Photodiode active pixel sensor with shared reset signal row select

[0002] This invention relates to the field of solid state photo-sensors and imagers referred to as Active Pixel Sensors (APS) that have active circuit elements associated with each pixel, and more specifically to Solid State Imagers that employ photodiode type photodetectors 4 transistor pixels and correlated double sampling (CDS).

[0003] APS are solid state imagers wherein each pixel contains the typical solid state pixel elements including a photo-sensing means, reset means, a charge to voltage conversion means, and additionally all or part of an amplifier. The photocharge collected within the pixel is converted to a corresponding voltage or current within the pixel as discussed in prior art documents such as “Active Pixel Sensors: Are CCD's Dinosaurs?”, SPIE Vol. 1900-08-8194-1133 July 1993, by Eric Fossum. APS devices have been operated in a manner where each line or row of the imager is selected and then read out using a column select signal as discussed by E. Fossum in “Active Pixel Sensors: Are CCD's Dinosaurs?”, SPIE Vol. 1900-08-8194-1133 July 1993 and by R. H. Nixon, S. E. Kemeny, C. O. Staller, and E. R. Fossum, in “128×128 CMOS Photodiode-type Active Pixel Sensor with On-chip Timing, Control and Signal Chain Electronics”. Proceedings of the SPIE vol. 2415, Charge-Coupled Devices and Solid-State Optical Sensors V, paper 34 (1995). The selection of rows and columns within an Active Pixel Sensor is analogous to the selection of words and bits in memory devices. Here, the selection of an entire row would be analogous to selecting a word and the reading out of one of the columns of the Active Pixel Sensor would be analogous to selecting or enabling a single bit line within that word. Conventional prior art photodiode devices teach architectures employing 4 transistor designs, where the 4 transistors (4T) are typically the Transfer, Row Select, Reset, and Source Follower Amplifier transistors. While this architecture provides the advantages of yielding APS devices having the capability to easily perform CDS and provide low readout noise, these 4T pixels suffer from low fill factor. Fill factor is the percentage of pixel area that is devoted to the photosensor. Since these 4 transistors and their associated contact regions and signal busses are placed in each pixel, and since contact regions typically consume a large amount of pixel area due to the required overlap and spacings of various layers, the fill factor for the pixel is reduced because of the large area consumed that could otherwise be used for the photodetector. Connection to each of these components to the appropriate timing signal is done by metal busses that traverse the entire row of pixels. These metal busses are optically opaque and can occlude regions of the photodetector in order to fit them into the pixel pitch. This also reduces the fill factor of the pixel. Decreasing the fill factor reduces the sensitivity and saturation signal of the sensor. This adversely affects the photographic speed and dynamic range of the sensor, performance measures that are critical to obtaining good image quality.

[0004] Prior art devices employing 3 transistor (3T) based pixels have a higher fill factor than 4T pixels, but these 3T pixels cannot easily perform CDS. Sensors that perform CDS employing 3 transistor based pixels, typically first read out and store an image frame comprising a reset level for each pixel on the sensor. Next the signal frame is captured and read out. The reset level frame stored in memory must then be subtracted from the signal frame at each pixel to provide a pixel signal level that is referenced to the pixel reset level prior to integration. This requires an extra frame of memory in the imaging system, and an extra step in the digital signal processing chain, thus adversely affecting the speed, size and cost of the system.

[0005] A typical prior art Photodiode APS pixel is shown in FIG. 1. The pixel in FIG. 1 is a prior art 4 transistor pixel that comprises: a photodiode (PD), and transfer transistor (TG); floating diffusion (FD); reset transistor with a reset gate (RG); row select transistor with a row select gate, (RSG); a source follower input signal transistor (SIG); a row select signal buss (RSSB); a reset gate signal buss (RGSB), and a transfer gate signal buss (TGSB). 2 adjacent pixels are shown, each containing identical but separate transistors and row control signal busses for RG, TG and RSG. As stated above these 4 transistor pixels provide low readout noise with CDS by inclusion of an extra transistor per pixel. However the area required to implement the 4^thtransistor reduces the fill factor of the pixel compared to the 3 transistor pixel.

[0006] It should be readily apparent that there remains a need within the art to provide an alternate pixel architecture that has higher fill factor, and the capability to perform CDS without the need to capture and store entire frames of image data.

