|Publication number||USRE41865 E1|
|Application number||US 11/412,249|
|Publication date||Oct 26, 2010|
|Filing date||Apr 26, 2006|
|Priority date||Dec 14, 2000|
|Also published as||US6727486, US20020118289|
|Publication number||11412249, 412249, US RE41865 E1, US RE41865E1, US-E1-RE41865, USRE41865 E1, USRE41865E1|
|Original Assignee||Soo-Chang Choi|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (17), Referenced by (2), Classifications (16), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The invention relates to image sensors and, more particularly, to a complimentary metal oxide semiconductor (CMOS) image sensor able to perform analog correlated double sampling (CDS).
Generally, an image sensor is an apparatus that captures images from objects by using the property that silicon semiconductors react with visible light. Most previous image sensors have used charge coupled devices (CCD) as image capturing devices.
However, current CMOS technology has matured to the point that the imagers implemented using CMOS transistors are becoming more popular. CMOS imagers have an advantage over CCD imagers in that supplementary analog and digital circuits can be integrated together with a CMOS image sensing portion on a single chip with very low cost, which makes it possible for the CMOS image sensor to have analog-to-digital conversion circuits and other image processing logic circuits integrated on a single imager.
The on-chip analog-to-digital conversion circuits are comprised of as many comparators as columns in a pixel array of the CMOS image sensor and the picture quality of the CMOS image sensor depends largely on the quality of these comparators that convert analog pixel signals into digital signals.
When the row decoder 600 selects a row line of the pixel array 100, the analog pixel signals are input to the comparator array 200, along with the ramp signal produced by the ramp signal generator 400. The comparators of the comparator array 200 compare the analog pixel signals with the ramp signal to find the digital pixel signals for analog-to-digital conversion.
The comparator array 200 has as many comparators as columns in the pixel array 100 and these comparators perform the analog-to-digital conversion on a row-by-row basis. The converted digital data (signals) are stored in the line buffer 300 on a column by column basis. The digital pixel signals stored in the line buffer 300 are then transferred to the digital controller 500, which performs the image processing on them and then outputs the digital image signals through the output pins of the CMOS image sensor.
The unit pixel 110 includes a photodiode 32 to generate a voltage from an image of an object; a transfer transistor Tx to cut the current pass, which will give the photodiode the chance to collect the photo-generated electrons to produce the pixel voltage; and a source-follower (or drive) transistor Dx driven by the photodiode voltage transferred through the transfer transistor Tx, which has a function to safely transfer the pixel voltage to the comparator. The unit pixel 110 also includes a reset transistor Rx that has two functions, to flush out all the electrons in the photodiode and to apply a reset signal to a gate of the source-follower transistor Dx; a selection transistor Sx to let the source-follower voltage out to a comparator 210; and a bias current source Is to supply the bias current to the source-follower transistor Dx.
To reduce fixed pattern noise (FPN), correlated double sampling (CDS) is used when reading the pixel data. CDS includes two phases, reading reset voltage and reading data voltage. To read the reset voltage, the transfer transistor Tx should be turned off, the reset transistor Rx is to be on for a time long enough to charge the floating node connected to the gate of source-follower transistor Dx up to VDD and then off, and the select transistor Sx must be on to apply the output voltage of the source-follower to the comparator. After the completion of AD(analog-to-digital) conversion cycle, the digital value of the pixel reset voltage is stored in the reset bank of line buffer.
To read the data voltage, the transfer transistor Tx is turned on for some time long enough to complete the process of charge sharing of the photodiode and the floating node of the Dx transistor and then off, and the select transistor Sx is turned on to apply the data voltage of the transistor Dx to the comparator for AD conversion. During the second phase, the Rx transistor is always off. After the second phase, the digital value of pixel data is stored in the data bank of line buffer. The actual CDS process is carried out by the digital control block 500, which digitally subtracts the reset value from the data value, to filter out all the signal sources of fixed pattern noise.
