Dynamic, single photodiode pixel circuit and operating method thereof

ES 2 811 152 T3

DESCRIPTION

Single photodiode dynamic pixel circuit and method of operation

Context and background of the invention

The invention relates to a pixel circuit for an image sensor. More specifically, it relates to a pixel circuit and a method of operation thereof, wherein an exposure metering circuit is configured to measure the intensity of light exposure from a photoreceptor signal derived from exposure to light. light from a single photoreceptor, upon detection, by a transient detector circuit, of a change in said photoreceptor signal.

Conventional image sensors capture visual information quantified over time with a predetermined frame rate. Each frame carries the information of all the pixels, regardless of whether or not this information has changed since the last frame was captured. Obviously, depending on the dynamic content of the scene, this approach results in a more or less high degree of redundancy in the recorded image data. The problem is getting worse as today's imaging sensors progress toward increasing spatial and temporal resolution. The complexity and cost of the hardware required for data post-processing increases, the demand on transmission bandwidth and data storage capacity skyrockets and power consumption increases, resulting in strict limitations in all the types of display applications, from demanding high-speed machine vision systems to battery-powered consumer mobile devices.

One approach to dealing with temporal redundancy in video data is difference encoding between frames. This very simple form of video compression includes the transmission of only pixel values ​​that exceed a defined intensity change threshold from frame to frame after an initial key frame. Imaging devices known for frame differentiation are based on capturing and processing entire frames of image data and are not capable, in a self-consistent manner, of removing temporal redundancy and providing a compressed video output in real time. Furthermore, even when the processing and quantization of the differences is performed at the pixel level, the temporal resolution of the scene dynamics capture, as in all frame-based imaging devices, is still limited to the achievable frame rate and is quantized in time relative to this frame rate.

The negative effects of data redundancy are most effectively avoided by not logging the redundant data in the first place and by directly reducing the volume of data at the sensor output level. The immediate benefits are reduced bandwidth, memory, and computational power requirements for data transmission and post-processing, reducing system power, complexity, and cost. Furthermore, the principle of frame-based, clock-driven operation of conventional CMOS or CCD image sensors leads to limitations in temporal resolution as the dynamics of the scene are quantized with respect to the frame rate. at which the field of view of the pixels is read, and also leads to poor dynamic range.

The problem that the present invention aims to solve is the provision of a method and an apparatus for the continuous capture of the complete visual information of a dynamic scene observed as a high intensity and temporal resolution, over a wide dynamic range (of light intensity recordable and actionable) and thus generating the minimum necessary amount of data volume. In this way, the data generated is not constituted by a succession of frames containing the image information of all the pixels, but by a continuous (asynchronous) flow of information of change and intensity (that is, gray level) of pixels. individual pixels, which are recorded and transmitted only if an actual change in light intensity of the individual pixel's field of view has been detected by the pixel itself.

This method results in a substantial reduction in generated data through a complete removal of the temporal redundancy in the image information that is typical of conventional image sensors, but in such a way that the data includes the same information content, or even higher. The picture element corresponding to an image sensor implementing the aforementioned method, as well as the required asynchronous data reading mechanism can be realized based on analog electronic circuitry. An image sensor with a multiplicity of such image elements is typically embodied and manufactured in the form of a system on a single chip, integrated, for example, in CMOS technology.

Implementing such a sensor and thus preventing the aforementioned drawbacks of conventional image data capture would be beneficial for a wide variety of machine vision applications including high speed machine vision (e.g. e.g. fast object recognition, motion detection and analysis, object tracking, etc.), automotive (e.g. 3D stereoscopic vision in real time for collision warning and avoidance, smart rear view mirrors, etc.), surveillance and security (scene surveillance) or robotics (autonomous navigation, SLAM)

ES 2 811 152 T3 as well as biomedical and scientific imaging applications. Insofar as the sensor's operation is inspired by the functional principles of the human retina, one of the advantageous exemplary applications is the treatment of a degenerated retina of a blind patient with an implantable prosthetic device based on data provided by a sensor of the type mentioned.

