The
V0G gate 302 also can be made by implantation of an ion with a potential voltage equal to the V0G gate potential.
As depicted in potential diagrams FIGS. 12(b), 12(c), and 12(d), the transfer gate 31 is used to transfer a photocharge 303 from the photoelement 301 to the holding capacitor 32. As shown in FIG. 12(b), a potential diagram during one integration period, within an integration period the transfer gate 31 is turned off and the document 2 is illuminated by the LED light source array 1. The photocharge 303 is generated from photoelement 301 and stored in photoelement 301. As shown in FIG. 12(c), a potential diagram during one transfer period, within a transfer period the transfer gate 31 is turned on and the LED light source array 1 is shut off. The stored charge 303 on photoelement 301 is transferred to the holding capacitor 32.
As shown in FIG. 12(d), a potential diagram following a transfer, after all of the charge 303 is transferred to the holding capacitor 32, the transfer gate 31 is turned off again. At this point, all of the photocharge 303 from photoelement 301 has been transferred to the holding capacitor 32. The photoelement 301 is then ready to integrate the next line of the document 2. Because the photoelement 301 is separated from the holding capacitor 32 by the transfer gate 31 , the charge on the photoelement 301 differs from that on the holding capacitor 32. The charge on the holding capacitor 32 represents the signal from the previous or (n-1)the line. The charge on photoelement 301 represents the signal from the current or nth line. Therefore, one can read the signal from the holding capacitor 32 and integrate the signal on the photoelement 301 simultaneously. The period of time to read a line of the document 2 equals either the integration time 22, or the readout time 23, significantly increasing the line scan speed. The use of holding capacitor 32 increases not only the line scan speed but also the device's sensitivity and also reduces the reset noise, significantly improving the signal-to-noise ratio. The voltage across the holding capacitor 32 equals the photocharge 303 divided by the capacitance. If the photocharge 303 generated by the photoelement 301 is ΔQ, and the capacitance of the holding capacitor 32 is CH, then the signal ΔVH = ΔQ / CD.
Normally, the capacitance of the photoelement 301 is large due to its large size (in a preferred embodiment, approximately 125 μm by 125 μm for 200 dpi). The voltage across the photoelement 301 is ΔVD = ΔQ / CH. Because the video signal is increased from ΔVD across the photoelement 301 to ΔVH across the holding capacitor 32, the device amplifies the signal by a factor of G.
G = CD / CH Eq. (1) where CD is the capacitance of the photoelement 301 , and CH is again the capacitance of the holding capacitor 32.
As shown in Eq. (1), a small value for the capacitance CH of the holding capacitor 32 will increase the device's sensitivity G.
The other advantage of a small holding capacitor 32 is the reduction of reset noise. Because the charge on the holding capacitor 32 must be reset after readout, the reset mechanism generates a reset noise when resetting the holding capacitor 32 to DC level. The reset noise is proportional to the square root of the capacitance of the holding capacitor 32. Reduction of the capacitance of the holding capacitor reduces reset noise by a factor of K.
K = (CD / CH) 1/2 Eq. (2)
For a given holding capacitor 32 capacitance CH, the reset noise NR is given by: NR = [(k x T ) / q ] x (CH)1 2 = 400 (CH)1/2 electrons @ 25°C Eq. (3)
If CH equals 0.1 pF, the reset noise is 126 electrons at room temperature. This reset noise is negligible compared to a photocharge of one million electrons on the photoelement 301. The signal-to-noise (S/N) ratio is therefore significantly improved. The charge on the holding capacitor 32 is converted to a voltage signal by the emitter-follower 33a. As mentioned above, the emitter-follower 33a is composed of an MOS transistor 34, a depletion transistor 35, and an energy-saving transistor 36. The depletion transistor 35 acts likes a load resistor, stabilizing the signal by providing a constant current while voltage varies. The gate of the energy-saving transistor 36 triggers a frame pulse φF from the timing generator 27. The emitter-follower 33a comes on only when this frame pulse is activated and is turned off during the holding period, significantly reducing power dissipation. This function reduced the power dissipation on the emitter-follower 33a. The voltage signal ΔV output by the emitter-follower 33a
equals the gain A of the emitter-follower 33a times the voltage across the holding capacitor 32.
ΔV = (Δ Q / CD) x A Eq. (4) where A is the gain of the emitter-follower 33a.
This signal is then read by the multiplexing switch 37 that is controlled by the n-stage shift register 38. After the signal is read by the n-stage shift register 38, the holding capacitor 32 is then reset to the reset voltage by turning on the reset transistor 33. The holding capacitor 32 is then ready to receive the signal from the second line of the active photoelement 30.
