JP2016129397A - Imaging apparatus and imaging system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To achieve a high bit, the enlargement of a dynamic range, and/or reduction in quantization error.SOLUTION: An AD conversion part compares the output signal of an amplifier circuit 20-1 after a pixel is reset with a reference signal which is from a lamp signal generation circuit 25 and changed with respect to time, to output a first digital value, sets the gain of the amplifier circuit to a first gain, when the output signal of the amplifier circuit in the non-reset state of the pixel is larger than the reference signal and sets the gain of the amplifier circuit to a second gain larger than the first gain, when the output signal of the amplifier circuit in the non-reset state of the pixel is smaller than the reference signal, and further, compares the output signal of the amplifier circuit in the non-reset state of the pixel with the reference signal changed with respect to the time, after the gain of the amplifier circuit is set to the first gain or the second gain, to output a second digital value. When the resolution of the first digital value is different from that of the second digital value, a correction part 40-1 corrects a difference between the resolutions.SELECTED DRAWING: Figure 4

Description

  The present invention relates to an imaging apparatus and an imaging system.

  In the image sensor element, different gain processing is performed on the pixel signal, AD (analog-digital) conversion is performed, the signal is read from the memory, the gain is corrected, and the dynamic range is expanded by selecting the AD data according to the signal level. Technology to do is released. There exists patent document 1 in the technique.

  Further, in the current intra-element AD conversion technique, it is common to perform differential processing on the AD data of the reset noise of the comparator from the AD data of the pixel signal having the same resolution.

JP 2008-124842 A

  The reset noise AD data has a problem in AD conversion accuracy due to the reset noise of the comparator. Patent Document 1 does not describe correction of reset noise in an AD converter with different gain signals.

  An object of the present invention is to provide an imaging apparatus and an imaging system that can reduce noise.

  The imaging apparatus of the present invention includes a pixel that generates a signal by photoelectric conversion, an amplification circuit that amplifies the signal of the pixel with a set gain, an AD conversion unit, and a correction unit, and the AD conversion unit Compares the output signal of the amplifier circuit after resetting the pixel with a reference signal that changes with time, and outputs a first digital value, and the output signal of the amplifier circuit in a non-reset state of the pixel Is larger than the reference signal, the gain of the amplification circuit is set to the first gain, and when the output signal of the amplification circuit in the non-reset state of the pixel is smaller than the reference signal, the gain of the amplification circuit is set, After setting the gain of the amplification circuit to the first gain or the second gain, the gain in the non-reset state of the pixel is set to a second gain larger than the first gain. The output signal of the amplifier circuit and a reference signal that changes with time are compared to output a second digital value, and the correction unit has different resolutions of the first digital value and the second digital value In this case, the difference in resolution is corrected.

  Noise can be reduced.

It is a block diagram of the image pick-up element by the 1st Embodiment of this invention. It is a figure which shows the structural example of a pixel. It is explanatory drawing about the noise contained in a pixel signal, and the comparison method of a signal level and a reference voltage. It is a block diagram of an AD converter. FIG. 5 is a timing chart of the AD conversion unit in FIG. 4. It is explanatory drawing of the bit shift of a counter. It is a block diagram of the amplifier circuit of the 2nd Embodiment of this invention. It is a timing diagram of an image sensor. It is a block diagram of the AD conversion part of the 3rd Embodiment of this invention. FIG. 10 is a timing chart of the AD conversion unit in FIG. 9. It is explanatory drawing of a pixel signal. It is explanatory drawing of the bit shift of a counter. It is explanatory drawing of the amplifier gain at the time of imaging | photography, and a ramp signal. It is bit explanatory drawing at the time of high sensitivity photography. It is a block diagram of an imaging system.

(First embodiment)
FIG. 1 is a schematic configuration diagram of an image sensor 100 according to the first embodiment of the present invention. Reference numeral 100 denotes an imaging device called a CMOS image sensor, which photoelectrically converts a received subject image and outputs an electrical signal as a digital signal. The image sensor 100 includes a pixel unit 10, a vertical scanning circuit 15, an amplification unit 20, a ramp signal generation circuit (reference signal generation circuit) 25, a comparison unit 30, a counter unit 40, a memory unit 50, an output circuit 60, and a horizontal scanning circuit 65. And a timing generation circuit (TG) 70. The pixel unit 10 includes a plurality of pixels 10-1 arranged in a two-dimensional matrix. The pixel 10-1 outputs a pixel signal by photoelectric conversion. The vertical scanning circuit 15 outputs drive pulses X-1, X-2,. The amplifying unit 20 amplifies the pixel signal of the pixel unit 10 with the set gain. The ramp signal generation circuit 25 generates a ramp signal (reference signal) that changes with time as a comparison signal with the pixel signal. The comparison unit 30 compares the pixel signal amplified by the amplification unit 20 with the ramp signal. The counter unit 40 counts until the comparison unit 30 outputs the comparison result. The memory unit 50 holds count data of the counter unit 40. The horizontal scanning circuit 65 transfers data from the memory unit 50 to the output circuit 60 by horizontal scanning. The timing generation circuit 70 controls the timing of each circuit block.

  In the pixel unit 10, a plurality of pixels 10-1 are arranged on an area, and a configuration example thereof will be described later with reference to FIG. The row of each pixel 10-1 is sequentially driven by drive pulses X-1 and X-2 from the vertical scanning circuit 15, and signals output from each pixel 10-1 are transmitted through vertical signal lines V-1 to V-n. Then, it is guided to the amplifying unit 20. Each circuit is provided for each of the vertical signal lines V-1 to Vn from the amplifying unit 20 to the memory unit 50. Each circuit may be provided separately in the vertical direction of the vertical signal lines V-1 to Vn. For example, each circuit may be provided so that a signal is transmitted to the lower part of the pixel portion in the drawing for the pixels in the even column and to the upper portion of the pixel portion in the drawing for the pixels in the odd column. Each amplification circuit 20-1 of the amplification unit 20 may have only a function of amplifying a signal from the pixel 10-1, or may have a CDS processing function for performing noise reduction processing by correlated double sampling. . The CDS process may be performed at the input unit of the comparison unit 30.

  The comparison unit 30 includes a plurality of comparison circuits 30-1 corresponding to a plurality of pixel columns. First, the comparison circuit 30-1 compares the N signal from the amplification circuit 20-1 with the ramp signal from the ramp signal generation circuit 25. The N signal is a signal corresponding to the reset of the amplifier circuit 20-1 when the amplifier circuit 20-1 has the CDS function, and when the amplifier circuit 20-1 does not have the CDS function. , A signal corresponding to the reset of the pixel 10-1. The counter circuit 40-1 of the counter unit 40 counts down from the ramp signal ramp start until the magnitude relationship between the N signal and the ramp signal is reversed in the comparison circuit 30-1 with a count setting that corrects the gain difference. . Thereafter, the comparison circuit 30-1 compares the S signal from the amplification circuit 20-1 with the reference signal from the ramp signal generation circuit 25. The S signal is a signal corresponding to the non-reset state of the amplifier when the amplifier circuit 20-1 has the CDS function, and the pixel 10- when the amplifier circuit 20-1 does not have the CDS function. 1 is a signal based on photoelectric conversion by 1. The comparison circuit 30-1 determines or selects whether the gain of the amplification circuit 20-1 of all the pixel columns is low gain or high gain according to the comparison result, and the S signal and the ramp signal at the gain are Make a comparison. The counter circuit 40-1 has a function of correcting the gain difference. This correction method includes a method for correcting a high gain for a low gain and a method for correcting a low gain for a high gain. Details will be described later. Here, the former will be mainly described. When the gain of the amplifier circuit 20-1 is low, up-counting is performed without correcting the gain difference, and when the gain of the amplifier circuit 20-1 is high, up-counting is performed with the gain difference corrected. The count in which the gain difference is corrected is to perform counting by changing the increase / decrease of the count value with respect to the clock signal given to the counter circuit 40-1. In other words, when the gain difference from the AD conversion of the N signal is not corrected, the fluctuation range of the count value is the same as that of the conversion of the N signal, and when the gain difference is corrected, The fluctuation range of the count value is different. The memory circuit 50-1 of the memory unit 50 holds the count value (digital data) of the counter circuit 40-1. The digital data held in the memory circuit 50-1 is transferred to the output circuit 60 by the scanning pulse from the horizontal scanning circuit 65. The above-described counter unit 40 has been described with respect to the operation of the column counter method in which counting is performed by an up / down counter provided for each column. However, in the common counter system using a common counter in the plurality of comparison circuits 30-1, the counter signal is latched according to the comparison result of the comparison circuit, and the count data of the N signal and the count data of the S signal are individually stored. The gain difference may be corrected later. Thereafter, differential processing of the count data of the S signal and the N signal is performed.