One of the solutions to achieve the complete suppression of the aforementioned temporal redundancy relies on the pre-processing and acquisition, by individual pixels, of the image information, controlled by events (that is, independent with respect to an external timing control, such as clock, shutter or reset signals) and conditionally (that is, only when changes in the scene have been detected). As explained later, the control of image data acquisition is transferred to the pixel level and can be performed with a very high temporal resolution (eg fully asynchronous).

In the case of the optical transient sensor, or dynamic vision sensor (DVS), an electrical circuit, "a transient detector", described in US patent 7,728,269, detects changes in the intensity of illumination received by the individual pixels that they work autonomously.

The United States patent application US 2010/0182468 A1 discloses the combination of transient detector circuits, that is, circuits detecting changes in the intensity of exposure to light, and circuits for conditional measurement of exposure. A transient detector circuit initiates, individually and asynchronously, the measurement of a new exposure measurement only if a change in brightness of a certain magnitude has been detected - and immediately after it - in the field of view of one pixel . Said pixel is not based on external timing signals and independently requests access to an output channel (asynchronous and arbitrated) only when it has a new gray scale value to communicate. Consequently, a pixel that is not visually stimulated does not produce any output. Additionally, asynchronous operation avoids time quantization of frame-based pickup and scan reading.

For each pixel, the transient detector circuit monitors a photoreceptor voltage derived from a first photodiode to detect relative voltage changes that exceed a threshold. Upon such detection, the transient detector circuit outputs a command to the exposure measurement circuit of the same pixel to initiate an absolute intensity measurement, that is, an absolute gray level measurement. The exposure metering circuit uses a second pixel photodiode, located adjacent to the first photodiode, and derives its measurement from the time span to discharge the photodiode junction capacity with the instantaneous photocurrent.

However, the pixel circuit disclosed in US 2010/0182468 A1 is not optimal since it consumes a large area corresponding to one pixel element and therefore cannot achieve high resolution. Furthermore, time-based exposure metering, through direct integration of photocurrent, typically results in a prohibitively long metering time from a new exposure value, especially at low levels of pixel illuminance, due to the corresponding reduced photocurrents. Finally, the use of two independent photodiodes for the detection of changes and the measurement of exposure results in a spatial divergence and a dependence, with respect to the direction of movement, of the image data acquisition process, resulting in a reduction quality in the treatment of images.

Summary of the invention

The invention aims to provide a pixel circuit with lower area requirements, allowing larger sizes of the arrays or smaller dimensions of the sensor chips. The invention also aims to speed up individual measurement processes and consequently increase the temporal resolution. Furthermore, the invention aims to avoid the spatial divergence between the detection of changes and the exposure measurement, caused by the use of two independent photodiodes, improving the precision of the measurements and, consequently, the quality of the images.

In this regard, the invention relates to a pixel circuit comprising:

- a front stage circuit comprising a single photodiode and presenting an output, said front stage circuit being configured to deliver at said output a photoreceptor signal derived from exposure to light from said single photodiode;

- a transient detector circuit configured to detect a change in said photoreceptor signal delivered at said output;

- an exposure measurement circuit configured to measure said photoreceptor signal delivered at said output upon detection, by the transient detector circuit, of a change in the photoreceptor signal.

ES 2 811 152 T3

In contrast to prior art circuits in which exposure changes were detected at one photodiode and exposure measurements were made at another photodiode, the proposed pixel circuit requires only one photodiode per pixel. Consequently, the surface consumption of the pixel element can be significantly reduced, allowing larger sizes of the arrays or smaller dimensions of the sensor chips. The resolution can also be increased. Furthermore, the spatial divergence between the detection of changes and the exposure measurement is avoided, improving the precision of the measurements and, consequently, the quality of the images. Very advantageously, the time span of a gray level measurement can be significantly reduced as explained below, significantly improving the temporal resolution of the image data acquisition process.

