Photodiode CMOS imager with column-feedback soft-reset for imaging under ultra-low illumination and with high dynamic range

[0001] This application claims the benefit of U.S. Provisional Patent Application No. 60/249,864, filed on Nov. 16, 2000, and U.S. Provisional Patent Application No. 60/311,475 filed on Sep. 10, 2001, the disclosures of which are hereby incorporated by reference.

[0002] 1. Field of the Invention

[0003] The present invention relates to the field of image sensing devices. More specifically, the present invention relates a to photodiode CMOS imager with column-feedback soft-reset for imaging under ultra-low illumination and with high dynamic range.

[0004] Portions of the disclosure of this patent document contain material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure as it appears in the Patent and Trademark Office file or records, but otherwise reserves all rights whatsoever.

[0005] 2. Background Art

[0006] Image sensors are devices capable of converting an image into a digital image. Image sensors are also referred as silicon film or silicon eyes. These devices are made of silicon since silicon has the properties of both being sensitive to light in the visible spectrum and being able to have circuitry integrated on-board.

[0007] A CMOS imager is a image sensor made from silicon. The process by which a CMOS imager is made is called CMOS, which stands for Complementary Metal Oxide Semiconductor. CMOS imagers include an array of photo-sensitive diodes, one diode within each pixel. FIG. 1 illustrates a CMOS imager array. Each pixel, 100, in a CMOS imager has a radiation sensitive element, 110, with each radiation sensitive element connected to an amplifier, 120. A CMOS pixel which converts an optical image into an electronic signal with an arrangement of having an amplifier attached to each radiation-sensitive diodes is called an active pixel.

[0008] The active pixels in a CMOS imager can be arranged in a matrix form and be utilized to generate video signals for video cameras, still photography, or anywhere incident radiation needs to be quantified. When an incident radiation interacts with a radiation sensitive element, 110, in a CMOS imager, charge carriers are liberated and can be collected for sensing. The number of carriers generated in a pixel, 100, is proportional to the amount of the incident light impinging on the radiation sensitive element and the sensitivity of radiation sensitive element to light. The electronic signal generated by a pixel, 100, in a CMOS imager is then read directly on an x-y coordinate system.

[0009] FIG. 2 is an illustration of a schematic circuit diagram of an active pixel, 200, of a conventional CMOS image sensor with a photodiode, 220, as a radiation-sensitive element. A photodiode, 220, comprising a p-n junction diode, wherein the p-doped junction side, 210, is connected to a negative bias voltage V_sub at 296. The n-doped junction side 230a is connected via connection 230 to the drain terminal, 230c, of a reset transistor, 240, and to the gate terminal, 230b, of a charge sensing transistor, 260. The source terminal, 270, of the reset transistor, 240, is biased to a positive reference bias voltage V_REF at 292. The gate terminal, 242, of the reset transistor, 240, is connected to a common row reset line, 250.

[0010] The source terminal, 255, of the sense transistor, 260, is connected to reference bias voltage V_DD at 294. The drain terminal of a sense transistor, 260, and the source terminal of a row select transistor, 280, are connected to each other at 265. A row select transistor, 280, has its gate terminal, 282, connected to a row select signal line, 285, and its drain terminal, 290, connected to the signal line at 298. Identical pixels represented by the active element 220, 240, 260, and 280 are laid out in form of a matrix pattern comprised of rows and columns to form the CMOS image sensor array.

[0011] FIG. 3 is an illustration of a schematic circuit diagram of an active pixel of a conventional CMOS image sensor operating under ultra-low illumination (e.g., taking photograph from outer space or at night without any flash light). The radiation-sensitive element is a SENSE element, 305, comprising a p-n junction, wherein the p-doped junction side, 310, is exposed to a low optical radiation, 315, and the n-doped junction side 330a is connected via connection 330 to the drain terminal, 330c, of a reset transistor, 340, and to the gate terminal, 330b, of a charge sensing transistor, 370. The source terminal, 360, of a reset transistor, 340, and the source terminal, 365, of a charge sensing transistor, 370, is biased to a positive reference bias voltage V_DD at 301. The gate terminal, 355, of the reset transistor, 340, is connected to a common row reset line at 350. A row select transistor, 380, has its gate terminal, 382, connected to a row select signal line, 385, and its drain terminal, 387, connected to a column signal bus, 390.