  As described above, since the image sensor 100 corrects and counts the gain difference when comparing the N signal and the ramp signal, there is an effect of reducing the quantization error of the digital data. Further, by using the counter circuit 40-1 having a function of correcting and counting the gain difference, the memory circuit 50-1 can be simplified, and the difference processing circuit can be reduced from the memory circuit 50-1. In addition, the gain correction function can increase the number of bits and increase the speed by performing AD conversion processing with a small number of bits.

  FIG. 2 is a circuit diagram illustrating a configuration example of the pixel 10-1. The pixel 10-1 includes a photodiode 101, a transfer transistor 102, a reset transistor 103, an amplification transistor 104, and a selection transistor 105. The photodiode 101 is a photoelectric conversion element that generates charges by photoelectric conversion. The transfer transistor 102 transfers the charge accumulated in the photodiode 101 to the floating diffusion portion FD by the control pulse φT. The amplification transistor 104 amplifies the charge on the floating diffusion portion FD by source follower reading. The reset transistor 103 resets the charge on the floating diffusion portion FD with the power supply potential by the control pulse φR. The selection transistor 105 outputs the output signal of the amplification transistor 104 to the vertical signal line V-1 by the control pulse φSEL. The current source 106 is connected to the vertical signal line V-1. The pixel 10-1 is not limited to the configuration example of FIG. 2, but has a configuration in which the selection transistor 105 is eliminated and pixel selection control is performed at a potential set in the floating diffusion portion FD, and a plurality of photoelectric conversion elements 101 have a common amplification transistor 104 may be shared.

  FIG. 3 is an explanatory diagram of the noise included in the pixel signal and a method of comparing the signal level with the reference voltage. In FIG. 3, the horizontal axis represents the amount of light incident on the pixel 10-1, and the vertical axis represents the signal level used for AD conversion. G1 and G8 indicated by solid lines in FIG. 3 indicate pixel signals having different gains of the amplifier circuit. A dotted line 301 is circuit system noise (noise caused by power supply, ground, amplifier circuit, AD conversion, and the like). A broken line 302-1 represents pixel noise at the input of the amplifier circuit after CDS (pixel noise <input conversion noise of the amplifier <circuit system noise). The circuit noise 301 is larger than the pixel noise 302-1. If the circuit noise 301 is 0.2 mV, the SN ratio that is the ratio of the signal level 1V to the pixel noise 0.2 mV is 74 dB. In order to perform AD conversion while covering this SN ratio, a resolution of about 14 bits is required in consideration of quantization bit errors. The higher the resolution, the longer the count period, the longer the AD conversion time, the lower the signal readout speed for the image sensor 100, and eventually the higher speed imaging cannot be performed.

  Therefore, in this embodiment, high-speed reading is achieved by reducing the number of AD conversion bits. When the pixel signal is large, the light shot noise is larger than the circuit noise, so that the circuit noise has a small influence on the SN ratio. Therefore, for example, the gain of the amplifier circuit 20-1 is set to 1 (G1 characteristic). When the pixel signal is small, the circuit noise is larger than the pixel noise in the incident light-output characteristics of G1, and this becomes a dominant factor of the SN ratio. Therefore, by making the gain of the amplifier circuit 20-1 8 times (G8 characteristic), the pixel noise 302-2 becomes larger than the circuit noise 301, and the pixel noise 302-2 becomes a dominant factor of the SN ratio. If the G1 characteristic is used, a signal having a wide light amount range until the output is saturated can be obtained, and if the G8 characteristic is used, a signal having a better SN ratio than the G1 characteristic can be obtained. As described above, either the G1 characteristic or the G8 characteristic is selected according to the amount of incident light, that is, the signal level. As a result, the signal obtained from the imaging device has a good S / N ratio for pixels that output a small signal, and the dynamic range of a pixel that outputs a large signal can be wider than a pixel that outputs a small signal.

  The determination of the signal level includes a method using the characteristic G1 and a method using the characteristic G8. FIG. 3A shows a case where the signal level is determined by comparing the characteristic G1 indicated by the solid line with the reference voltage VREF1 indicated by the alternate long and short dash line 401. On the other hand, FIG. 3B shows a case where the signal level is determined by comparing the characteristic G8 indicated by the solid line with the reference voltage VREF2 indicated by the alternate long and short dash line 402.

  3A and 3B, the signal level corresponding to the light quantity L2 is set as the reference voltage VREF1 or VREF2. When the gain signal is larger than the reference voltage, the characteristic G1 is used, and when the gain signal is smaller than the reference voltage, the characteristic G8 is used.

  It is desirable to set the reference voltage VREF2 to a value smaller than the saturation signal where the G8 characteristic is linear, and the reference voltage VREF1 is obtained by dividing the reference voltage VREF2 by a gain ratio between the G8 characteristic and the G1 characteristic. It is desirable to set to a different value. However, it does not have to be a strict setting value. This is because, since a large signal has a large optical shot noise, even if the signal is determined with a slightly different reference voltage, the dominant factor of the SN ratio does not change.

  A case where the amount of incident light is L1 will be described. In the case where the amount of incident light is L1, as shown in FIG. 3A, the case where the signal level is determined using the signal having the G1 characteristic and the reference voltage VREF1 is considered. In the characteristic of G1, the signal level when the incident light quantity is L1 is V2, which is smaller than the reference voltage VREF1. Therefore, the signal having the characteristic of G8 is selected. As shown in FIG. 3B, a case is considered in which the signal level is determined using the signal having the characteristic G8 and the reference voltage VREF2. In the characteristics of G8, the signal level when the amount of incident light is L1 is V1, which is smaller than the reference voltage VREF2. Therefore, the characteristic G8 is selected.

  A G8 characteristic pixel signal V1 obtained by amplifying the signal obtained when the incident light amount L1 is 8 times is AD converted, and after the conversion, gain correction (returning to the original signal amplitude) is performed to obtain digital data of the pixel signal V2. . As a result, the circuit system noise is 1/8 in calculation, and there is an effect that a high S / N ratio can be obtained. Further, the pixel signal level is not particularly limited to the light amount L1, but the same effect can be obtained in a range smaller than the reference voltage of the light amount L2.

  As described above, gain correction is performed on the gain difference, that is, digital data is bit-shifted (bit correction, 3 bits in the above example), so that ideally, AD conversion data with 13-bit accuracy can be obtained by a 10-bit AD converter. Will be obtained. However, considering that the N signal digital data is differentially processed from the S signal digital data, the least significant bit of the digital data of the pixel signal G1 having a gain of 1 is quantized by the differential processing of the S signal and the N signal. Since the conversion error is large, there is no accuracy of 10 bits. In order to reduce this quantization error, the digital data of the N signal corrects the gain difference, that is, shifts by 3 bits, thereby reducing the quantization error due to the difference processing between the S signal and the N signal. .

  As described above, when the pixel signal is small, a method of correcting the digital data of the pixel signal G8 amplified with a high gain of 8 times to the original gain of 1 and improving the SN ratio as a result. explained. When the pixel signal is large, in order to give priority to the expansion of the dynamic range of the pixel signal, the gain difference is obtained by using the digital data of the characteristic G1 obtained by amplifying the large signal at a low gain of 1 times as the upper bits with respect to the digital data of the G8 characteristic. The bit shift is explained.