[0007] The present invention provides a high fill factor Photodiode Active Pixel Architecture with the capability to perform Correlated Double Sampling, (CDS). The functionality of a 4 transistor pixel is maintained while eliminating the separate row select transistor. This is done by sharing the RG control signals in one row with the row select means of an adjacent row.

[0008] The present invention provides a Photodiode Active Pixel sensor with true Correlated Double Sampling (CDS), using only 3 transistors resulting in a higher fill factor. The advantage gained is high fill factor, and lower temporal noise. No disadvantages are foreseen.

[0009] FIG. 1 is a schematic diagram of two adjacent pixels of prior art architecture for a four transistor Photodiode Active Pixel Sensor pixel;

[0010] FIG. 2 is a schematic diagram of two adjacent pixels of a new Photodiode pixel architecture for an Active Pixel Sensor;

[0011] FIG. 3 is a timing diagram illustrating the operation of the pixels shown in FIG. 2;

[0012] FIG. 4 is a top view layout of the pixel shown in FIG. 2.

[0013] FIG. 2 is a schematic illustration of the 3 transistor photodiode pixel architecture for an Active Pixel Sensor (APS) as envisioned by the present invention. The preferred embodiment shown in FIG. 2 illustrates the best mode known to inventor. Other physical embodiments are realizable that are obvious variations of the embodiment shown in FIG. 2, as will be discussed further below. The pixels 10 shown in FIG. 2 are single pixels within an array of pixels having numerous rows and columns. The two row adjacent pixels 10 are shown in FIG. 2, to indicate how control the signal lines in one row are used for another purpose in an adjacent row. For the purpose of illustration of timing and operation, FIG. 2 also contains an example of per column analog signal processing used to read out the sensor. The per column signal processing referred to is the correlated double sampling that is performed under control of the Sample and Hold Signal (SHS) and Sample and Hold Reset (SHR) transistors that activate that sampling of the pixel signal voltage by the Signal Capacitor Cs and the pixel reset voltage by Reset Capacitor Cr.

[0014] As seen in FIG. 2, the pixel 10 comprises: a photodiode photodetector 12 (PD); transfer gate (TG) 23; floating diffusion (FD) 25; a reset transistor 14 with a reset gate 15 (RG); a reset drain 18 for the reset transistor 14; a source follower input signal transistor 21 (SIG); a transfer gate signal buss (TGSB) 26; and a reset gate signal buss (RGSB) 27. The row select and the row select signal buss of the prior art pixel shown in FIG. 1 is eliminated, and the row select process is accomplished in a novel manner discussed hereinbelow. As shown in FIG. 2, the reset gate signal buss (RGSB) 27 of any given row, referred to herein as Row_i, is connected to the reset drains 18 of the next row in the readout sequence, referred to herein as Row_i+1. The transistor, gate, and signal buss components within the 2 pixels have also been annotated with a subscript to indicate the row that they reside in.

[0015] Referring to FIG. 3, which is a timing diagram used to describe the operation of the 3 transistor pixels 10 shown in FIGS. 2, the operation of the new pixel architecture for FIGS. 2 will be described. It should also be understood that the pixel operation described takes place for an entire row of pixels 10 in the context of the standard per line rolling shutter operation of CMOS active pixel sensor devices. The diagram shows the timing sequence for the readout of 3 rows within the sensor; Row_i−1, Row_i, and Row_i+1. Each of the signals in the timing diagram have been annotated with a subscript indicating the row that they are associated with. It should be understood that the levels in the timing diagram are relative and can be set to be any predetermined signal level to provide optimized sensor performance.

[0016] The timing diagram begins at a point in time indicated as τ₀, where the integration time for Row_i−1has elapsed and the readout of Row_i−1commences. Row_i−2has been deselected, RG_i−2and RG_i−1are both high at time τ₁to reset the floating diffusions in Row_i−1(FD_i−1) to ˜VDD, or some other appropriate predetermined potential. The reset potential now turns on the source follower input transistors in Row_i−1(SIG_i−1), effectively selecting that row for readout of the reset level of the floating diffusion which is sampled and held on Cr by pulsing SHR at time τ₂. The transfer gates for Row_i−1(Tg_i−1) are pulsed on at time τ₃to transfer the signal charge from the photodiodes in Row_i−1onto the floating diffusions (FD_i−1) of Row_i−1. The signal levels of the floating diffusions in Row_i−1(FD_i−1) are still at a level that effectively selects Row_i−1for readout and the signal levels of these floating diffusions (FD_i−1) of Row_i−1are then sampled and held on Cs by pulsing SHS at time τ₄. Each column in Row_i−1is then readout differentially through the per column difference amplifier 32 in the period indicated in the timing diagram by “Readout Row_i−1” at time τ₅.