[0012] When the p-doped junction side, 310, of a SENSE element, 305, is exposed to the optical radiation of low illumination, 315, the p-doped junction side, 310, releases electrons, 320. The released electrons, 320, flow towards the n-doped junction side, 330a, of a SENSE element. The number of electrons which are released and flow depend upon the intensity of the optical radiation, 315, and the sensitivity of p-dope junction side, 310, to the optical radiation. This process of conduction can be compared to a p-n junction diode or to a photodiode.

[0013] The charge from n-doped side, 330a, flow to the gate terminal, 330b, of a sensing transistor, 370. The positive voltage at the gate terminal, 330b, would make a sensing transistor, 370, to conduct a reference bias voltage V_DD, 301. When a sensing transistor, 370, conducts, the charges are passed to the source terminal, 375, of a row select transistor, 380. The row select line, 385, would have a positive voltage when a particular row is selected to sense. When a row select line, 385, has a positive voltage, the gate terminal, 382, of a row select transistor, 380, is turned on to make a row select transistor, 380, to conduct. When a row select transistor, 380, to conducts, the charges from the source terminal, 375, are passed on to the drain terminal, 387, of a row select transistor, 380, and finally can be read-out by the column bus line, 390.

[0014] By applying a positive reset voltage to the row reset line 350, all the SENSE elements, 305, in that row are reverse biased to the reference bias voltage V_DD, 301. When the reset voltage is removed while all SENSE elements, 305, is exposed to an optical radiation, the charge stored in the SENSE elements, 305, in the respective row decreases due to the induced leakage (photo) current generated by the photo-induced electron-hole pair, causing the voltage at the gate 330b of the charge sensing transistor, 370, to decreasing proportionately. By applying a row select voltage to the gate, 382, of a row select transistor, 380, a signal representing the voltage at the gate 330b (and therefore also the charge stored in SENSE element 305) can be read out column-wise visa signal lines, 390, connected to the drain terminal, 387, of each row select transistors 380 in a respective column.

[0015] After processing the signals from all the active pixel in a CMOS imager array, the final image is reproduced in a digital form. FIG. 4 is an illustration of a CMOS imager capturing a image under ultra-low illumination. An image, 401, is captured using a CMOS imager, 402, under low illumination or low intensity of light, especially when capturing images from space or places were there is no resource of providing a flash light over the object to capture good quality image.

[0016] A CMOS imager, 402, converts the optical radiation exposed on it into electrical signal and processed to produce a digital image, 403. The digital image, 403, is not an exact replication of the actual image, 402. The digital image, 403, does not have the exact range of illumination which is present in the original image, 401. The digital image produced by a CMOS imager is dull or dark and with very low intra-scene contrast. A CMOS imager is unable to capture any image under low illumination with high dynamic range and high intra-scene contrast.

[0017] Characteristic Analysis of a CMOS Imager:

[0018] A CMOS imager experiences some unwanted electrical signal which interferes with the image being read and transferred. These unwanted electrical signals which interfere with a CMOS imager is called a read noise or temporal noise. Read noise occurs randomly and is generated by the basic noise characteristics of electronic components in a CMOS imager circuit. This type of noise can be compared to a disturbance like the snow on a bad TV reception.

[0019] To capture an image using a CMOS imager with high intra-scene contrast and wide dynamic rage under ultra-low illumination, the noise level should be low and full-well value must be high with high quantum efficiency. Full-well value defines the maximum amount of charge (photons) an individual pixel can hold before saturating. Low noise is achieved with photogate CMOS active pixel sensor (APS) but at the cost of greater reduced quantum efficiency and reduction in full-well.

