CN106027921A - PWM (Pulse Width Modulation) digital pixel sensor with adaptive reference voltage and large dynamic range - Google Patents

Abstract

The invention relates to the field of CMOS (Complementary Metal Oxide Semiconductor) integrated circuits and provides pulse width modulation digital pixels with the adaptive reference voltage, so that the reference voltage is related to the illumination intensity of the pixels, and thus the expansion of the dynamic range is realized. The invention adopts a technical scheme as follows: a structure of the PWM digital pixel sensor with the adaptive reference voltage and the large dynamic range is that a pixel array is divided into pixel blocks with the same number of pixels, each pixel block includes N pixels and one reference voltage generator, and the whole pixel array further includes a global counter, a column level ADC (Analog to Digital Converter) and a column level storage; a positive end of a photodiode PD of each pixel is connected to the ground and a negative end of the photodiode PD is connected to a reset transistor Mrst and a positive input end of a comparator; an output of the comparator is connected to a writing control end of the storage; and data written into the storage is controlled by the global counter. The PWM digital pixel sensor is mainly applied to design and manufacturing of the PWM digital pixel sensor.

Description

Adaptive reference voltage Larger Dynamic scope PWM count word element sensor

Technical field

The present invention relates to CMOS integrated circuit fields, particularly relate to use Pixel-level analog-to-digital conversion and pulse width modulation The digital pixel image sensor field quantified.

Background technology

Cmos image sensor be broadly divided into digital pixel image sensor (Digital Pixel Sensor, DPS) and The big class of active picture element image sensor (Active Pixel Sensor, APS) two.Digital pixel image sensor is in pixel It is integrated with a/D converter circuit, in pixel, completes the light intensity conversion to time domain, it is possible to obtain higher signal to noise ratio, with Time overcome the restriction of the output voltage swing that low supply voltage in deep-submicron CMOS process brings, have in realtime imaging field Important application.Digital pixel image sensor can be divided into pulse width modulation type (PWM) and pulse frequency modulated by operation principle Two kinds of type (PFM).PFM DPS is iteratively produced multiple pulse signal in exposure process, and the power consumption and the circuit that add system are made an uproar Sound, and in pixel, it is integrated with enumerator, fill factor is low.And PWM DPS by contrast in terms of power consumption and fill factor more Have superiority, but its photoresponse is nonlinear.

Fig. 1 is the structure of tradition PWM count word pixel image sensor.Pixel comprises a photodiode PD, one Reset transistor Mrst, a Pixel-level comparator and a memorizer, enumerator is shared by whole pel array, its N position Output valve is connected to the input of each in-pixel memory.

The work process of PWM count word pixel is divided into three steps: (1) reseting procedure, reset transistor turns on, and PD is reset to Vrst；(2) integral process, Vref is set to high level, and outside Vcount signal control counter starts counting up, and VPD is along with exposure Process begins to decline, and when VPD drops to Vref, comparator output signal overturns, and output signal is connected to memorizer and writes Enter and control end, control the count value of memorizer storage nonce counter, end exposure；(3) readout, outside Read signal control Pixel-level memorizer reading number code processed.

VPD fall off rate is directly proportional to light intensity, and the pulse width that half-light produces is long, and the pulse width that light produces is short, no With pulse width values i.e. corresponding to the digital code in memorizer, it is achieved the time domain conversion of incident intensity to pulse width digital code. For different light intensity, have:

$t_{\mathrm{int}}=\frac{\left(V_{rst}-V_{ref}\right)\·C_{PD}}{I_{ph}}---\left(1\right)$

Can obtain the photoresponse curve of PWM DPS according to above formula, as in figure 2 it is shown, abscissa is light intensity, vertical coordinate is defeated Go out digital code value.

Have benefited from the progress of CMOS technology, can be with integrated more transistor, digital pixel in the pixel of unit are Image quality also improving constantly.The digital pixel image sensor technology processed based on Pixel-level is under small size technique The important development direction of cmos image sensor.But for traditional digital pixel, Pixel-level comparator uses fixing Reference voltage compares, and is limited by quantified precision and the time of integration, and it exists for the detectivity of high light and the low light level Deficiency, dynamic range still ratio is relatively limited, and this limits the application of digital pixel image sensor to a certain extent.