  FIG. 4 is a block diagram of an AD conversion unit for explaining the connection of the comparison circuit 30-1 of the present embodiment with the input / output circuit. Blocks having the same functions as those of the embodiment of FIG. To do. The AD converter can convert the photoelectrically converted analog signal into a digital signal at high speed. The comparison circuit 30-1 resets the N signal of the pixel signal Va and the input signal of the ramp signal VRAMP with the pulse φc from the timing generator 70.

  FIG. 5 is a timing chart of the AD conversion unit of FIG. Hereinafter, the AD conversion operation will be described with reference to FIGS. In FIG. 5, a period Tad is a period for performing AD conversion of the N signal and the S signal of the signal Va read from the pixel 10-1. The period Tdata is a digital data transfer period. In the period Tad, the period Td is an AD conversion period of the N signal, and the comparison signal for this is the ramp signal N-RAMP. A period Tj is a signal level determination period of the S signal, and a comparison signal for this is the reference signal VREF. The period Tu is an AD conversion period of the S signal, and the comparison signal for this is the ramp signal S-RAMP.

  The amplifier circuit 20-1 is set to a first gain (signal G1 in FIG. 3) of 1 at the initial stage of the operation shown in FIG. 5, and the amplifier circuit 20-1 outputs an N signal as an output signal Va. The S signal is sequentially output and guided to the input terminal of the comparison circuit 30-2 through the capacitor Ci. The signal VRAMP is input to the other input terminal of the comparison circuit 30-2 through another capacitor Ci. The ramp generation circuit 25 is controlled by the pulse CNT1 of the timing generation circuit 70 to generate the signal VRAMP. The signal VRAMP includes ramp signals N-RAMP, S-RAMP and a reference signal VREF. The amplifier circuit 20-1 is controlled by the signal CNT2 of the timing generation circuit 70.

  Here, the reference signal VREF will be described. Although the reference signal VRERF is generated by the ramp signal generation circuit 25, it may be generated from another power supply circuit. Since the reference signal VREF can be generated by stopping the charging current while the ramp signal is changing with the ramp signal being inclined with respect to time in the ramp signal generation circuit 25, there is an advantage that the circuit configuration can be simplified. . The reference signal VREF can be generated in a short period with respect to the ramp signal S-RAMP. In order to further shorten this period, the charging current may be increased. Further, the reference signal VREF needs to be larger than the maximum value that can be AD converted by the ramp signal N-RAMP. This is because the comparison processing is performed within a signal level range in which the small signal is always larger than the ramp signal N-RAMP by increasing the size. Specifically, in the example shown in FIG. 3, it is conceivable that the maximum value that can be AD-converted by the N-RAMP is set to 60 mV, and the reference signal VREF is set to about (60 + V11) mV.

  In the AD conversion period Td of the N signal, the amplifier circuit 20-1 outputs the N signal as the output signal Va. The ramp signal generation circuit 25 outputs a ramp signal N-RAMP as the output signal VRAMP. In order to increase the resolution of the N signal, the slope of the ramp signal N-RAMP is set to 1/8 of the slope of the ramp signal S-RAMP. The N signal and the ramp signal N-RAMP are compared by the comparison circuit 30-1, and the magnitude relationship between the two is reversed after the period Tr has elapsed. The counter circuit 40-1 counts down during the period Tr. That is, the counter circuit 40-1 starts down-counting when the ramp signal N-RAMP starts to be tilted, and ends down-counting when the magnitude relationship between the two is reversed. The counted down count value (first count value) is set in the counter circuit 40-1.

  Next, in the signal level determination period Tj, the amplifier circuit 20-1 outputs an S signal as the output signal Va. The ramp signal generation circuit 25 outputs the reference signal VREF as the output signal VRAMP. The comparison circuit 30-1 compares the S signal with the reference signal VREF. In this example, since the S signal is larger than the reference signal VREF, the comparison circuit 30-1 outputs a low-level gain switching signal HO to the amplification circuit 20-1. As a result, the gain of the amplifier circuit 20-1 maintains the first gain that is 1 time, and the amplifier circuit 20-1 outputs the S signal (signal G1 in FIG. 3) amplified by 1 time as the output signal Va. To do. If the S signal is smaller than the reference signal VREF, the comparison circuit 30-1 outputs a high-level gain switching signal HO to the amplification circuit 20-1. As a result, the gain of the amplifier circuit 20-1 is switched to the second gain of 8 times, and the amplifier circuit 20-1 outputs the S signal (signal G8 in FIG. 3) amplified by 8 times as the output signal Va. To do. The second gain (high gain: 8 times) is larger than the first gain (low gain: 1 time).

  After the above gain setting, in the AD conversion period Tu of the S signal, the amplifier circuit 20-1 continues to output the S signal as the output signal Va. The ramp signal generation circuit 25 outputs the ramp signal S-RAMP as the output signal VRAMP. The S signal and the ramp signal S-RAMP are compared by the comparison circuit 30-1, and the magnitude relationship between the two is reversed after the period Ts has elapsed. The counter circuit 40-1 that maintains the first count value set in the period Td performs up-counting during the period Ts with respect to the down-count value. That is, the counter circuit 40-1 starts up-counting when the ramp signal S-RAMP starts to tilt, and ends up-counting when the magnitude relationship between the two is reversed. The counted up count value (second count value) is set in the counter circuit 40-1. This value is equivalent to subtracting the N signal from the S signal to indicate the difference between the first count value (first digital value) and the second count value (second digital value) as a result. It becomes a count value. The down-count mode and up-count mode functions of the counter circuit 40-1 will be described later with reference to FIG.

  Note that the low gain set by the amplifier circuit 20-1 in FIG. However, in the imaging system described later with reference to FIG. 12, the low gain or the high gain of the amplifier circuit 20-1 is changed in order to perform sensitivity setting suitable for the imaging environment. Even if the gain of the amplification circuit 20-1 is changed, the correction is made if the gain ratio between the magnification of the low gain (first gain) and the magnification of the high gain (second gain) is made constant between different imaging sensitivities. The amount can be adjusted. When this gain ratio is set to a multiple of 2, digital signal correction is facilitated. Further, the gain ratio between the low gain (first gain) and the high gain (second gain) is constant within the same frame of the image signal. The gain ratio between the low gain (first gain) and the high gain (second gain) is constant within the same horizontal pixel row of the image signal. The sensitivity may be set by changing the slope of the ramp signal. Specifically, by reducing the slope of the ramp signal, AD conversion can be performed with high resolution, resulting in high sensitivity.

  6A to 6C are diagrams illustrating a configuration example of the counter circuit (correction unit) 40-1. The counter circuit 40-1 counts until the output of the comparison circuit 30-1 in the comparison between the N signal and the ramp signal N-RAMP and the comparison between the S signal and the ramp signal S-RAMP is reversed. When the comparison circuit 30-1 compares N signals, the counter circuit 40-1 counts down. On the other hand, when the comparison circuit 30-1 compares the S signals, the counter circuit 40-1 counts up. The memory unit 50 holds count data with the resolution ratio corrected.

  FIG. 6A is a diagram illustrating a configuration example of the counter circuit 40-1. 6B and 6C are diagrams for explaining the counting process including the correction process of the counter circuit 40-1. FIG. 6B shows a case where the S signal is larger than the reference signal VREF after comparing the N signal and the ramp signal N-RAMP, and the count data when the S signal and the ramp signal S-RAMP are compared is shown. FIG. FIG. 6C shows a case where the S signal is smaller than the reference signal VREF after the N signal and the ramp signal N-RAMP are compared, and the count data when the S signal and the ramp signal S-RAMP are compared is shown. FIG.

  The counter circuit 40-1 includes an inverter 601, a 3-bit up / down counter 602, a 10-bit up / down counter 603, and switches SW1, SW2. The up / down counter shown in this configuration example is an asynchronous counter. The count clock signal CLK is input to the switches SW1 and SW2. The inverter 601 outputs a logic inversion signal of the gain switching signal HO. The switch SW1 is controlled by the output signal of the inverter 601. The switch SW2 is controlled by a gain switching signal HO. The counter clock signal CLK is input to either the 3-bit up / down counter 602 or the 10-bit up / down counter 603 according to the gain switching signal HO.