[0020] Low noise can be also achieved with pinned-photodiode (PPD) APS but PPD APS has very low full-well and a poor quantum efficiency. Furthermore, it requires complicated processing and has great difficulty operating under advanced (deep sub-micron) process. Photodiode APS is most suited for advance sub-micron process. However, photodiode APS has noise value high when achieving a high full-well. This is not suited for a high quality imaging.

[0021] In order to capture scenes with high intra-scene contrast under a low-illumination with large range of illumination in the final image, the signal-to-ratio at low-light level of a CMOS imager has to be maximized along with the increase in the saturation signal level value of the CMOS imager. The signal-to-ratio at low-light level (SNR) is governed by the following equation:

[0022] where QE is the quantum efficiency and R_N is the read noise. The quantum efficiency, QE, is defined as the ratio between the number of generated electrons and the number of impinging photons and the read noise, R_N, is obtained by the root mean square (RMS) value of consecutive samples of the output voltage for one pixel. Thus, to achieve high quality imaging at low-light-level, quantum efficiency (QE) has to be increased simultaneously with a reduction in read noise (R_N).

[0023] On the other hand, to achieve imaging with high intra-scene contracts, the saturation signal level (full-well) value of the CMOS imager has to be increased. Currently, photodiode-type CMOS imager do not allow this. For a typical CMOS imager, the read noise level, R_N, remains high between 25 electrons and 70 electron and the full-well value only about 70,000 electrons at a very moderate low read noise level of 25 electrons. However, the full-well can be increased to 1,000,000 electrons, but this would increased the noise to about 100 electrons. The increased value of noise when full-well is increased in unacceptable to achieve high quality imaging. The main reason for this increase in noise is that photodiode active pixel read noise is governed by the sense node reset noise. Sense node reset noise can be expressed the following equation:

[0024] where Q_noise is the uncertainty on the charge stored on the capacitor, k is Boltzman's Constant, T is the absolute temperature and C_D is the sense node capacitance value. Thus, to achieve a low noise level, C_D value should be low. But reducing the value of C_D to a lower level would reduce the full-well value.

[0025] Hence reducing read noise by reducing C_D is a conflict with achieving large full-well which demands a large C_D value. This conflicting requirement on the sense node capacitance size, C_D, is one of the main limitations of a CMOS imager in simultaneously achieving large full well and low capacitance. All the above discussed technical limitations make a CMOS imager impossible to achieve high intra-scene contrast under a low-illumination with a large range of illumination in the final image.

[0026] The present invention provides a CMOS imager with a column feedback soft-reset scheme, by which a CMOS imager generates a sub-kTC noise so that read noise does not depend on the sense node capacitance. By using a column feedback circuit, reset noise can be suppressed to a negligible amount, so that a CMOS imager circuit can achieve noise performance to very efficient low noise level. This scheme allows increasing sense node capacitance without an associated noise penalty to it. This scheme provides a method to achieve a large full-well value without sacrificing read-noise performance. The feedback circuit is located at the column side of the circuit which provides a design with a minimal change to the pixel, as a result quantum efficiency or pixel size is not compromised. The present invention allows a CMOS imager to capture image with high intra-scene contracts and with high dynamic range under low illumination.

[0027] FIG. 1 illustrates a CMOS imager array.

[0028] FIG. 2 is an illustration of a schematic circuit diagram of a pixel of a conventional CMOS image sensor with a photodiode as a radiation-sensitive element.

[0029] FIG. 3 is an illustration of a schematic circuit diagram of an active pixel of a conventional CMOS image sensor operating under ultra-low illumination.

[0030] FIG. 4 is an illustration of a flow diagram of a CMOS imager capturing image under ultra-low illumination.

[0031] FIG. 5 is an illustration of circuit diagram of one of the embodiment of the present invention.

[0032] FIG. 6 is an illustration of a flow diagram of a CMOS imager with column feedback soft-reset scheme capturing image under ultra-low illumination.

[0033] FIG. 7 is an illustration of circuit diagram of one of the embodiment of the present invention with a R-C Slew circuit.

[0034] FIG. 8 is an illustration of circuit diagram of one of the embodiment of the present invention with a FET Slew circuit.