Summary of the invention

For overcoming the deficiencies in the prior art, improve the dynamic range of conventional digital pixel image sensor, it is contemplated that A kind of pulse width modulated digital pixel with adaptive reference voltage is proposed so that reference voltage height and the illumination of pixel Intensity is correlated with, thus realizes the extension of dynamic range.To this end, the technical solution used in the present invention is, adaptive reference voltage is big Dynamic range PWM count word element sensor, structure is, pel array is divided into the block of pixels comprising same number pixel, each Block of pixels comprises N number of pixel and a reference voltage generator, the most whole pel array further comprises global counter, Row level ADC, row level memorizer；The photodiode PD positive ending grounding of each pixel, negative terminal meets reset transistor Mrst and compares The positive input terminal of device, the output of comparator is connected to the write of memorizer and controls end, and in memorizer, the data of write are counted by the overall situation Number device controls；Reference voltage generator be one by sampling capacitance C_H, switch S₁-S₃The sample-hold circuit constituted with amplifier, Sampling capacitance C_HOne end connecting valve S₁、S₃, sampling capacitance C_HThe other end connects amplifier negative input end, switchs S₃The other end connects fortune Putting output, switch is connected across between amplifier negative input end and outfan；Sampling nodes is respectively connecting to same by N number of switch The negative terminal of photodiode in each pixel in block of pixels, the output of reference voltage generator is connected to each pixel comparator Negative input and row level adc circuit.

The working cycle that pixel is complete includes 2 times exposing, and second time time of exposure is 2 times of time of exposure for the first time, Definition total exposure time Tint, then time of exposure is Tint/3 for the first time, and time of exposure is 2Tint/3 for the second time.

When exposure starts for the first time, pel array resets, S, S₁、S₂Switch Guan Bi, now N number of photoelectricity two pole in block of pixels Parasitic capacitance CPD and the sampling capacitance CH of pipe are connected in parallel, and are equivalent to total photoelectric current and discharge the total capacitance after parallel connection； For a block of pixels containing N number of pixel, the voltage Vsample on CPD node is expressed as:

$V_{sample}=V_{rst}-\frac{\left(I_1+I_2+\mathrm{...}+I_N\right)\×T_{\mathrm{int}}/3}{N\×C_{PD}+C_H}=V_{rst}-\frac{{\mathrm{NT}}_{\mathrm{int}}/3}{{\mathrm{NC}}_{PD}+C_H}\stackrel{\‾}{I}=n_1\·\mathrm{\Δ}V---\left(2\right)$

Wherein, Vrst is pixel reset voltage, I_NFor the photoelectric current in n-th pixel in block of pixels,For flat in block of pixels All photoelectric currents, Δ V is the quantization unit of row level ADC, n₁For the digital value quantified for the first time；

For the first time after end exposure, pel array sum counter resets, and switchs S, S₁、S₂Disconnect, switch S₃Guan Bi, CPD ties Voltage on point keeps constant, this average current is stored and is converted to the reference voltage that second time exposes, lead to simultaneously Cross the output of row level ADC, process for subsequent conditioning circuit, because second time time of exposure is primary two times, if so one The photoelectric current of pixel is equal to the average current in block of pixels, then its photodiode voltage VPD can be at second time time of exposure Midpoint drops to reference voltage, for the second time exposure stage, and the time value that n-th pixel quantifies is:

$t_{PWM}=\frac{V_{rst}-V_{ref}}{I_{ph}}{C}_{PD}=\frac{V_{rst}-V_{sample}}{I_{ph}}{C}_{PD}=\frac{{\mathrm{NC}}_{PD}{T}_{\mathrm{int}}/3}{{\mathrm{NC}}_{PD}+C_H}\frac{\stackrel{\‾}{I}}{I_{ph}}\≈\frac{\stackrel{\‾}{I}}{I_{ph}}\frac{T_{\mathrm{int}}}{3}=n_2\·\mathrm{\Δ}t---\left(3\right)$