As mentioned above, upon receipt of two pulses, clock pulse φCP 15 and input pulse
ΦIP 16, the timing generator 27 generates the transfer pulse φτ 24, the reset pulse φR 25, and frame pulse φF 25b. Both the n-stage shift register 38 and an m-stage digital scanning shift register 42 associated with the dummy element array 29 operate in response to the same clock pulse φCP 15. Because the start pulses of these two shift registers are designed separately, one can read the active pixels and dummy pixels independently. An active start pulse φASP 43 is used to activate the n-stage shift register 38, whereas dummy start pulse φDSP 46 is used to trigger the m-stage shift register 42. The number of stages in the m-stage shift register 42 is smaller than the number of stages in the n-stage shift register 38. The time interval between successive activations of the m-stage shift register 42 will be longer than the time interval required to generate transfer pulse y 24 and reset pulse φR 25, and frame pulse y 25b.
FIG. 13 shows a timing diagram of an individual image sensor chip with LTPR structure. The dummy start pulse φDSP 46 on the dummy element array 29 connects to the input pulse φ|p 16, which activates the m-stage shift register 42. After m clock cycles, the dummy element array 29 generates the dummy end pulse φDEP 47. The segment of the dummy output signal VD048 between the input pulse φjP 16 and the transfer pulse φτ 24 is distorted by the reset and transfer process and cannot be used for black level reference. The segment of the dummy output signal VD0 48 starting with the falling edge of the transfer pulse φτ 24 and ending with the dummy end pulse φDEP 47
represents dark fixed pattern noise from the dummy element array 29. This dummy output signal VD048 is used to provide a black level reference that cancels out the dark fixed pattern noise from the active photoelement array 28. The dummy end pulse φDEP
47 of the dummy element array 29 connects to the active start pulse φASP 43 of the active photoelement array 28, inducing the active photoelement array 28 to activate the active photoelement 30 and read the active video signals one by one in sequence.
The use of the dummy element 41 to cancel out the dark fixed pattern noise from the photoelement 30 can be explained as follows. During the integration time 22, the charge on the capacitor associated with each active photoelement 30 is gradually removed by reverse current flowing across the associated active photoelement 30. The reverse current consists of two components: photocurrent and dark fixed pattern noise. The photocurrent equals the product of photodetector responsiveness and light intensity. During a line scan, the charge integrated from each photodetector is the product of the sum of photocurrent and dark fixed pattern noise ("dark leakage current") times the integration time 22. This charge is stored on the active photoelement 30, which builds up a potential voltage. Equation 5 describes the charge built up on photoelement site:
ΔQA = (lL + lD) x Tint Eq. (5) where lL is light current, lD is dark leakage current, and Tint is the integration time 22 of the image sensor.
Assuming the size of the dummy element 41 is the same as that of the active photoelement 30, the charge built up on the dummy element 41 equals the product of dark leakage current times the integration time Tint 22.
ΔQD = lD x Tint Eq. (6) The emitter-follower 33a converts charge on the holding capacitor 32 to its associated voltage level as explained above. The active output voltage Δ VA0 of active photo element array is linear proportional to charge on the active photoelement 30.
ΔVA0 = (Δ QA / CH) x A Eq. (7)
where CH is the capacitance of the holding capacitor 132, and A is the gain of the emitter-follower 33a.
The dummy output voltage ΔVD0 of the dummy element array 29 is linearly proportional to the charge on the dummy element array 29.
ΔVD0 = (Δ QD / CH) x A Eq. (8)
The capacitances of the holding capacitors 32 in the active photoelement array 28 and dummy holding capacitors in the dummy element array 29 are the same, as are the gains of the emitter-followers 33a in the two arrays. In order to cancel out the dark fixed pattern noise, the system is designed so that the net output signal ΔV equals the active output signal 45 from the active photoelement array 28 minus the dummy output signal 48 from the dummy element array 29.
ΔV =Δ VA0 - Δ VD0= [(lL + lD) - lD] x Tirrt = - |L x η int Eq. (9)
The integration time Tint 22 is constant for all of the active photoelements 30 and dummy elements 41. Therefore, the net output signal ΔV is linearly proportional to the photocurrent as shown in equation 9. The photocurrent is the product of the photoresponse R times the light intensity λ|. The photoresponse R of the active photoelement 41 is the same for each of the active photoelements 30 and dummy elements 41. Therefore, the net output signal ΔV is linearly proportional to the light intensity λ| from the reflected image of the document 2.