[0035] The present invention is a design of a photodiode CMOS imager with a column feedback soft-reset for imaging under ultra-low illumination and with high dynamic range. In the following description, numerous specific details are set forth to provide a more thorough description of embodiments of the invention. It is apparent, however, to one skilled in the art, that the invention may be practiced without these specific details. In other instances, well known features have not been described in detail so as not to obscure the invention.

[0036] The present invention incorporates a column feedback circuit which provides a scheme for imaging under ultra low illumination and with high dynamic range. The present invention provides a CMOS imager to capture scenes with high intra-scene contrast under a low-illumination with large range of illumination. This pattern as well as the individual active element are generated by a CMOS manufacturing process techniques.

[0037] FIG. 5 is an illustration of a schematic circuit diagram of one of the embodiment of the present invention having an active pixel of a CMOS image sensor with column feedback soft-reset operating under ultra-low illumination. The radiation-sensitive element in an active pixel, 500, is a photodiode, 520, comprising a p-n junction, wherein the p-doped junction side, 510, is connected to a negative bias voltage V_sub at 505. The n-doped junction side 530a is connected via 530 to the drain terminal, 530c, of a reset transistor, Mrst 540, and to the gate terminal, 530b, of a charge sensing transistor, Mst 560. The source terminal, 545, of a reset transistor, Mrst 540, and the source terminal, 565, of a charge sensing transistor, Mst 560, is biased to a positive reference bias voltage V_DD at 575.

[0038] The gate terminal, 547, of the reset transistor, Mrst 540, is connected to the drain terminal, 549, of a soft reset transistor, Msw 546. The gate terminal, 550, of a soft reset transistor, Msw 546, is connected to the common row reset line at 551. The drain terminal of a charge sensing transistor, Mst 560, is connected to the source terminal of a row select transistor, Msel 568 at 567. A row select transistor, Msel 568, has its gate terminal, 570, connected to a row select signal line, 571, and its drain terminal, 569, connected to a column signal bus, 574.

[0039] The feedback circuit has an operational amplifier (e.g. comparator), 581, which has a reference bus voltage line (V_ref), 585, connected to its positive input terminal (non-inverting terminal), 580, and column signal bus line, 574, connected to its negative input terminal (inverting terminal), 582. A comparator is a circuit which compares a signal voltage applied at one input of an operational amplifier with a known reference voltage at the other input. A comparator, 581, detects the presence in the signal in a column bus line, 574, and changes its output with reference to the voltage in the a reference bus voltage line (V_ref).

[0040] The output terminal, 584, of an operational amplifier, 581, is connected to the feedback bus line, 555. An amplifier of large gain is used as an operational amplifier (comparator, 581). The feedback bus, 555, has a two-pole switch, 590, connected to it. One terminal of the switch, 590, is biased to a ground, 505, and the other terminal, 591 is connected to the feedback bus, 555. The switch provides a scheme for manual reset of a pixel.

[0041] A pixel, 500, consists of the three conventional transistors (like FET): Mrst 540, Mst 560 and Msel 568. Unlike a conventional pixel, the present invention has an additional transistor Msw 546, a operational amplifier 581 and an additional line (a feedback bus line, 555) for providing the feedback to pixel, 500. The column feedback line is connected to the source terminal, 548, of the soft reset transistor, Msw 546. When the reset signal, 551, is at high level, the soft reset transistor, Msw 546 conducts the feedback signal. When the reset signal, 551, is at low level, the soft reset transistor, Msw 546 does not conducts the feedback signal. The power bus (V_DD), 575, is pulsed signal which provides imaging with zero-lag.