The process of AD conversion of this imager is simple. When the ramp generator 400, a simple switched-capacitor integrator, starts to generate a ramp signal, the digital control block 500 starts to count the gray code and the gates of digital latches in the line buffer 300 controlled by the comparator 200 that compares the ramp signal (+) and the pixel voltage (−), opens the gates of latches when the ramp signal is higher than the pixel voltage, and closes the gates when it is lower are open and ready for the digital latches to follow the codes of the gray counter. The comparator 200 then closes the gates of latches in the line buffer 300 when the ramp signal is the same as, or lower than, the pixel voltage, which means that the latches of the column controlled by the comparator of that column keep the digital value in gray code converted from the analog pixel voltage. In other words, the ramp generator scans from the voltage higher than the maximum possible pixel voltage to the voltage lower than the minimum possible pixel voltage so that the comparator can convert all the analog pixel voltages to digital codes. The gray codes in the line buffer are then transferred to the digital control block 500, converted to the binary codes, and processed with the CDS operation after the completion of AD conversion of a full row of pixel voltages.
Typically, a CMOS differential amplifier has an offset voltage and, for the case that a few hundreds of comparators are implemented with such differential amplifiers, the offset voltages of the comparators are not uniform. Therefore, these mismatches of offset voltages of comparators result in the fixed pattern noise in the image captured by this imager. That is why CDS is important in this type of AD conversion. But the traditional CDS performed in the images of
One weak point that the comparator implemented with the CMOS differential amplifier has is that when it is not actually comparing, the static bias currents are still flowing, which results in poor power efficiency. Poor efficiency is a serious defect when applying sensors to mobile applications.
Another weak point is that it is impossible to use a specific gamma correction for the pixel analog signals because the start voltage of the ramp signal is different from one another due to the various offset voltages of the comparators.
The disclosed CMOS image sensor may include an image capturer for capturing an image for analog image signal from an object and an analog-to-digital converter that converts the analog image signal to a digital value using a ramp signal. In such an arrangement, the analog-to-digital converter may include a chopper-type comparator receiving the analog image signal and the ramp signal and a first capacitor that receives a start voltage of the ramp signal and charging a voltage level corresponding the start voltage of the ramp signal in a reset mode and for receiving a down-ramping signal of the ramp signal in a count mode in order to remove an device offset voltage. The CMOS image sensor may also include a ramp signal generator providing the ramp signal to the analog-to-digital converter.
In a rest mode, the start voltage of the ramp signal is charged in the first capacitor and a reset voltage of the image capturer is simultaneously charged in the chopper-type comparator. In a charge transfer mode, the analog image signal from the image capturer is provided to the chopper-type comparator. In a count mode, the down-ramping signal of the ramp signal is provided to the chopper-type comparator in a count mode.
The disclosed apparatus may use analog correlated double sampling, which the CDS is carried out based on analog signals, rather than on conventional digital correlated double sampling carried out after converting analog signals from a pixel to digital signals.
In the disclosed apparatus, an analog signal and an offset voltage from a pixel of the CMOS image sensor are stored in a second capacitor, the ramp signal and the offset voltage are stored in a third capacitor, and then the offset voltage is removed by a switching operation between the second and third capacitors.
Also disclosed is a a method for removing a device offset voltage in a CMOS image sensor. The method may include charging a start voltage of a ramp signal in a capacitor and simultaneously charging a rest voltage of an image capturer in a chopper-type comparator in a reset mode and providing to the chopper-type comparator an analog image signal from the image capturer in a charge transfer mode. The method may also include providing a down-ramping signal of the ramp signal to the chopper-type comparator in a count mode.
Hereinafter, the disclosed apparatus will be described in detail referring to the accompanying drawings.
As mentioned above, the chopper-type comparator disclosed herein includes many switches S1 to S4. The switching operation of the switches S1 to S4 makes an offset voltage caused by charge injection as the following equation:
where Vth is a logic threshold voltage to subsequently connected next digital circuit and A1 and A2 are gains of the first and second stages, respectively. However, this offset voltage is weaker than that in the conventional differential amplifier. Further, the larger the size of the first and second stages 10 and 20, the smaller the offset voltage.
It is possible to reduce the offset voltage by increasing the gains of the first and second stages 10 and 20 and the fixed pattern noise can be considerably reduced by the smaller offset voltage.
On the other hand, a starting voltage (Vstart) of a ramp voltage (Vramp) is applied to node N2 and, on this time, the voltage level at node N2 is Vramp (=Vstart).