ΔV oc λ, Eq. (10)
The photoresponse linearity, defined as the net output signal Δ V divided by the light intensity λ|, is thus constant. While the linearity of the photoresponse is a desirable aspect for embodiments of the invention using CIS modules 6 designed to scan black and white documents 2, this feature is particularly important for embodiments involving color CIS modules 6. In the case of applications using color CIS modules 6, each pixel of the digitized representation of the color document 2 is composed of a combination of three colors: red, green and blue. A conventional color scanner offers at least 8 bits for
each color or 28 = 256 total options for each color. Therefore, considering that each pixel is composed of combinations of these three colors, a grand total of 224 = 16,777,216 different colors may be reconstructed. The color C of a given pixel is determined according to the formula:
C = A% x R + B% x G + C% x B Eq.
(11)
where
A% is the percentage representation of the red color under the three-color description of the pixel,
B% is the percentage representation of the green color, C% is the percentage representation of the blue color, R denotes the red color,
G denotes the green color, and B denotes the blue color.
Note that A + B + C = 100. The above demonstrated constancy of the photoresponse linearity strongly promotes accurate reconstruction of the color image of the document 2.
FIG. 14 shows a chip layout plan for the silicon butting CIS chip with LTPR structure. As pictured, the active photoelement array 28 is arranged on the upper half of the image sensor chip 11 , while the dummy element array 29 is located on the lower half of the chip 11. A total of n active photoelements 30 are set in a single line at equal distances from each other. A very long active photoelement 30 or series of photoelements 30 which may, for example, be A4 size, can easily be butted on the chip 11.
FIG. 15 is a block diagram of the silicon butting CIS board with LTPR structure, comprised of (1) a plurality of silicon butting CIS chips 11 ; (2) a flip-flop circuit 67; and (3) an operational amplifier 68a. In response to the start pulse φSP 14, the dummy element array 29 on a first chip 49 emits a dummy start pulse φDSP 52 which triggers the input pulse φiP 16. A dummy end pulse φDEP 53 is emitted by the dummy element array 29. The dummy end pulse φDEP 53 is coupled to an active start pulse φASP 54.
This design allows for great flexibility in that further connections to the dummy element array 29 may be arranged prior to the connection to the active photoelement array 28 in order to provide a longer time for determination of the black level reference. The dummy element array 29 produces a dummy output signal VD0 55 which is combined with an active output signal VA0 56 generated by the active photoelement array 28.
The active photoelement array 28 of the first chip 49 emits an active end pulse φAEP 58, which connects to a second chip active start pulse φASP 59 on a second chip 51 , and so on. The last chip 50 connects a last chip active start pulse φASP to the active end pulse of previous chip and outputs 60 to a last-but-one chip active end pulse φAEP 161 on a last-but-one chip 50. The active photoelement array 28 triggers a last chip active end pulse φAEP 62 from the last chip 50 which in turn connects to a dummy start pulse φDSP
63 on the dummy element array 29. A dummy output signal VD0 64 from the dummy element array 29 is combined with an active output signal VA0 65 generated by the active photoelement array 28. Three bonding pads exist for the active photoelement array 28, corresponding to the active start pulse φASP 54, the active end pulse φAEP 58, and the active output signal, VA065. Three bonding pads exist for the dummy element array 29, corresponding to the dummy start pulse φDSP 52, the dummy end pulse φDEP
53, and the dummy output signal VD0 55. The dummy end pulse ΦDEP 53 and the start pulse φSP 14 are connected to alternate sides of the flip-flop circuit 67. The flip-flop circuit 67 generates an LED output pulse φLED 68. This pulse causes the automatic illumination of the document 2. The dummy element arrays 29 of all chips 51 other than the first chip 49 and the last chip 50 are disabled by disabling the dummy start pulse φDSP 52a, the dummy end pulse φDEP 53a and the dummy output signal VD055a. All of the analog signals generated by the chips 11 are combined and then amplified by the operational amplifier 68a.
FIG. 16 is a timing diagram for the hybrid, silicon butting CIS board with LTPR structure. A number m-) of preceding dummy pixels 69 from dummy signal precede active pixels
70, which carry the video signal, while a number m2 of following dummy pixels 71 follow the active pixels 70. As discussed above, the preceding and following dummy pixels 69 and 71 enable the cancelling out of the dark fixed pattern noise. In operation, upon receipt by the CIS module 6 of a start pulse φSP, the timing generator 27 simultaneously generates the reset pulse φR 25, the frame pulse φF, and the transfer pulse φj 24. The
capacitance of the holding capacitor 32 is reset to the reset voltage by the reset pulse φR 25. The transfer pulse φτ 24 triggers the operation of a series of active photoelements 30. This electronic signal is isolated and then parallel transferred to the holding capacitors 32 associated with the active photoelements 30.