[0042] A photodiode, 520, in a CMOS imager is exposed to a optical radiation, 502. The exposure to the optical radiation, 502, of a photodiode, 520, would generate a charge flow in the photodiode. The generated charges would flow from n-doped side, 530a, to the gate terminal, 530b, of a sensing transistor, 560. The positive voltage at the gate terminal, 530b, would make a sensing transistor, 560, to conduct a reference bias voltage V_DD, 575. When a sensing transistor, 560, conducts, the charges are passed to the source terminal, 567, of a row select transistor, 568. The row select line, 571, would have a positive voltage when a particular row is selected to sense. When a row select line, 571, has a positive voltage, the gate terminal, 570, of a row select transistor, 568, is turned on to make a row select transistor, 568, to conduct. When a row select transistor, 568, to conducts, the charges from the source terminal, 567, are passed on to the drain terminal, 569, of a row select transistor, 568, and finally can be read-out by the column bus line, 574.

[0043] A column feedback operational amplifier, 581, has an inverting terminal, 582, and a non-inverting terminal, 580, and connected to column bus line, 574, and a reference bus voltage line, 585, respectively. The output of an operational amplifier is fed back to the source terminal, 548, of a soft reset transistor, 546. By applying a positive reset voltage to the gate terminal, 550, from row reset line 551, would make the soft-reset transistor, 546, conduct the feed-back voltage from the column feedback line, 555.

[0044] When a soft reset transistor, 546, conducts, it passes a voltage to the gate terminal, 547, of a reset transistor, 540. A reset transistor, 546 conducts a positive reference bias voltage V_DD, 575, when the gate terminal, 547, is high (i.e., has a positive voltage). The scheme of achieving the conduction of the reset transistor, 540, is called soft-reset. This process eliminates the noise associated with the capacitance of the pixel circuit. An in-depth analysis of various electrical properties of one of the embodiments of the present invention is discussed below.

[0045] Low-Noise Photodiode Pixel Using Column-Feedback Reset:

[0046] FIG. 3 shows the pixel schematic of a conventional photodiode APS. Noise in photodiode-type CMOS active pixel sensor (APS) is primarily due to the reset noise (kTC) present at the node 330a which is also called a sense node (in FIG. 5, the sense node is 530a). A signal integrated on a photodiode sense node is calculated by measuring difference between the voltage on the column bus, 390, before and after the row reset, 350, is pulsed. Uncertainly in the number of electrons at the sense node following a reset generates a reset noise. The reset noise (N) is governed by the following equation: $\begin{array}{cc}{N}^2=\frac{{\mathrm{kTC}}_{\mathrm{SENSE}}}{q^2}& \left(3\right)\end{array}$

[0047] Where C_SENSE is the sense node capacitance, q is the electronic charge, k is the Boltzmann constant and T is the temperature in degree Kelvin.

[0048] A noise value lower than kTC noise can be achieved with a photodiode-type pixel by employing soft-reset technique. Soft reset refers to resetting with both drain and gate of the n-channel reset transistor kept at the same potential, as is obtained if both V_DD, 575, and the RST, 551, are kept at the maximum permitted voltage for a given CMOS technology (e.g. 3.3V in a 0.35 m technology). Under such conditions, sense node can charge up approximately to V_DDV_T, where V_T is the threshold voltage. In fact, as long as reset line voltage 551 remains high, the sense node continues to charge up under sub-threshold MOSFET current flow, causing the node to rise in a logarithmic fashion.

[0049] The current flow in the rest FET during Soft-reset: $\begin{array}{cc}I=\exp \left[\frac{q\left(\mathrm{Vrst}-\mathrm{Vsense}-\mathrm{Vm}\right)}{\mathrm{mkT}}\right]=\exp \left[\left(\mathrm{Vrst}-\mathrm{Vsense}\right)\right]& \left(4\right)\end{array}$ $\begin{array}{cc}\mathrm{where}=\left[\frac{q}{\mathrm{mkT}}\right]& \left(5\right)\end{array}$

[0050] By implementation of the present invention, sub-kTC reset noise can be achieved under soft-reset. As long as the sense node swing is large enough (q>>1), noise at the sense is governed by the following equation: $\begin{array}{cc}\begin{array}{ccc}{N}^2=\left[\frac{\mathrm{Csense}}{2}\right]& & \mathrm{when}q>>1\end{array}& \left(6\right)\end{array}$

[0051] Thus, if the feedback factor () can be made large, the read noise can be made very small. Union conventional soft-reset, is relatively small, since the minimum value for the non-ideality factory (m) is approximately equal to the value 1. Hence, only about a factor of 2 reduction is rest noise is possible.