  In FIG. 6B, the S signal is larger than the reference signal VREF, and the gain of the amplifier circuit 20-1 is set to 1. In the period Tr, the gain switching signal HO is at a low level, and N signal comparison processing is performed. Then, the counter clock signal CLK is input to the clock terminal of the 3-bit up / down counter 602 by the switch SW1. The carry output (carry out) co of the 3-bit up / down counter 602 is output to the clock terminal of the 10-bit up / down counter 603 by the switch SW2. The 3-bit up / down counter 602 counts down in synchronization with the counter clock signal CLK and outputs data D0 to D2. The 10-bit up / down counter 603 performs down-counting in synchronization with the carry output co of the 3-bit up / down counter 602, and outputs data D3 to D6. The down count value (first count value) of the N signal is data D0 to D6. Next, in the period Tj, since the S signal is larger than the reference signal VREF, the gain switching signal HO becomes low level, the gain of the amplification circuit 20-1 is set to 1 time, and the amplification circuit 20-1 is 1 time. A low gain signal G1 is output. Next, in the period Ts, the gain switching signal HO is at a low level, and the counter clock signal CLK is not input to the clock terminal of the 3-bit up / down counter 602 by the switch SW1. The counter clock signal CLK is output to the clock terminal of the 10-bit up / down counter 603 by the switch SW2. The 10-bit up / down counter 603 counts up to the first count value in synchronization with the counter clock signal CLK, and outputs the up-count value (second count value) to the memory unit 50. The memory unit 50 shifts the up-count value by 3 bits and holds 10-bit data D3 to D12 shifted by 3 bits as data Da3 to Da12. The memory unit 50 holds the output 3-bit data D0 to D2 of the 3-bit up / down counter 602 as data Da0 to Da2. As a result, data obtained by performing the difference between the S signal and the N signal in the 3-bit up / down counter 602 and the 10-bit up / down counter 603 becomes Da0 to Da12. The 13-bit data Da0 to Da12 correspond to the data D0 to D12, respectively, and are held in the memory circuit 50-1. Thus, the AD conversion data D3 to D12 of the low gain S signal are shifted by 3 bits from the low gain N signal data D0 to D6 and subjected to differential processing. Thereby, highly accurate 13-bit AD conversion data Da0 to Da12 are obtained.

  In FIG. 6C, the S signal is smaller than the reference signal VREF, and the gain of the amplifier circuit 20-1 is set to 8 times. In the period Tr, the N signal is down-counted as in FIG. The down count value (first count value) is data D0 to D6. Next, in the period Tj, since the S signal is smaller than the reference signal VREF, the gain switching signal HO becomes high level, the gain of the amplifier circuit 20-1 is set to 8 times, and the amplifier circuit 20-1 is 8 times higher. The high gain signal G8 is output. Next, since the gain switching signal HO is at a high level during the period Ts, the counter clock signal CLK is input to the clock terminal of the 3-bit up / down counter 602 by the switch SW1. The carry output (carry out) co of the 3-bit up / down counter 602 is output to the clock terminal of the 10-bit up / down counter 603 by the switch SW2. The 3-bit up / down counter 602 counts up in synchronization with the counter clock signal CLK. The 10-bit up / down counter 603 performs up-counting in synchronization with the carry output co of the 3-bit up / down counter 602 and outputs 10-bit data D0 to D9 to the memory unit 50. The dummy data D10 to D12 are “0”. The data D0 to D9 are held as data Da0 to Da9, the dummy data D10 to D12 are held as data Da10 to Da12, and the 13-bit data Da0 to Da12 are held in the memory unit 50, respectively. As a result, data obtained by performing the difference between the S signal and the N signal in the 3-bit up / down counter 602 and the 10-bit up / down counter 603 becomes Da0 to Da9. The dummy data D10 to D12 are added as data Da10 to Da12. The 13-bit data Da0 to Da12 are held in the memory circuit 50-1. Since the dummy data D10 to D12 are small amplitude data (high gain data), it means that the upper bits are zero.

  In the present embodiment, as described above, regardless of the gain of the S signal, as a result, the N signal uses count data that is compared and processed with a low resolution and a high resolution. For this reason, it is possible to obtain highly accurate digital data in which the influence of quantization noise of the low gain signal is reduced. Further, 13-bit digital data can be acquired by shifting the AD-converted 10-bit count data by 3 bits. Further, since the counter circuit 40-1 performs differential processing between the S signal and the N signal and gain correction, there is an effect that the circuit is simplified. For the above-described bit shift, the counter circuit 40-1 having a count function in the down-count mode and the up-count mode is used, but the counter circuit 40-1 may be a common counter system as described above.

  The comparison circuit 30-1 compares the N signal with the ramp signal N-RAMP in the period Td, and the counter circuit 40-1 compares the period Tr until the magnitude relationship between the N signal and the ramp signal N-RAMP is reversed. First, the first count value is counted. Thereafter, the comparison circuit 30-1 compares the S signal with the reference signal VREF in the period Tj. The comparison circuit 30-1 sets the gain of the amplification circuit 20-1 to the first gain (1 times) when the S signal is larger than the reference signal, and the amplification circuit 20-1 when the S signal is smaller than the reference signal. Is set to the second gain (8 times). After that, the comparison circuit 30-1 compares the S signal with the ramp signal S-RAMP in the period Tu, and the second count in the period Ts until the magnitude relationship between the S signal and the ramp signal S-RAMP is reversed. Count the values. The counter circuit 40-1 and the correction unit of the memory unit 50 correct the difference in resolution between the first count value and the second count value corresponding to the difference in gain between the N signal and the S signal. Then, the memory unit (correction unit) 50 outputs the difference data Da0 to Da12 of the corrected first count value and second count value.

  In the above description, the example in which the first count value is down-counted in the period Tr and the second count value is up-counted in the period Ts has been described. The counter circuit 40-1 up-counts the first count value in the period Tr and down-counts the second count value in the period Ts, whereby the difference data Da0 between the first count value and the second count value. ~ Da12 may be output. That is, the counter circuit 40-1 down-counts or up-counts the first count value, and counts the second count value so as to be opposite to the up-down direction of the first count value. Thereby, the memory part 50 can hold | maintain the difference data Da0-Da12 of the correct | amended 1st count value and 2nd count value.

(Second Embodiment)
FIG. 7 is a diagram illustrating a configuration example of the amplifier circuit 20-1 in the imaging device according to the second embodiment of the present invention. Hereinafter, the points of the present embodiment different from the first embodiment will be described. The amplifier circuit 20-1 is a configuration example using a circuit that can reduce offset fluctuation at the time of gain switching. The output signal Va of the amplifier circuit 20-1 is input to the comparison circuit 30-1. After the signal Va and VRAM, which are input signals of the comparison circuit 30-1, are initially reset by the pulse φc, the change in the AD conversion data is small because the variation in the offset potential is small even when the gain is switched. Further, if the gain switching is switched from the low gain to the high gain, the high gain AD conversion data is subjected to gain correction, and at this time, there is an effect that the change in the offset potential is further reduced.

  Next, the operation of the amplifier circuit 20-1 will be described. The amplifier circuit 20-1 has an operational amplifier 20-2, clamps the N signal of the pixel 10-1 as an input signal with a clamp capacitor Co, and compares the amplified signal Va according to the gain setting (for example, 1 time). Output to 30-1. The operational amplifier 20-2 has an inverting input terminal, a non-inverting input terminal, and an output terminal. The clamp capacitor (second capacitor) Co is connected between the inverting input terminal of the operational amplifier 20-2 and the pixel 10-1. The feedback circuit includes a switch SW1 for clamping the clamp capacitor Co to the reference voltage Vr, a capacitor C1 and a capacitor C8 for amplifying an input signal together with the clamp capacitor Co, and switches SW2 and SW3 for controlling connection of these to an input / output terminal. The series connection circuit of the first capacitor C1 and the first switch SW2 and the series connection circuit of the first capacitor C8 and the first switch SW3 are connected in parallel between the inverting input terminal and the output terminal of the operational amplifier 20-2. Connected. In addition, the feedback circuit completely transfers the charge of the capacitor to the switched capacitor when the gain is switched, and performs offset compensation. Further, switches SW4 and SW5 for gain compensation at the time of gain switching are connected to the reference voltage Vr. The second switch SW4 is connected between the interconnection point of the first capacitor C1 and the first switch SW2 and the non-inverting input terminal of the operational amplifier 20-2. The second switch SW5 is connected between the interconnection point of the first capacitor C8 and the first switch SW3 and the non-inverting input terminal of the operational amplifier 20-2. In this embodiment, the capacitance values of the capacitors C1, C8 and Co are set so that the gains are Co / C1 = 1 and Co / C8 = 8. In this case, the parasitic capacitance is omitted for the sake of simplicity. Each of the switches SW1 to SW5 has an equivalent circuit configuration shown on the drawing.