Wherein, Δ t is the counting interval of enumerator, n₂The digital value quantified for temporal information, I_phFor flowing through photoelectricity two pole The reverse current of pipe PD.

n₁、n₂It is all inverse ratio with light intensity, if Vrange is the scope of voltage quantization, respectively with Vrange/ Δ V-n₁、Tint/ Δt-n₂Represent, and deduct average current second step exposure quantization time value Tint/ (2 Δ t), light intensity always export expression For:

$N=\left(\frac{V_{range}}{\mathrm{\Δ}V}-n_1\right)+\left(\frac{T_{\mathrm{int}}}{\mathrm{\Δ}t}-n_2\right)-\frac{T_{\mathrm{int}}}{2\mathrm{\Δ}t}---\left(4\right).$

The feature of the present invention and providing the benefit that:

On the basis of conventional digital pixel, propose the circuit structure improved, improve circuit performance, expand digital image The dynamic range of element imageing sensor.In pel array, merely add less accessory circuit, can also obtain higher simultaneously Fill factor.

Accompanying drawing illustrates:

Fig. 1 PWM DPS structure.

Fig. 2 PWM pixel light intensity-digital code response curve.

Fig. 3 adaptive reference voltage pwm digital pixel structure.

Fig. 4 adaptive reference voltage pwm digital pixel work schedule.

Fig. 5 minimum detectable light intensity.

The maximum detectable light intensity of Fig. 6.

Detailed description of the invention

The present invention is directed to the problems referred to above, on the basis of conventional digital pixel, propose the variable reference voltage of improvement PWM count word pixel layout, it is possible to effectively expand the dynamic range of digital pixel.Dot structure is as shown in Figure 3.

Pel array is divided into the block of pixels comprising same number pixel, comprises N number of pixel and one in each block of pixels Reference voltage generator, further comprises the structures such as global counter, row level ADC, row level memorizer in the most whole pel array. The structure of each pixel is identical with tradition PWM count word pixel, and the photodiode PD positive ending grounding of pixel, negative terminal connects reset crystal Pipe Mrst and the positive input terminal of comparator, the output of comparator is connected to the write of memorizer and controls end, write in memorizer Data are by global counter control.Reference voltage generator be one by sampling capacitance CH, switch S1-S3 and adopting of constituting of amplifier Sample-holding circuit, sampling nodes photodiode in each pixel in being respectively connecting to same block of pixels by N number of switch Negative terminal, the output of reference voltage generator is connected to negative input and the row level adc circuit of each pixel comparator.

The working cycle that pixel is complete includes 2 times exposing, and second time time of exposure is 2 times of time of exposure for the first time. Definition total exposure time Tint, then time of exposure is Tint/3 for the first time, and time of exposure is 2Tint/3 for the second time, pixel operation Sequential chart is as shown in Figure 4.

When exposure starts for the first time, pel array resets, and S, S1, S2 switch Guan Bi.Now N number of photoelectricity two pole in block of pixels Parasitic capacitance CPD and the sampling capacitance CH of pipe are connected in parallel, and are equivalent to total photoelectric current and discharge the total capacitance after parallel connection. For a block of pixels containing N number of pixel, the voltage Vsample on CPD node is represented by:

$V_{sample}=V_{rst}-\frac{\left(I_1+I_2+\mathrm{...}+I_N\right)\×T_{\mathrm{int}}/3}{N\×C_{PD}+C_H}=V_{rst}-\frac{{\mathrm{NT}}_{\mathrm{int}}/3}{{\mathrm{NC}}_{PD}+C_H}\stackrel{\‾}{I}=n_1\·\mathrm{\Δ}V---\left(2\right)$

Wherein, Vrst is pixel reset voltage, I_NFor the photoelectric current in n-th pixel in block of pixels,For flat in block of pixels All photoelectric currents, Δ V is the quantization unit of row level ADC, n₁For the digital value quantified for the first time.