Also, the switches S1 and S2 are turned on and a capacitor C3 stores a voltage level of VC3. Subsequently, the switch S2 is turned off immediately after a predetermined time to maintain such a stored voltage as shown in FIG. 9. The voltage Vc3 stored in the capacitor C3 is given by:
To apply an actual data from the unit pixel 120 to the comparator 220, the reset transistor Rx is tuned off and the transfer and selection transistors Tx and Sx are turned on so that the photocharges generated in the photodiode are applied to a gate of the source-follow transistor Dx. At this time, because the gate voltage of the source-follow transistor Dx is Vpixel, a voltage level on node N1 is Vn1 (=Vpixel−(Vth+Voffset)).
Subsequently, the switches S3 and S4 are turned on and then voltage levels of Vclamp1 and Vclamp2 are respectively induced in the capacitors C2 and C3 based on the operation voltage of the inverting amplifiers IN1 and IN2.
On the other hand, because the switch Si is continuously turned on, the capacitors C2 and C1 respectively stores voltage levels of Vc2 and Vc1 as follows:
In summary, the first and second stages mentioned above, ‘Vreset−(Vth+Voffset)−Vstart’ is sampled at the first stage and ‘Vpixel−(Vth+Voffset)−Vclamp1’ is sampled at the second stage. Accordingly, a double sampling for removing the offset voltages in the capacitors C3 and C2 can be achieved, which is called an analog correlated double sampling in the present disclosure.
At the third stage, the switches S1, S3 and S4 are turned off and the switch S2 is turned on to compare the ramp signal from the ramp signal generator 410 to the pixel voltage. Because the switches S1, S3 and S4 are turned off, the voltage levels of the capacitor C1, C2 and C3 are kept continuous, even if the switch S2 is turned on.
At this time, the input voltage (N3) of the inverting amplifier IN1 is given by:
On the other hand, because the start voltage of the ramp signal is Vstart, VN3 is expressed as follow:
As shown in the above polynomial of VN3, the voltage levels of Vth and Voffset, which exist within the polynomials of VC3 and VC2, are removed; thereby achieving the analog correlated double sampling. The voltage level of “Vreset−Vpixel” is a net image data caused by the analog pixel data. Also, since the voltage of Vclamp1 is an operation voltage of the inverting amplifier IN1, the comparison can be obtained while the input voltage of the inverting amplifier IN1 becomes Vclamp1.
A latch enable signal Latch_EN is set to a high voltage level to drive the latch circuit 310 and a clock counting value of the counter 510 increases one by one as the ramp signal from the ramp signal generator 410 gradually decreases.
On the other hand, the ramp signal from the ramp signal generator 410 can be expressed as follow:
Accordingly, the voltage level of VN3 can be expressed as follow:
According to the feature of the ramp signal, the voltage level of ΔV gradually increases with the lapse of time and eventually it is the same as “Vreset−Vpixel.” An input voltage of the inverting amplifier IN1 becomes “Vclamp1” and an input voltage of the inverting amplifier IN12 becomes “Vclamp2 simultaneously, so that the two inverting amplifiers IN1 and IN2 are at the operation voltage at the same time.
This point in time is the comparison moment and, if the ramp signal is dropped a little, the signal is amplified by the gains of the inverting amplifiers IN1 and IN2 and Vo is dropped to a ground voltage level.
If Vo is dropped to the ground voltage level, the final value, which is continuously counted by the counter 510, is stored in the latch circuit 320. Accordingly, the latched value is a digital value from the unit pixel 120.
Finally, the latch enable signal Latch_EN is set to a logic low level in order to store the digital values in the latch circuit 320 until the data stored to latch 310 is transmitted to the digital controller (reference numeral 500 of FIG. 1).
The current of the comparator is consumed in the inverting amplifiers IN1 and IN2 only when the comparison is carried out so that there is little static current and it is possible to reduce the power consumption sharply. Also, because the comparator stores the reset level in the capacitor C3 in the analog signal level, only one ramp signal is required to obtain the digital signal corresponding to the input analog signal with the simple digital control algorithm and operations used in the CMOS image sensor. Further, because it is not necessary to store the digital value corresponding to the reset level of the CMOS image sensor, the entire size of the memories can be reduced by half.
As apparent from the above, the disclosed comparator can reduce the fixed pattern noise, such as the offset voltage, in the CMOS image sensor by considerably removing the offset voltage that exist between pixels using the analog correlated double sampling. The comparator can be made by a simple circuit design without a subtractor because only one ramp signal is used to obtain the digital value. Also, the ramp signal generator for the comparison can has a simple structure so that the chip size of the CMOS image sensor using the disclosed analog correlated double sampling is smaller than others using the digital correlated double sampling. Further, the disclosed apparatus may be employed in other integration circuits in which a low-voltage operation is required to reduce a power consumption or it is necessary to remove the offset value to obtain an exact digital value.