  A driving method of the amplifier circuit 20-1 of FIG. 7 will be described with reference to the timing chart of FIG. Similar to the timing chart of FIG. 5, the period Tad is the AD conversion period for the N signal and the S signal, the period Td is the AD conversion period for the N signal, the period Tj is the signal level determination period for the S signal, and the period Tu is the AD signal period for the S signal. Conversion period. Before performing the AD conversion operation, a period during which the amplifier circuit 20-1 and the comparison circuit 30-1 are initially set is Tc during N signal readout from the pixel unit 10. In the period Tc, the amplification transistor 104 is operated by the high levels of the pulses φSEL and φR of the pixel 10-1. At the same time, the amplifier circuit 20-1 is reset to the initial state by turning on the switches SW1, SW2, and SW3, and the comparison circuit 30-1 is reset to the initial state by the high level of the pulse φc. By setting the pulse φR to a low level and turning off the reset transistor 103, the floating diffusion portion FD enters a floating state. In order to set the amplification circuit 20-1 to an initial gain (for example, gain of 1), the switch SW2 is kept on and the switches SW1 and SW3 are turned off, so that the output signal Va of the amplification circuit 20-1 It becomes an offset voltage at a gain of 1 after clamping the N signal.

  Next, by controlling the reset pulse φc of the comparison circuit 30-1 to a low level, the comparison circuit 30-1 becomes a potential obtained by clamping the N signal of the pixel signal. Next, in the period Td, AD conversion of the N signal is performed using the ramp signal N-RAMP as described above. When the AD conversion period Td ends, the transfer transistor 102 is turned on by the high level of the pulse φT, and the charge of the photodiode 101 is transferred to the floating diffusion portion FD. The amplifier circuit 20-1 outputs the N signal amplified by a gain of 1 to the comparison circuit 30-1 as a signal Va. The transfer transistor 102 is turned off by the low level of the pulse φT. Next, in the period Tj, as described above, the signal level of the S signal is determined using the reference signal VREF. When the S signal is smaller than the reference signal VREF, the gain switching signal HO is input to the amplifier circuit 20-1 as a high level signal. As a result, the switch SW2 is turned off, the switches SW3 and SW4 are turned on, the gain of the amplifier circuit 20-1 is switched to 8 times, and the S signal amplified by 8 times is output as the signal Va. When the S signal is larger than the reference signal VREF, the gain switching signal HO remains at a low level, the switches SW1 to SW5 do not change, and the gain of the amplifier circuit 20-1 remains unchanged. By this operation, the offset of the amplifier circuit 20-1 can be reduced. Next, in the period Tu, AD conversion of the S signal is performed using the ramp signal S-RAMP as described above. As described above, AD conversion data in which a change in the offset voltage of the amplifier circuit 20-1 is reduced can be obtained.

  In the above description, the low gain signal is the initial gain. However, the high gain signal may be the initial gain, and the low gain signal may be switched to the low gain by the gain switching signal. In the timing chart of FIG. 8, the ramp signals N-RAMP and S-RAMP have the same slope. In this case, the resolution of the N signal and S does not change, and the counting process at this time will be described with reference to FIG. Although the quantization noise of the low gain signal cannot be reduced, the bit is increased.

(Third embodiment)
FIG. 9 is a diagram illustrating a configuration example of the amplifier circuit 20-1 and the comparison circuit 30-1 in the imaging device according to the third embodiment of the present invention. Hereinafter, differences of the present embodiment from the first and second embodiments will be described. In the present embodiment, an amplifier circuit 20-1 having two systems of amplifier circuits A and B is provided to perform gain switching. In the present embodiment, the circuit configurations of the two systems of amplifier circuits A and B are the same as those of the amplifier circuit 20-1 in FIG. The first amplifier circuit A includes a first operational amplifier 20-2, capacitors Co, C8, C16, and switches SW10 to SW14. The first operational amplifier 20-2 has an inverting input terminal, a non-inverting input terminal, and an output terminal. The series connection circuit of the first capacitor C1 and the first switch SW12 is connected between the inverting input terminal and the output terminal of the first operational amplifier 20-2. A series connection circuit of the second capacitor C16 and the second switch SW13 is connected between the inverting input terminal and the output terminal of the first operational amplifier 20-2. The third switch SW10 and the third capacitor Co are connected between the inverting input terminal of the first operational amplifier 20-2 and the pixel 10-1. The fourth switch SW11 is connected between the inverting input terminal and the output terminal of the first operational amplifier 20-2. The fifth switch SW5 is connected to the output terminal of the first operational amplifier 20-2.

  The second amplifier circuit B includes a second operational amplifier 20-2, capacitors Co, C1, C2, and switches SW21 to SW24. The second operational amplifier 20-2 has an inverting input terminal, a non-inverting input terminal, and an output terminal. A series connection circuit of the fifth capacitor C1 and the fifth switch SW22, the sixth capacitor C2, and the sixth switch SW23 is connected between the inverting input terminal and the output terminal of the second operational amplifier 20-2. The seventh switch SW21 is connected between the inverting input terminal and the output terminal of the second operational amplifier 20-2. The fourth capacitor Co is connected between the inverting input terminal of the second operational amplifier 20-2 and the pixel 10-1. The eighth switch SW24 is connected to the output terminal of the second operational amplifier 20-2.

  When the pixel reset signal is read out, the two amplifier circuits A and B and the comparison circuit 30-1 are simultaneously reset at the same time, and then the gain is set. Although the gain setting varies depending on the concept of sensitivity setting of the imaging system, in the circuit of FIG. 9 of the present embodiment, the amplifier circuit A selects a high gain considering the imaging sensitivity of the imaging device, and the amplifier circuit B has a low gain. The basic operation of the amplifier circuit 20-1 is the same as that in FIG. The first amplifier circuit A is a high gain amplifier circuit, and the second amplifier circuit B is a low gain amplifier circuit. The gain is determined by the input capacitance Co and the feedback capacitance Cn. The first amplifying circuit A can be set to gain 8 times by using the capacitor C8 and gain 16 times by using the capacitor C16. Similarly, the second amplifier circuit B can be set to a gain of 1 by using the capacitor C1, and can be set to a gain of 2 by using the capacitor C2. In order to simplify the explanation, a case where two gains are switched is described as an example, but it may be configured such that three or more gains can be switched. An operation of increasing the gain of the first amplifier circuit A by 8 times and an operation of increasing the gain of the second amplifier circuit B by 1 will be described as examples, but the operation is the same even if the gain setting is changed. Generally, the gain of an image signal for one frame is determined according to the sensitivity setting set in the imaging system. In this embodiment, since either gain is selected for each pixel row according to the signal level and AD conversion is performed, the above-described bit shift, which is a resolution correction process from the AD conversion data, can be easily performed by the same circuit or the same data processing. There is an effect that can be done. First, the first amplifier circuit A is multiplied by 8 times, and the comparison circuit 30-1 determines the S8 signal level. When the S signal is smaller than the reference signal VREF, the high gain signal is compared. When the S signal is greater than the reference signal VREF, the low gain signal of the amplifier circuit B is selected and comparison processing is performed. Increase the gain ratio to 8 times. The reason for comparing the high gain signal with the reference signal VREF to determine the signal level will be described below.