This sampled voltage contains the information of average photo-current in block of pixels.For the first time after end exposure, pel array Sum counter resets, and S, S1, S2 switch off, and S3 switchs Guan Bi, and the voltage on CPD node keeps constant, and this is the most electric Stream stores and is converted to the reference voltage that second time exposes, and is exported by row level ADC simultaneously, processes for subsequent conditioning circuit. Because second time time of exposure is primary two times, if so the photoelectric current of a pixel is equal to the average electricity in block of pixels Stream, then its photodiode voltage VPD can drop to reference voltage at the midpoint of second time time of exposure.Second time exposure rank Section, the time value that n-th pixel quantifies is:

$t_{PWM}=\frac{V_{rst}-V_{ref}}{I_{ph}}{C}_{PD}=\frac{V_{rst}-V_{sample}}{I_{ph}}{C}_{PD}=\frac{{\mathrm{NC}}_{PD}{T}_{\mathrm{int}}/3}{{\mathrm{NC}}_{PD}+C_H}\frac{\stackrel{\‾}{I}}{I_{ph}}\≈\frac{\stackrel{\‾}{I}}{I_{ph}}\frac{T_{\mathrm{int}}}{3}=n_2\·\mathrm{\Δ}t---\left(3\right)$

Wherein, Δ t is the counting interval of enumerator, and n2 is the digital value that temporal information quantifies, I_phFor flowing through photoelectricity two pole The reverse current of pipe PD..

N1, n2 are all and light intensity is inverse ratio.If Vrange is the scope of voltage quantization, in order to obtain positively related output, point Do not represent with Vrange/ Δ V-n1, Tint/ Δ t-n2, and deduct quantization time value Tint/ of average current second step exposure (2 Δ t), total output of light intensity is represented by:

$N=\left(\frac{V_{range}}{\mathrm{\Δ}V}-n_1\right)+\left(\frac{T_{\mathrm{int}}}{\mathrm{\Δ}t}-n_2\right)-\frac{T_{\mathrm{int}}}{2\mathrm{\Δ}t}---\left(4\right)$

As shown in Figure 5 and Figure 6, when the minimum intensity of light that this structure can detect is to make reference voltage with Vref_max, The photoelectric current of reference voltage is arrived at the end of two-stage.Detectable largest light intensity be with Vmin as the reference voltage time, one The photoelectric current of reference voltage is arrived at the end of individual time quantization unit.Therefore, dynamic range is represented by:

$DR=20{\log }_{10}\left(\frac{I_{phs\max }}{I_{phs\min }}\right)=20{\log }_{10}\left(\frac{\frac{\mathrm{\Δ}V\·2^{k_1}}{\mathrm{\Δ}T}\·C_{PD}}{\frac{\mathrm{\Δ}V}{\mathrm{\Δ}T\·2^{k_2}}{C}_{PD}}\right)=20\left(k_1+k_2\right)\·{\log }_{10}2---\left(5\right)$

Wherein Δ V, Δ T are respectively minimum quantifiable voltage and time, and k1, k2 are respectively ADC and Pixel-level memorizer Resolution.Its dynamic range increases than Typical Digital pixel digital image sensor.

Using SMIC65nm CMOS technology to achieve the block of pixels circuit of 4 × 4 herein, supply voltage is 1.2V, comparator Working range is 0.5～1.1V.The operational amplifier that reference voltage generation module is used is two foldings formula cascade knot Structure, input range 0.5～1.1V, DC current gain is 88dB, and phase margin is 71 °, and speed is 20M.Row level ADC, enumerator and Pixel internal storage storage unit is 8, and CH value is 60fF.When resetting voltage is 1.1V, dynamic range ideally can Reach 96dB.

Claims (4)

1. an adaptive reference voltage Larger Dynamic scope PWM count word element sensor, is characterized in that, structure is, pel array It is divided into the block of pixels comprising same number pixel, each block of pixels comprises N number of pixel and a reference voltage generator, separately Outer whole pel array further comprises global counter, row level ADC, row level memorizer；The photodiode PD of each pixel Positive ending grounding, negative terminal connects the positive input terminal of reset transistor Mrst and comparator, and the output of comparator is connected to writing of memorizer Entering and control end, in memorizer, the data of write are by global counter control；Reference voltage generator be one by sampling capacitance C_H、 Switch S₁-S₃The sample-hold circuit constituted with amplifier, sampling capacitance C_HOne end connecting valve S₁、S₃, sampling capacitance C_HThe other end Connect amplifier negative input end, switch S₃The other end connects amplifier output, and switch is connected across between amplifier negative input end and outfan； Sampling nodes negative terminal of photodiode in each pixel in being respectively connecting to same block of pixels by N number of switch, with reference to electricity The output of pressure generator is connected to negative input and the row level adc circuit of each pixel comparator.