The comparator may have a simple structure that connects, in series, signal processing stages to process input data and the ramp signal. Further, the disclosed device may include a CMOS inverter with a low-operation voltage and a chopper type voltage comparator. Because the chopper type voltage comparator uses an inverter as a voltage amplifier, which consumes the current only when the comparison of inputs is carried out, the disclosed device can reduce the power consumption thereof.
The disclosure introduces a new architecture of CMOS image sensor that has many advantages over the previous one. Such advantages include smaller size of chip area, reduced power consumption, reduced FPN and possibility of implementing analog gamma correction. The disclosed CMOS image sensor is capable of reducing power consumption and a size of chip through the reduction of an offset voltage efficiently therein.
Although certain apparatus constructed in accordance with the teachings of the invention have been described herein, the scope of coverage of this patent is not limited thereto. On the contrary, this patent covers all embodiments of the teachings of the invention fairly falling within the scope of the appended claims either literally or under the doctrine of equivalents.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5877715||Jun 12, 1997||Mar 2, 1999||International Business Machines Corporation||Correlated double sampling with up/down counter|
|US5920274||Aug 5, 1997||Jul 6, 1999||International Business Machines Corporation||Image sensor employing non-uniform A/D conversion|
|US5982318||Oct 10, 1997||Nov 9, 1999||Lucent Technologies Inc.||Linearizing offset cancelling white balancing and gamma correcting analog to digital converter for active pixel sensor imagers with self calibrating and self adjusting properties|
|US6025875||May 23, 1996||Feb 15, 2000||National Semiconductor Corporation||Analog signal sampler for imaging systems|
|US6040570||May 29, 1998||Mar 21, 2000||Sarnoff Corporation||Extended dynamic range image sensor system|
|US6064239 *||Jun 16, 1998||May 16, 2000||Nec Corporation||Chopper-type voltage comparator|
|US6346696||Oct 14, 1999||Feb 12, 2002||Hyundai Electronics Industries Co., Ltd.||Image sensor and method for driving the same|
|US6423957||Dec 30, 1999||Jul 23, 2002||Hyundai Electronics Industries Co., Ltd.||CMOS image sensor having automatic reference voltage controller|
|US6456170||Mar 30, 2000||Sep 24, 2002||Fujitsu Limited||Comparator and voltage controlled oscillator circuit|
|US6727486 *||Dec 14, 2001||Apr 27, 2004||Hynix Semiconductor Inc||CMOS image sensor having a chopper-type comparator to perform analog correlated double sampling|
|US20020051067 *||Aug 17, 2001||May 2, 2002||Stmicroelectronics Ltd.||Modification of column fixed pattern column noise in solid state image sensors|
|JPH01125071A||Title not available|
|JPH04196688A||Title not available|
|JPH09247494A||Title not available|
|JPH09270961A||Title not available|
|JPH10304133A||Title not available|
|JPS57139973A||Title not available|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8319869||Jun 9, 2010||Nov 27, 2012||Panasonic Corporation||Solid-state imaging device|
|US20100245642 *||Jun 9, 2010||Sep 30, 2010||Panasonic Corporation||Solid-state imaging device|
|U.S. Classification||250/208.1, 250/214.1|
|International Classification||H04N5/369, H04N5/357, H04N5/363, H01L27/146, H04N5/335, H04N5/374, H04N5/378, H01L27/00|
|Cooperative Classification||H04N5/374, H04N5/3575, H04N5/378|
|European Classification||H04N5/374, H04N5/378, H04N5/357B|
|May 30, 2009||AS||Assignment|
Owner name: CROSSTEK CAPITAL, LLC, DELAWARE
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:MAGNACHIP SEMICONDUCTOR, LTD.;REEL/FRAME:022764/0270
Effective date: 20090514
|Jul 22, 2011||AS||Assignment|
Owner name: INTELLECTUAL VENTURES II LLC, DELAWARE
Free format text: MERGER;ASSIGNOR:CROSSTEK CAPITAL, LLC;REEL/FRAME:026637/0632
Effective date: 20110718
|Sep 23, 2011||FPAY||Fee payment|
Year of fee payment: 8
|Sep 24, 2015||FPAY||Fee payment|
Year of fee payment: 12