  For example, when the gain ratio is 16, if the signal level is determined from the low gain signal, the signal level determination level is about 62 mV when the saturation signal is 1 V. The N signal amplitude (variation of the reset signal) of the comparator is assumed to be about 60 mV, and the amplitude range of the N signal is large, so that the signal level of the S signal cannot be accurately determined. If the signal is a high gain signal, the signal level is determined using a signal sufficiently larger than the N signal, so that the influence of the N signal can be reduced. For example, if the gain ratio is four times, the signal level determination level is 250 mV, and the influence of the N signal amplitude is small. Accordingly, when such a gain ratio is small, the signal level may be determined using a low gain signal.

  The amplifier circuit 20-1 only needs to be a circuit that can output two systems of low gain and high gain signals. Further, an amplifier circuit of another type, for example, a common source amplifier circuit may be used. The common source amplifier circuit has an effect of reducing the circuit area. Low gain and high gain signals may be obtained simultaneously, or the signal level may be first determined from one gain signal and then the other gain may be obtained.

  With reference to the timing diagram of FIG. 10, a driving method for simultaneously obtaining a low gain signal and a high gain signal and selecting a gain with the circuit of FIG. 9 will be described. Similarly to the timing chart of FIG. 8, the period Tad is the AD conversion period of the N signal and the S signal, the period Td is the AD conversion period of the N signal, the period Tj is the signal level determination period of the S signal, and the period Tu is the AD signal of the S signal. Conversion period. Before performing the AD conversion operation, a period during which the amplifier circuit 20-1 and the comparison circuit 30-1 are initially set is Tc during N signal readout from the pixel unit 10. In the period Tc, the amplification transistor 104 is operated by the high levels of the pulses φSEL and φR of the pixel 10-1. The switches SW10, SW14 and SW24 are turned on, the amplifiers A and B are reset to the initial state by the control of the switches SW11, SW12, SW13, SW21, SW22 and SW23, and the comparison circuit 30-1 is initialized by the high level of the pulse φc. Reset to state. By setting the pulse φR to a low level, the floating diffusion portion FD enters a floating state, and the N signals are input to the amplifier circuits A and B. In order to set the gain of the amplifier circuit A to 8 times, the switch SW12 is kept on and the switches SW11 and SW13 are turned off, so that the output signal Va-L of the amplifier circuit A is 8 times the gain after clamping the N signal. Is the offset voltage. Similarly, in the amplifier circuit B, the switch SW22 is kept on in order to set the gain to 1 and the switches SW21 and SW23 are turned off, so that the output signal Va-H of the amplifier circuit B is obtained after clamping the N signal. The offset voltage at a gain of 1 is obtained. Next, by controlling the reset pulse φc of the comparison circuit 30-1 to a low level, the input of the comparison circuit 30-2 becomes a voltage obtained by clamping the N signal.

  Next, the switches SW10, SW14, and SW24 are kept on, and AD conversion of the N signal is performed using the ramp signal N-RAMP as described above in the period Td. When the AD conversion period Td of the N signal ends, the switch SW24 is controlled to be off and the amplifier circuit B is electrically disconnected from the comparison circuit 30-1. That is, only the output signal Va-L of the amplifier circuit A is input to the comparison circuit 30-2 through the capacitor Ci. Here, an example has been described in which the switch SW24 is turned off after the period Td ends. However, the switch SW24 is any period from the end of the period Tc to the start of the period Tj as long as it is a period excluding the period Td. It may be switched off.

  Then, the transfer transistor 102 is turned on by the high level of the pulse φT, the charge of the photodiode 101 is transferred to the floating diffusion portion FD, and the amplifier circuit A outputs an S signal having a gain of 8 times to the comparison circuit 30-1. The transfer transistor 102 is turned off by the low level of the pulse φT, and the signal level of the S signal is determined using the reference signal VREF in the period Tj as described above. When the S signal is larger than the reference signal VREF, the gain switching signal HO is input to the amplifier circuit 20-1 as a high level signal. Then, the switch SW10 is turned off and the switch SW11 is turned on, and the output signal Va-L of the amplifier circuit A becomes an offset signal of the amplifier circuit. That is, the input capacitance Ci of the amplifier circuit A and the comparator returns to the initial reset potential. Next, when the switch SW24 is turned on, a signal having a gain of 1 of the amplifier circuit A is input to the comparator, and AD conversion of the S signal is performed using the ramp signal S-RAMP in the period Tu. Switching of the switch when the S signal of the amplifier circuit A is larger than the reference signal VREF is the timing indicated by the broken line in the figure.

  Further, when the S signal of the amplifier circuit A is smaller than the reference signal VREF, AD conversion of the S signal is performed using the ramp signal S-RAMP in the period Tu. In the ramp signal VRAMP shown in FIG. 10, when the inclination of the N-RAMP and S-RAMP with respect to time is reduced, AD conversion processing of a low amplitude signal is performed, so that the shooting sensitivity corresponds to high-sensitivity shooting. At this time, the comparison reference signal VREF also needs to be reduced in proportion to the slope of the RAMP signal (broken line in the figure). In this case, since the variation of the N signal does not change, the amplitude of the ramp signal N-RAMP does not change. The ramp signal N-RAMP can be changed to a plurality of types of ramp signals having the same amplitude and different inclinations. Thus, by reducing the high gain of the amplifier circuit 20-1 and the slope of the ramp signal, highly sensitive AD conversion processing can be performed.

  As described above, the output signals of the two amplifier circuits A and B are selected by the control of the switches SW14 and SW24, and the signal amplified with the set gain is output to the comparison circuit 30-1. When the first gain (high gain) is set, the amplifier circuit 20-1 outputs only the output signal Va-L of the first amplifier circuit A to the comparison circuit 30-1. In addition, when the second gain (low gain) is set, the amplifier circuit 20-1 outputs the output signal Va-H of the second amplifier circuit B to the comparison circuit 30-1. With the outputs of the two systems of amplifiers A and B and the comparison circuit 30-1 connected, the amplification circuit 20-1 and the comparison circuit 30-1 are reset to the initial state. Alternatively, the two amplification circuits A and B may be individually connected to the comparison circuit 30-1 to reset the amplification circuit 20-1 and the comparison circuit 30-1 to the initial state. As a result, even if the gain is switched, AD conversion of the N signal, which is an initial reset signal, can be performed only once, and the AD conversion data of the N signal can be used regardless of the gain. Further, by turning on the switches SW14 and SW24, a signal obtained by adding the output signal Va-L of the first amplifier circuit A and the output signal Va-H of the second amplifier circuit B is output to the comparison circuit 30-2. be able to.

  Since the signal level is low in a low illumination environment, the gain is increased by the amplifier circuit to increase the signal level, or the slope of the lamp signal is decreased. On the other hand, in a high illumination environment, light shot noise is large, so the influence of circuit noise is very small. Therefore, in this embodiment, the dynamic range of the signal is expanded by correcting the data obtained with the low gain setting.

  With reference to FIG. 11, the relationship between the pixel signal and the ramp signal in the case of high gain and high ISO sensitivity with a small slope of the ramp signal will be described. For example, when the photographing sensitivity described in FIG. 13 is high ISO 1600, the saturation signal of the amplifier circuit is set to 1V, the signal has a low gain of 2 times (G2), a high gain of 16 times (G16), and the slope of the ramp signal is 1. This is an example that was set to / 2. The ramp signal amplitude 501 of the S signal is VL, and a signal smaller than this level is AD converted. In the embodiment of FIG. 11, the amount of light that is AD-converted is a signal up to the light amount L6 for the characteristic G2 signal and a signal up to the light amount L4 for the characteristic G16.

  Consider a signal at the amount of light L3. Since the signal V3 of the characteristic G16 is equal to or lower than the reference voltage VREF3 of the one-dot chain line 403, it is used after AD conversion. When the signal of the characteristic G16 is larger than the reference voltage VREF3, the low gain characteristic G2 is used. The digital data is level-shifted and used so that the digital data from the light quantity L4 to the light quantity L6 corresponds to the light quantity L4 or more of the characteristic G16. As a result, signals from the light quantity L4 to the light quantity L6 that could not be used conventionally can be used, and the dynamic range is expanded. In the above description, the low gain signal is corrected. However, the high gain signal may be corrected, and the video signal processing unit in the subsequent imaging system may increase the gain by 3 bits.