[0052] The feedback factor () can be increased by using circuit feedback, FIG. 5 shows the schematic of a photodiode pixel that suppresses the reset noise arbitrarily beyond the conventional kTC value. The reset noise suppression is achieved by resetting the pixel, 500, in soft-reset with the reset level determined by column-feedback, 555.

[0053] The feedback is provided by the column-feedback amplifier, 581, by minimizing the error between the voltage on the column bus, 574, and the reference-bus, V_ref 585. The reference bus is fixed D.C. level and is common to all columns in an APS array. Under feedback, the gate of Mrst, 540, is continuously adjusted so that the pixel output reaches the voltage level set by V_ref, 585, and the pixel is reset under soft-reset. As the gain of a column amplifier, 581, is large, the feedback factor (f) during soft-reset can be arbitrarily high and provide reset noise suppression.

[0054] The feedback factor for the circuit can be determined by the following current equation for the Mrst current:

[0055] where f(1A), and A is the gain of the column amplifier, 581. From the equation (1), it can be seen that by using a column-feedback, the reset noise is reduced to: $\begin{array}{cc}\begin{array}{ccc}{N}^2=\left[\frac{\mathrm{Csense}}{2\left(1+A\right)}\right]& & \mathrm{for}q>>1\end{array}& \left(8\right)\end{array}$

[0056] Thus, through an appropriate choice of column-amplifier gain A, reset noise can be suppressed to a negligible amount. By using a column-amplifier, 581, of a gain of 100 would reduce the noise by a factor of 10, which is more that sufficient for low-noise applications. The noise from the column-amplifier can also be made negligible by increasing the feedback bus capacitance.

[0057] Furthermore, low-noise no longer requires small sense node capacitance (Csense), since the same noise value can be achieved by keeping the ratio of Csense to amplifier gain (A) constant. This is extremely important, since large full-well requires large sense node capacitance. Thus, the scheme presented here simultaneously achieves low noise and large full-well, and hence removes one of the most critical limitations on CMOS photodiode APS performance.

[0058] High gain amplifiers usually require complicated circuits for implementation, and as such, cannot be integrated in a small pixel. However, in the scheme presented here, the high gain amplifier is at the bottom of the column. Only one additional transistor (typically a FET) and an additional line is needed in the pixel. As a result, reset noise is suppressed without sacrificing pixel fill-factor and quantum efficiency or pixel size. Thus, the noise suppression scheme is entirely compatible with realistic and advanced imager design. The present invention provides a scheme which simultaneously supports high QE, low noise, small pixel size, and large full-well, enabling high performance imaging in CMOS technology.

[0059] FIG. 6 is an illustration of a process of the present invention which provides a CMOS imager, 602, with a soft-reset scheme, by which a CMOS imager generates a sub-kTC noise so that read noise does not depend on the sense node capacitance. By using a column feedback circuit, rest noise can be suppressed to a negligible amount so that photogate APS or CCD-like circuits can achieve noise performance to very efficient value (low noise level). This scheme allows increasing sense node capacitance without an associated noise penalty to it. This scheme provides a method to achieve a large full-well value without sacrificing read-noise performance. As the result the original image, 601, is captured under low illumination to a digital image form, 603, with very high intra-scene contract and high dynamic range.

[0060] The feedback circuit in one of the embodiment of the present invention is located at the column side of the circuit. In FIG. 5, an operational amplifier, 581, and a soft reset transistor, 546, are located outside a pixel, 500. The design in which the feedback circuit is placed in one of the embodiment of the present invention provides a scheme to implement a CMOS imager with a column feedback soft-reset circuit without any changing the pixel area. This design makes it all the more practical for manufacturing a CMOS device without complicating the pixel design. As the feedback circuit does not need a change in pixel area or interfere with the pixel area, a pixel of a greater quantum efficiency is produced using existing pixel design. A CMOS imager with column feedback soft-reset circuit yields larger full-well and greater quantum efficiency with no compromise in pixel area.