The series of active photoelements 30 integrates the next line of the document 2. The charge on the active holding capacitor 32 is converted to a voltage signal by the emitter-follower 33a as described by equation (4). This signal is then read by the multiplexing switch 37 that is controlled by the n-stage shift register 38. After the signal is read by the n-stage shift register 38, the holding capacitor 32 is then reset to the reset voltage by turning on the reset transistor 33. The holding capacitor 32 is then ready to receive the signal from the second line of the photoelement 30. This mechanism is repeated again to read the next line of the document 2.
In one embodiment, a passive pixel sensor (PPS) can be used to make an active photoelement.
FIG. 12 shows the active photodiode PPS comprises a photodiode 301 and a V0G gate. The voltage level of the V0G gate is adjusted between the high and low levels of the transfer pulse φτ in order to set a threshold level for the photodiode. The photodiode accumulates charge and then converts a charge signal to a voltage signal. In one embodiment, the active photodiode is an active pn junction photodiode capable of converting a charge signal to a voltage signal. In an alternative embodiment, the diode is a p-i-n photodiode, which has ten times lower dark fixed pattern noise than does a conventional pn junction photodiode. In one embodiment, the V0G gate is an gate made by ion implantation. In this case, a single polysilicon wafer process can be used.
In an embodiment having a photodiode with an image sensor chip principally comprises: (1) a buffer, (2) a timing generator, (3) an active photoelement array, and (4) a dummy element array. The active photoelement array itself comprises several components: (1) a plurality of active PPS's, (2) a plurality of active holding capacitors, (3) a plurality of reset transistors; (4) a plurality of emitter-followers, (5) a plurality of multiplexing switches, (6) the n-stage shift register, (7) the active chip selector, (8) the video buffer, and (9) a plurality of transfer gates capable of separating the charges on the active PPS's from the charges on the active holding capacitors and capable of
transferring the charges on the active PPS's to the active holding capacitors. This arrangement permits the simultaneous reading of the charge on the active holding capacitor and integration of the charge on the photoelement.
A dummy element array contains a lesser number of the same components as the active photoelement array, replacing the plurality of active PPS's by a plurality of dummy PPS's. Again, the dummy element array may be simplified by separating out selected components from the rest of the dummy array. For example, if the m-stage shift register 42 is separated out, a dummy output signal VD0 of constant voltage results.
Each emitter-follower is composed of one MOS transistor 34, one depletion transistor 35 and one energy-saving transistor 36. The depletion transistor acts likes a load resistor, stabilizing the signal by providing a constant current while voltage varies. The gate of the energy-saving transistor triggers a sample pulse. The emitter-follower comes on only during the sampling period and is turned off during the holding period, significantly reducing power dissipation. During the integration period, photocharges are generated by the photodiode and stored by its associated holding capacitor. The photocharge on the holding capacitor is converted to a voltage signal by the emitter-follower.
In another embodiment, an active pixel sensor (APS) can be used to make a CIS chip 11. The chip principally comprises: (1) a buffer, (2) a timing generator, (3) an active photoelement array, and (4) a dummy element array. The active photoelement array 128 itself comprises several components: (1) a plurality of active APS's, (2) a plurality of sample/hold switches, (3) a plurality of holding capacitors, (4) a plurality of multiplexing switches, (5) the n-stage shift register, (6) the active chip selector, and (7) the video buffer. The dummy element array again contains a lesser number of the same components as the active photoelement array, replacing the plurality of APS's by a plurality of dummy APS's. The dummy elements have the same size as the active photoelements but are covered with aluminum to block out the light. Again, the dummy element array may be simplified by separating out selected components from the rest of the dummy array. For example, if the m-stage shift register is separated out, a dummy output signal VD064 of constant voltage will result.