(Fourth embodiment)
FIGS. 12A to 12B are diagrams for explaining the counting process of the counter circuit (correction unit) according to the fourth embodiment of the present invention. In the embodiment of FIG. 6, the count processing is performed with the N signal having a high resolution and the S signal having a low resolution when the gain is low. In the present embodiment, the N signal and the S signal are counted with a low resolution. . The counter performs SN processing using a 10-bit counter at the time of count processing at low gain and high gain. The memory has a 13-bit configuration, and a 10-bit data bit shift and 3-bit dummy data are added when data is held in the memory from the counter. Alternatively, as another embodiment, the memory may have an 11-bit configuration, and the added 1 bit may be used as gain information when data is transferred from the memory to the horizontal transfer line, or may be bit-shifted outside the imaging device. Thus, the circuit scale can be reduced by making the counter 10 bits instead of 13 bits with bit shift added.

  FIG. 13 is a table showing combinations of imaging sensitivity, amplifier gain, and slope of the ramp signal RAMP of the imaging system. In a conventional imaging system, the signal gain is increased in accordance with ISO sensitivity that is imaging sensitivity. For example, the gain is 1 time for ISO 100, 2 times for ISO 200, and 4 times for ISO 400. As described above, in the method of uniformly increasing the gain according to the sensitivity of the imaging system, the signal is easily saturated in the amplifier circuit, so that the usable light amount range is narrowed and the dynamic range is lowered as the ISO sensitivity is increased.

  In the present embodiment, the low gain, the high gain, and the slope of the ramp signal are changed depending on the photographing sensitivity. The gain ratio between the low gain and the high gain is related to the expansion of the number of bits, the improvement of the SN ratio, or the expansion of the dynamic range. The slope of the ramp signal limits the signal amplitude for AD conversion, but has the same function as high sensitivity by reducing the slope.

  In the table shown in FIG. 13, up to the photographing sensitivity ISO400, the low gain is 1 time, the high gain is 8 times, and the ramp signal slope is 1 time. ISO 800 to ISO 3200 have a low gain of 2 times and high gain of 16 times, and ISO 6400 to ISO 12800 have a low gain of 4 times and high gain of 16 times. The slope of the ramp signal is 1/2 for ISO 1600, 1/4 for ISO 3200 and ISO 6400, and 1/8 for ISO 12800. From ISO100 to ISOO3200, the bit is increased by 3 bits, and when ISO6400 or higher, the bit is increased by 2 bits.

  In FIG. 13, the noise and dynamic range (DR) after correction of the high gain signal are exemplary. For example, in the photographing sensitivity 100, the circuit noise is 1/8. This means that the high gain signal is reduced to 1/8 after AD conversion, which means that the circuit noise is 1/8 in calculation. Yes. The signal-to-noise ratio of the signal is improved by reducing the circuit system noise. The improvement in the S / N ratio due to the reduction of the circuit noise has the same meaning as the expansion of the dynamic range. Here, the DR value of 8 indicates that the dynamic range becomes 8 times when the shooting sensitivity is 800. ISO 200 to ISO 800 have higher gain than the conventional gain setting. As the final ISO sensitivity, the amount of gain that is higher than the conventional method is adjusted. The ISO 200 will be explained. Since the gain is 2 times in the conventional sensitivity setting and the high gain is 8 times in the present embodiment, the gain is corrected to ¼ in consideration of that amount, and the low gain is obtained. Double the data and adjust the sensitivity. As a result, the SN ratio is improved. ISO 400 and ISO 800 perform gain correction based on the same concept. In ISO800, the DR value is 1. This is because the dynamic range is doubled by assuming that the circuit noise is 1/2 by high gain correction, but the low gain is doubled, so the dynamic range becomes 1/2 due to signal saturation. As a result, the DR value is 1. However, since the DR value is 1/8 in the conventional method of increasing the gain, in this embodiment, the dynamic range is expanded by 8 times.

  Since ISO1600 has a high gain of 16 times, the gain is the same as the conventional sensitivity setting. As described with reference to FIG. 11, the dynamic range of the signal is expanded by multiplying the low gain digital data by 8 which is the gain ratio. However, since the slope of the ramp signal is ½, the substantial dynamic range expansion is four times the DR value 1/16. ISO 3200 to ISO 12800 can be considered similarly. When the dynamic range is increased as compared with the present embodiment, the gain ratio may be increased. As described above, in this embodiment, in low-sensitivity imaging, the noise of the circuit system is substantially reduced by correcting the gain of high gain digital data, and the signal-to-noise ratio of the signal can be improved. In high-sensitivity shooting, the dynamic range of the signal to be used can be expanded by widening the light amount range in which the signal can be used by correcting the gain of the low gain digital data.

  The higher shooting sensitivity means that the subject image becomes darker, the amount of received light is reduced, and light shot noise predominately determines the signal-to-noise ratio of the signal. Therefore, in this embodiment, the ISO 3200 or later does not have a high gain for increasing sensitivity as in the prior art. This is because the gain effect of the amplifier is small, and whether to further increase the gain may be determined by how the imaging system is constructed. When the gain is high, it is necessary to increase the input capacity of the amplifier, which causes a problem that the image pickup apparatus becomes large and the current consumption increases. With high sensitivity of ISO 1600 or higher, the high gain of the amplifier circuit 20-1 does not change, and the slope of the ramp signal is changed. In the present embodiment, high sensitivity is achieved by reducing the slope of the ramp signal.

  FIG. 14A is an explanatory diagram of a counter process in which the ratio of the low gain to the high gain at the time of low ISO sensitivity is 8 times and 3 bits of bit shift are combined with 10-bit AD conversion data. FIG. 14B is an explanatory diagram of a counter process in which the ratio of the low gain to the high gain at the time of high ISO sensitivity is 4 times and 2 bits of bit shift are combined with 9-bit AD conversion data. Optical shot noise is very large at high ISO sensitivity, so that the number of bits of the counter is reduced to 9 bits, and if the number of data is reduced without using 1 bit or 2 bits of high resolution data, higher-speed shooting is possible. The slope of the ramp signal shown in FIG. 13 becomes equal to the relationship of the amplitude that the ramp signal can take when the length of the AD conversion period is made uniform between different imaging sensitivities.

(Fifth embodiment)
FIG. 15 is a diagram illustrating a configuration example of an imaging system according to the fifth embodiment of the present invention. The imaging system 800 includes, for example, an optical unit 810, an imaging device 100, a video signal processing circuit unit 830, a recording / communication unit 840, a timing control circuit unit 850, a system control circuit unit 860, and a reproduction / display unit 870. The imaging device 820 includes the imaging device 100 and a video signal processing circuit unit 830. As the image sensor 100, the image sensor 100 described in the above embodiments is used.

  An optical unit 810, which is an optical system such as a lens, focuses (condenses) light from a subject on the pixel unit 10 (FIG. 1) in which a plurality of pixels of the image sensor 100 are two-dimensionally arranged. Form an image of The image sensor 100 outputs a signal corresponding to the light imaged on the pixel unit 10 at a timing based on the signal from the timing control circuit unit 850. A signal output from the image sensor 100 is input to a video signal processing circuit unit 830 that is a video signal processing unit, and the video signal processing circuit unit 830 performs an operation on the input signal according to a method determined by a program or the like. Perform signal processing. The signal obtained by the processing in the video signal processing circuit unit 830 is sent to the recording / communication unit 840 as image data. The recording / communication unit 840 sends a signal for forming an image to the reproduction / display unit 870 and causes the reproduction / display unit 870 to reproduce / display a moving image or a still image. The recording / communication unit 840 receives a signal from the video signal processing circuit unit 830 and communicates with the system control circuit unit 860, and records a signal for forming an image on a recording medium (not shown). Also works.