[0017] Following the readout of Row_i−1, Row_i−1is deselected by pulsing RG_i−1on while leaving RG_i−1low at time τ₆. This sets the voltage of the floating diffusions for Row_i−1(FD_i−1) to a level that shuts off the source follower transistors for Row_i−1(SIG_i−1), effectively disconnecting the source followers for Row_i−1from the output signal column buss, thus deselecting Row_i−1. Next Row_iis selected by pulsing on RG_iwhile RG_i−1is held high at time τ₇. It should be noted that the re-pulsing of RG_i−1at time τ₇simply resets the floating diffusions for the previous row that was read (FD_i−1) to a low voltage, keeping the source follower signal transistors for that row (SIG_i−1) in a non-enabled state, effectively disconnecting the floating diffusions for the row that was previously read (FD_i−1) and prevents it from interfering with the output of the current row being read, in this case ROW_i. The reset level of the floating diffusions for ROW_i(Fd_i), are sampled and held by pulsing SHR at time τ₈. TG_iis then pulsed “high” to transfer the signal charge from the photodiodes for ROW_i(Pd_i) to the floating diffusions for ROW_i(FD_i) at time τ₉, and the signal level is sampled and held by pulsing SHS at time τ₁₀. Readout of Row_iis then completed similarly to Row_i−1.

[0018] This process described for ROW_i−1and ROW_iis repeated for Row_i+1. Row_iis deselected by setting the FD_ito a low voltage level at τ₁₁by pulsing RG_i“high” while RG_i−1is held low disabling the source follower transistor for ROW_i. Row_i+1is reset, which operates to select Row_i+1at time τ₁₂because Rg_iis “high” at that time and Row_i+1is effectively selected with all other rows deselected; the reset level is sampled and held at time τ₁₃by activating SHR; the signal level is transferred from PD_i+1to FD_i+1at time τ₁₄by TG_i+1; the signal level is then sampled and held at τ₁₅; and the Row_i+1readout occurs at time τ₁₆as previously described.

[0019] This process repeats for all rows of the sensor until the entire frame or sequence of frame readouts is complete. It should be noted that the row select and deselect process could also be combined by eliminating the first of the 2 RG pulses and having enough overlap of the adjacent RG signals as indicated in FIG. 3. The timing illustrated by FIG. 3 is that which the inventor believes that best mode operating the invention.

[0020] From the description of operation provided it is evident that this new 3 transistor photodiode pixel architecture provides true CDS output signal without the need for storing a reset frame.

[0021] The typical prior art Photodiode APS pixel shown in FIG. 1 comprised 4 transistors, and 3 row control signal busses. By comparison the new pixel architecture in FIG. 2 comprises only 3 transistors and 2 row control signal busses. The elimination of 1 transistor and buss and the associated contact regions provides more pixel area that can be allocated to the photodetector. This provides substantially higher pixel fill factor and consequently higher optical sensitivity.

[0022] FIG. 4 is a top view of the new pixel architecture as discussed above shown to further illustrate the concept of the present invention. In FIG. 4 the gate of the reset transistor in ROW_i−1(RG_i−1) is the reset drain for the reset transistor of the next row to be read out in sequence, ROW_i.

[0023] The foregoing description details the embodiments most preferred by the inventor. Variations of these embodiments will be readily apparent to those skilled in the art. Accordingly, the scope of the invention should be measured by the appended claims.

[0024] 10 pixel

[0025] 12 photodiode

[0026] 14 reset transistor

[0027] 15 reset gate

[0028] 18 reset drain

[0029] 21 source follower input transistor

[0030] 23 transfer gate

[0031] 25 floating diffusion

[0032] 26 transfer gate signal bus

[0033] 27 row select and photogate signal buss

[0034] 32 difference amplifier

[0035] SHS sample and hold signal transistor

[0036] CS signal storage capacitor

[0037] SHR sample and hold reset transistor

[0038] CR reset storage capacitor