FIG. 17 shows a simplified schematic diagram of a phototransistor APS. The phototransistor APS comprises: (1) a phototransistor 85, (2) a pn junction photodiode
86, (3) a base reset transistor 87, (4) a APS capacitor 88, and (5) an emitter-reset transistor 89. The phototransistor 85 converts an optical signal to an electronic signal. The phototransistor 85 switches on when the base voltage exceeds a fixed non-zero emitter voltage such as 0.7 V. In order to prevent problems in low lighting levels, the pn junction photodiode 86 is inserted between the base of the phototransistor 85 and the base reset transistor 87. The base voltage resets to a fixed non-zero voltage such as 0.7 V instead of to ground as with a conventional phototransistor structure. The other advantage of the pn junction diode 86 is it allows for rapid temperature equalization between the photodiode 86 and the base-emitter of the phototransistor. During reset, the emitter-reset transistor 89 resets the emitter to ground, whereas the base reset transistor 87 resets the voltage across the base of the phototransistor to a fixed non-zero value such as 0.7 V. Both the base and the emitter of the phototransistor are reset to the reset voltage at the same time. The APS capacitor 88 stores the charge corresponding to the converted signal. The combination of the phototransistor 85 and the APS capacitor 88 effectively acts as an emitter-follower. This design follows a simple schematic, lowering the cost of the board.
FIG. 18 shows a block diagram of the image sensor chip 11 with an active photodiode APS 74 as an active image sensing photoelement 30. Each photodiode APS 74 is comprised of one photodiode 301 , one reset transistor 33, and one emitter-follower 33a. The photodiode 301 converts a charge signal to a voltage signal on every pixel. Each emitter-follower 33a is in turn composed of one MOS transistor 34, one depletion transistor 35 and one energy-saving transistor 36. The depletion transistor 35 acts likes a load resistor, stabilizing the signal by providing a constant current while voltage varies. The gate of the energy-saving transistor 36 triggers a sample pulse. The APS 74 comes on only during the sampling period in response to the chip selector pulse and is turned off during the holding period, significantly reducing power dissipation. During the integration period, photocharges are generated by the photodiode and stored by its associated capacitor 32. The photocharge is converted to a voltage signal by the emitter-follower 33a. In one embodiment, the active photodiode is an active pn junction photodiode capable of converting a charge signal to a voltage signal on every pixel. In an alternative embodiment, the diode is a p-i-n photodiode, which has ten times lower dark fixed pattern noise than does a conventional pn junction photodiode.
For both the PPS and APS embodiments, the voltage signal ΔV generated by the emitter-follower 33a is given by the equation:
ΔV = (ΔQ / CD) x A Eq. (12) where Δ Q is the photocharge, CD is the photodiode capacitance, and A is the gain of the emitter-follower 33a.
For the APS system, during the transfer period, the sample/hold switch 75 generates a sample/hold pulse φs H which triggers the sampling of the voltage signal ΔV and then the holding of the signal on the holding capacitor 32. After the sampling and holding has taken place, the reset transistor 33 resets the charge on the photodiode 301 , causing the voltage signal ΔV to be reset. The photodiode 301 is then ready to integrate the charge from the next line of the document 2. The voltage ΔVC across the holding capacitor 32 thus represents the video signal from the previous line of the document 2. The discharge of the voltage ΔVc(t) across the holding capacitor 32 as a function of time t is described by:
ΔVc(t) = ΔV x exp (φt/RCH) Eq.
(13) where R is the effective resistance of the system, and CH is again the capacitance of the holding capacitor 32.
Therefore, the capacitance CH of the holding capacitor 32 must be large enough that the voltage signal ΔV can be retained for one transfer period. The voltage signals across the holding capacitors 32 are then read one by one in sequence by the multiplexing switch 37 that is controlled by the n-stage shift register 138.
After a signal has been read by the n-stage shift register 38, the reset transistor 33 resets the voltage signal ΔV. Reset noise in the form of kTC noise results and may be significant given the large surface area of the active photoelement 30. Therefore a preferred embodiment incorporates correlated double sampling (CDS) circuitry to cancel the reset noise. This is achieved by creating more than one row of active sample/hold switches, active holding capacitors, and active multiplexing switches as well as more than one row of dummy sample/hold switches, dummy holding capacitors, and dummy multiplexing switches. Again, the dummy element array may be simplified by
separating out selected components from the rest of the dummy array. FIG. 19 shows a block diagram of the CIS chip with LTPR structure, an APS image sensing photoelement and Correlated Double Sampling (CDS) circuitry.
FIG. 20 is a timing diagram for the same chip. Instead of using one sample/hold pulse as in the previous embodiment, this embodiment uses two sample/hold pulses φs Hι
93 and φs H2 94 to cancel reset noise.
The above disclosure is not intended as limiting. Those skilled in the art will readily observe that numerous modifications and variations of the present invention are possible while retaining the teachings of the invention. Accordingly, the invention may be practiced otherwise than as specifically described herein.