  The system control circuit unit 860 controls the operation of the imaging system in an integrated manner, and controls the driving of the optical unit 810, the timing control circuit unit 850, the recording / communication unit 840, and the reproduction / display unit 870. Further, the system control circuit unit 860 includes a storage device (not shown) that is a recording medium, for example, and a program and the like necessary for controlling the operation of the imaging system are recorded therein. Further, the system control circuit unit 860 supplies a signal for switching the driving mode in accordance with, for example, a user operation into the imaging system. Specific examples include a change in a line to be read out and a line to be reset, a change in an angle of view associated with electronic zoom, and a shift in angle of view associated with electronic image stabilization. The timing control circuit unit 850 controls the drive timing of the image sensor 100 and the video signal processing circuit unit 830 based on control by the system control circuit unit 860 which is a control unit.

  According to the first to fifth embodiments, image noise can be reduced by performing a difference process between an N signal having a high resolution and a low gain and an S signal having an appropriate gain. Further, the circuit scale can be reduced by performing the difference processing between the S signal and the N signal and the gain correction by the same counter circuit 40-1.

  The above-described embodiments are merely examples of implementation in carrying out the present invention, and the technical scope of the present invention should not be construed in a limited manner. That is, the present invention can be implemented in various forms without departing from the technical idea or the main features thereof. For example, as the reference signal, a ramp signal whose level changes linearly with respect to time has been described, but a ramp signal that changes stepwise may be used. In each of the above-described embodiments, the configuration in which the counter circuit is provided for each comparison circuit has been described. However, a configuration in which a common counter circuit is provided for a plurality of comparison circuits and a memory corresponding to each comparison circuit is provided. good. Each memory may obtain the first and second count values described above by holding the count value of the common counter circuit based on the output of the corresponding comparison circuit.

10-1 pixel, 20-1 amplifier circuit, 30-1 comparison circuit, 40-1 counter circuit, 50-1 memory circuit

The imaging device of the present invention includes a pixel, an amplifier circuit, an AD conversion unit, and a correction unit, and the pixel includes a first signal generated when the pixel is reset and a second signal generated by photoelectric conversion. To the amplifier circuit, the amplifier circuit amplifies the signal based on the first signal, the signal based on the second signal, the first gain and the A second output signal amplified with any of the second gains greater than the first gain is output to the AD conversion unit, and the AD conversion unit AD converts the first output signal To generate a first digital value, and to perform AD conversion on the second output signal to generate a second digital value, and the correction unit generates the first output signal and the second output signal. said first digital value generated by the gain difference between the second de And correcting at least one of said first digital value and the second digital value so as to reduce the difference in resolution tal values.

Claims (19)

A pixel that generates a signal by photoelectric conversion;
An amplification circuit for amplifying the signal of the pixel with a set gain;
An AD conversion unit;
A correction unit,
The AD converter is
Comparing the output signal of the amplifier circuit after resetting the pixel with a reference signal that changes with time, and outputting a first digital value;
When the output signal of the amplifier circuit in the non-reset state of the pixel is larger than a reference signal, the gain of the amplifier circuit is set to a first gain,
When the output signal of the amplifier circuit in a non-reset state of the pixel is smaller than a reference signal, the gain of the amplifier circuit is set to a second gain larger than the first gain,
Further, after setting the gain of the amplifier circuit to the first gain or the second gain, the output signal of the amplifier circuit in a non-reset state of the pixel is compared with a reference signal that changes with time. Output a second digital value,
The correction unit is
An imaging apparatus, wherein when the first digital value and the second digital value have different resolutions, the difference in resolution is corrected.   The imaging apparatus according to claim 1, wherein the correction unit corrects the difference in resolution by bit-shifting the second digital value.   The correction unit includes a counter circuit, down-counts or up-counts the first digital value, and counts the second digital value in a direction opposite to the up-down direction of the first digital value. The imaging apparatus according to claim 1, wherein: A plurality of the pixel, the amplifier circuit, the AD conversion unit, and the correction unit;
Further, the plurality of correction units share a counter circuit,
3. The imaging according to claim 1, wherein each of the plurality of correction units includes a memory that holds count data output from the counter circuit as the first digital value and the second digital value. apparatus.   The AD converter compares the output signal of the amplifier circuit and the reference signal after resetting the pixel in a state where the gain of the amplifier circuit is set to the first or second gain. The imaging device according to any one of claims 1 to 4.   The AD converter compares the output signal of the amplifier circuit and the reference signal in a non-reset state of the pixel in a state where the gain of the amplifier circuit is set to the first or second gain. The imaging apparatus according to claim 1, wherein the imaging apparatus is characterized. The AD converter compares the output signal of the amplifier circuit after resetting the pixel and the reference signal, and the correction unit includes a counter circuit, and the output signal of the amplifier circuit after reset of the pixel The first digital value is counted until the magnitude relationship with the reference signal is reversed,
Thereafter, the AD converter compares the output signal of the amplifier circuit in the non-reset state of the pixel with the reference signal, and the counter circuit of the correction unit outputs the output of the amplifier circuit in the non-reset state of the pixel The imaging apparatus according to claim 1, wherein the second digital value is counted until a magnitude relationship between a signal and the reference signal is reversed. The amplifier circuit is
An operational amplifier having an inverting input terminal, a non-inverting input terminal and an output terminal;
A series connection circuit of a plurality of first capacitors and first switches connected in parallel between the inverting input terminal and the output terminal of the operational amplifier;
A plurality of second capacitors connected between a plurality of first capacitors and interconnection points of the first switches and a non-inverting input terminal of the operational amplifier;
The imaging apparatus according to claim 1, further comprising a second capacitor connected between an inverting input terminal of the operational amplifier and the pixel. The amplifier circuit is
A first amplifier circuit that can be set to the first gain;
The imaging apparatus according to claim 1, further comprising: a second amplifier circuit that can be set to the second gain. The first amplifier circuit includes:
A first operational amplifier having an inverting input terminal, a non-inverting input terminal and an output terminal;
A series connection circuit of a plurality of first capacitors and first switches connected in parallel between the inverting input terminal and the output terminal of the first operational amplifier;
A second capacitor connected between the inverting input terminal of the first operational amplifier and the pixel;
The second amplifier circuit includes:
A second operational amplifier having an inverting input terminal, a non-inverting input terminal and an output terminal;
A series connection circuit of a plurality of third capacitors and third switches connected in parallel between the inverting input terminal and the output terminal of the second operational amplifier;
The imaging apparatus according to claim 9, further comprising a fourth capacitor connected between the inverting input terminal of the second operational amplifier and the pixel.   When the first gain is set, the amplifier circuit outputs only the output signal of the first amplifier circuit to the AD converter, and when the second gain is set, the first gain is set. 11. The imaging apparatus according to claim 10, wherein a signal obtained by adding the output signal of the amplifier circuit and the output signal of the second amplifier circuit is output to the AD converter.   The imaging apparatus according to claim 10 or 11, wherein the amplification circuit and the AD conversion unit are reset to an initial state in a state where a plurality of outputs of the amplification circuit are connected to the AD conversion unit.   13. The reference signal for comparison with the output signal of the amplifier circuit after resetting the pixel can be changed to a plurality of types of reference signals having the same amplitude and different slopes. The imaging device according to any one of the above.   The imaging apparatus according to claim 1, wherein when imaging sensitivity changes, the gain of the amplifier circuit does not change, and the slope of the reference signal changes.   The imaging apparatus according to claim 1, wherein the gradient of the reference signal does not change even at different shooting sensitivities, and the gain of the amplifier circuit changes.   The imaging apparatus according to claim 1, wherein a gain ratio between the first gain and the second gain is constant between different imaging sensitivities.   The imaging apparatus according to claim 1, wherein a gain ratio between the first gain and the second gain is constant within the same frame of the image signal.   The imaging apparatus according to claim 1, wherein a gain ratio between the first gain and the second gain is constant within the same horizontal pixel row of the image signal. The imaging device according to any one of claims 1 to 18,
An optical unit for condensing light on the pixel;
An imaging system comprising: a video signal processing unit that processes a signal output from the imaging device.

Images