2. adaptive reference voltage Larger Dynamic scope PWM count word element sensor as claimed in claim 1, is characterized in that, as The element complete working cycle includes 2 times exposing, and second time time of exposure is 2 times of time of exposure for the first time, defines total exposure Time Tint, then time of exposure is Tint/3 for the first time, and time of exposure is 2Tint/3 for the second time.

3. adaptive reference voltage Larger Dynamic scope PWM count word element sensor as claimed in claim 1, is characterized in that, the When single exposure starts, pel array resets, S, S₁、S₂Switch Guan Bi, the now parasitic electricity of N number of photodiode in block of pixels Hold CPD and sampling capacitance CH to be connected in parallel, be equivalent to total photoelectric current and the total capacitance after parallel connection is discharged；One is contained Having the block of pixels of N number of pixel, the voltage Vsample on CPD node is expressed as:

$V_{sample}=V_{rst}-\frac{(I_1+I_2+\mathrm{...}+I_N)\×T_{\mathrm{int}}/3}{N\×C_{PD}+C_H}=V_{rst}-\frac{{\mathrm{NT}}_{\mathrm{int}}/3}{{\mathrm{NC}}_{PD}+C_H}\stackrel{\‾}{I}=n_1\·\mathrm{\Δ}V---\left(2\right)$

Wherein, Vrst is pixel reset voltage, I_NFor the photoelectric current in n-th pixel in block of pixels,For average light in block of pixels Electric current, Δ V is the quantization unit of row level ADC, n₁For the digital value quantified for the first time；

For the first time after end exposure, pel array sum counter resets, and switchs S, S₁、S₂Disconnect, switch S₃Guan Bi, on CPD node Voltage keep constant, this average current is stored and is converted to the reference voltage that exposes of second time, simultaneously by row Level ADC output, processes for subsequent conditioning circuit, because second time time of exposure is primary two times, if so pixel Photoelectric current equal to the average current in block of pixels, then its photodiode voltage VPD can be at the midpoint of second time time of exposure Dropping to reference voltage, for the second time exposure stage, the time value that n-th pixel quantifies is:

$t_{PWM}=\frac{V_{rst}-V_{ref}}{I_{ph}}{C}_{PD}=\frac{V_{rst}-V_{sample}}{I_{ph}}{C}_{PD}=\frac{{\mathrm{NC}}_{PD}{T}_{\mathrm{int}}/3}{{\mathrm{NC}}_{PD}+C_H}\frac{\stackrel{\‾}{I}}{I_{ph}}\≈\frac{\stackrel{\‾}{I}}{I_{ph}}\frac{T_{\mathrm{int}}}{3}=n_2\·\mathrm{\Δ}t---\left(3\right)$

Wherein, Δ t is the counting interval of enumerator, n₂The digital value quantified for temporal information, I_phFor flowing through photodiode PD Reverse current.

4. adaptive reference voltage Larger Dynamic scope PWM count word element sensor as claimed in claim 3, is characterized in that, n₁、 n₂It is all inverse ratio with light intensity, if Vrange is the scope of voltage quantization, respectively with Vrange/ Δ V-n₁、Tint/Δt-n₂Carry out table Show, and deduct the exposure of average current second step quantization time value Tint/ (2 Δ t), total output of light intensity is expressed as:

$N=(\frac{V_{range}}{\mathrm{\Δ}V}-n_1)+(\frac{T_{\mathrm{int}}}{\mathrm{\Δ}t}-n_2)-\frac{T_{\mathrm{int}}}{2\mathrm{\Δ}t}---\left(4\right).$