[0061] The present invention, a photodiode CMOS imager with column feedback, provides low-noise, high QE and high full-well imaging. By employing the present invention, a CMOS imager can achieve read noise reduced down to less than 5 electrons and provide an excellent low-light-detection capability. A full-well can be also achieved as large as 1 million electrons, providing over 105 dB dynamic range and increasing quantum efficiency as high as about 60% resulting in a higher performance imaging.

[0062] Other Designs

[0063] One of the embodiments of the present invention has a R-C circuit in addition to a feed back amplifier and soft-reset. FIG. 7 is an illustration of a schematic diagram of one of the embodiment of the present invention with column-feedback and R-C circuit providing a ramp for the reference level bus.

[0064] The radiation-sensitive element in an active pixel is a photo capacitor, 703. A photo capacitor, 703 has one end, 704, connected to a negative bias voltage V_sub (ground voltage) at 702 and the other end 705a is connected via 705 to the drain terminal, 705c, of a reset transistor, Mrst 707, and to the gate terminal, 705b, of a charge sensing transistor, Mst 720.

[0065] The source terminal, 709, of a reset transistor, Mrst 707, and the source terminal, 722, of a charge sensing transistor, Mst 720, is biased to a positive reference bias voltage V_hts at 710. The voltage V_hts, 710, is depends upon the selection of hard or soft transistors. A transistor which is responsible the hard picture quality is a N-MOS transistor, 738, and a transistor which is responsible for soft picture quality is a P-MOS transistor, 732.

[0066] The gate terminal, 745, of a hard N-MOS transistor, 738, is connected to a hard signal line, 747. The gate terminal, 735, of a soft P-MOS transistor, 732, is connected to a soft signal line, 736. The source terminals of both N-MOS transistor, 738, and P-MOS transistor, 732, are connected to a reference bias voltage power bus (V_DD, 700) at 700a and 700b respectively. The drain terminal, 740, of N-MOS transistor, 738, and the drain terminal, 733, of P-MOS transistor, 732, are connected together at a node, 710a. The node 710a is connected to 710.

[0067] The N-MOS transistor, 738, conducts when the signal value in hard signal line, 747, is high level (e.g., a positive voltage value). When the N-MOS transistor, 747, conducts a reference bias voltage, V_DD 700, a positive voltage value is passed to the node 710a via the drain terminal, 740. The P-MOS transistor, 732, conducts when the signal value in soft signal line, 736, is low level (e.g., a zero voltage value). When the P-MOS transistor, 732, conducts a reference bias voltage, V_DD 700, and a positive voltage value is passed to the 710a via the drain terminal, 733. The voltage at the nodes 710a and 710 is V_hts, which depends upon the selection of signal at the gate terminal of N-MOS transistor, 738, and P-MOS transistor, 732.

[0068] The gate terminal, 711a, of the reset transistor, Mrst 707, is connected to the drain terminal, 711b, of a soft reset transistor, Msw 712. The gate terminal, 715, of a soft reset transistor, Msw 712, is connected to the common row reset line at 718. The drain terminal of a charge sensing transistor, Mst 720, is connected to the source terminal of a row select transistor, Msel 726 at 724. A row select transistor, Msel 726, has its gate terminal, 729, connected to a reference bias voltage, V_DD 700, at 700c, and its drain terminal, 727, connected to a column signal bus, 728.

[0069] A load transistor, 776, is connected to the column signal bus, 728. A current supply source, 728, provides the gate terminal, 780, of a load transistor, 776, a positive pulse for conduction. The source terminal, 779, is connected to the column signal bus, 728. The transistor, 776, serves as a load to an active pixel. A capacitor, C2 772, is connected between the inverting terminal, 751, of a operational amplifier, 750, and to a negative bias voltage (ground) at 702b. The capacitor, 772, charges when there is a positive signal in the column bus line, 728 and discharges when there is no signal in the in the column bus line, 728. By controlling the capacitance value of the capacitor, C2 772, the noise level is reduced along with a higher full-well feed back.

[0070] The feedback circuit has an operational amplifier, 750, which has a reference bus voltage line (V_ref), 703, connected to its positive input terminal (non-inverting terminal), 752, and column signal bus line, 728, connected to its negative input terminal (inverting terminal), 751. The output terminal, 754a, of an operational amplifier, 750, is connected to a pass transistor, 756. A pass transistor, 756, has the output of the operational amplifier, 750, and a feedback pulse is supplied to the input terminals at 754a and 755 respectively. The output signal of a pass transistor is a product of the two inputs. Hence, the pass transistor output will have a positive signal only when both the feedback pulse, 755, and the operational amplifier, 754a, are positive signal.

[0071] A resistance-capacitance slew circuit is provided as a reference voltage source. A resistance-capacitance slew circuit provides a reference bias voltage which slowly increase to a saturation point and falls back to a zero or a low voltage level. The time rate at which it increases, also known as ramp up, is slower than the feedback closed-loop response time in one of the embodiments of the present invention. A reset transistor, 707, is maintained to conduct in closed loop feedback during the entire process providing a further reduction in noise. Without the ramp on the reference voltage, such closed loop feedback is not maintained.

[0072] A resistance, 793, connected to a positive voltage source, 792b, at 794 and to non inverting terminal, 752, via 703b. A capacitor, 796, is connected between the resistance, 793, at 703b and to a negative bias voltage, 702 at 702e. A positive voltage source, 792b, generates a pulse signal. The capacitor, 796, charges when the voltage at 792b is a positive voltage level and it discharges when the voltage at 792b is zero.

[0073] A N-MOS transistor, 788, is connected to the voltage reference bus line, 703, at 703a to act as a switch to provide a by-pass path for the current from discharge of the capacitor, 796. A N-MOS transistor, 788, conducts when a positive signal is supplied at the gate terminal, 790 from a pulse line 755. When the N-MOS transistor, 788, conducts it provides a low resistance path for the R-Slew circuit, 799. As current always flows through a low resistance path, the charges are not passed to the feedback amplifier, 750.

[0074] A pixel consists of the three conventional transistors (like FET): Mrst 707, Mst 720 and Msel 726. One embodiment of the present invention has an additional transistor Msw 712, a operational amplifier, 750, a R-C Slew circuit, contrast control transistors (738 & 732), pass transistor, 756, bias transistor (776, 782,788 and 765) and an additional feedback bus line, 701. The power bus (V_DD), 700, is pulsed signal which provides imaging with zero-lag.

[0075] FIG. 8 is an illustration of one of the embodiment of the present invention with a FET providing the ramp scheme. The ramp circuit consist of a N-MOS transistor, 893, and a capacitor, 896. The source terminal, 894, of a N-MOS transistor, 893, is connected to a voltage source, 892b. The drain terminal, 895, of a N-MOS transistor, 893, is connected to a capacitor, 896, via voltage reference bus at 803b. The other input terminal, 898, of the capacitor, 896 is connected to a negative bias voltage, 802, at 802e.

[0076] The gate terminal, 891, of a N-MOS transistor, 893, is connected to a voltage source, 891a. When the signal at the gate terminal, 891, is high (positive value) the capacitor, 896, gets charged. When the voltage at 803b is a zero voltage level and it gets discharged. The time rate of increase of the signal generated when the capacitor, 896, is slower than the feedback closed-loop response time in one of the embodiments of the present invention.

[0077] A reset transistor, 807, is maintained to conduct in closed loop feedback during the entire process providing a further reduction in noise. Without the ramp on the reference voltage, such closed loop feedback is not maintained.

[0078] Thus, a method and apparatus for a photodiode CMOS imager with column feedback soft-reset for imaging under ultra-low illumination and with high dynamic range is described in conjunction with one or more specific embodiments. Although the present invention has been described in considerable detail with regard to the preferred versions thereof, other versions are possible. The invention is defined by the claims and their full scope of equivalents.