Other preferred, but not limiting, aspects of the pixel circuit are the following, in isolation or in a technically feasible combination:

- the exposure metering circuit comprises

- an input connected to the output of the front stage circuit to receive the photoreceptor signal,

- a capacitor connected by a first switch to said input, said first switch being configured to disconnect said capacitor from said input,

- a current source in series with a second switch, parallel to said capacitor, said second switch being configured to control a discharge from said capacitor;

- the exposure measurement circuit comprises a voltage comparator having a signal input connected to one of the terminals of the capacitor and a reference input connected to a reference voltage;

- the voltage comparator shows:

- a signal input connected to one of the capacitor terminals, and

- a reference input connected to a reference switch configured to selectively connect said reference input to at least two reference voltages;

- the transient detector circuit comprises an amplifier having two single-ended common source inverting stages, with a capacitive feedback, separated by a follower buffer, a first capacitor being charged by means of the photoreceptor signal, and at least one threshold detector is arranged to detect if a voltage on another capacitor exceeds a threshold value;

- the front stage circuit comprises a photoreceptor circuit connected to the single diode, the photoreceptor circuit comprising:

- an output to deliver the photoreceptor signal derived from exposure to light from said single photodiode,

- a first photoreceptor transistor having a drain and a gate, the gate of said first photoreceptor transistor being connected to said output,

- an additional photoreceptor transistor having a drain, a source and a gate, the source of said additional photoreceptor transistor being connected to said single photodiode and, wherein said first photoreceptor transistor and said additional photoreceptor transistor present a source common;

- the gate of the additional photoreceptor transistor is biased by a bias voltage or is connected to the common source of the first photoreceptor transistor and the additional photoreceptor transistor;

- the front stage circuit further comprises a gain stage to amplify the photoreceptor signal delivered at the output of the front stage circuit, said gain stage comprising:

- an input connected to the output of a photoreceptor circuit,

- an exit,

- a first gain transistor having a drain, a source and a gate, the gate being

ES 2 811 152 T3 of the first gain transistor connected to the input of the gain stage, the source of the first gain transistor being connected to a bias voltage and the drain of said first gain transistor being connected to the output of said gain stage,

- a plurality of gain transistors in series, each gain transistor of the series presenting a drain, a source and a gate, each gain transistor of the series presenting its drain connected to its gate, and presenting one of said plurality of gain transistors in series their drain connected to the drain of the first gain transistor.

The invention also relates to an image sensor comprising a plurality of pixel circuits according to a possible embodiment of the invention.

The invention also relates to a method of operating a pixel circuit according to one of the possible embodiments of the invention, wherein a measurement cycle of the exposure to light of a photodiode by means of the measurement circuit exposure is initiated by the detection, by the transient detector circuit, of a change in the photoreceptor signal derived from the intensity of the incident light in said photodiode.

Other preferred, but not limiting, aspects of the pixel circuit are the following, in isolation or in a technically feasible combination:

- a measurement cycle of the exposure to light of a photodiode by means of the exposure measurement circuit can also be initiated via an externally applied control signal independent with respect to any detection in the intensity-derived photoreceptor signal of the light incident on said photodiode;

- the light exposure of the photodiode is measured by determining the time for the voltage across a discharge capacitor of the exposure measurement circuit to reach at least a reference voltage;

- a first reference voltage and a second reference voltage are provided, said first reference voltage being greater than said second reference voltage, and wherein the exposure to light from the photodiode is measured by determining and comparing:

- a first time slot corresponding to the time for the voltage across a discharge capacitor of the exposure measurement circuit to reach said first reference voltage, and

- a second time slot corresponding to the time for the voltage across said discharge capacitor of the exposure measurement circuit to reach said second reference voltage;

- before the exposure metering cycle, the capacitor of the exposure metering circuit is charged with a voltage corresponding to the photoreceptor signal;

- the exposure metering circuit comprises:

- an input connected to the output of the front stage circuit to receive the photoreceptor signal,

- a capacitor connected by a first switch to said input, said first switch being configured to disconnect said capacitor with respect to said input,

- a current source in series with a second switch, parallel to said capacitor, said second switch being configured to control a discharge from said capacitor, and the exposure measurement cycle comprises at least the following stages:

- opening a first switch to disconnect the metering capacitor from the input of the metering exposure circuit,

- close the second switch to allow the discharge of the capacitor,

- determine the time for the exposure discharge capacitor to reach at least a reference voltage,

- determine the light exposure of the photodiode from the determined discharge time of the capacitor.

ES 2 811 152 T3

Brief description of the drawings

Other aspects, objectives, and advantages of the present invention will become more clearly apparent upon reading the following detailed description of preferred embodiments thereof, provided as non-limiting examples, and made with reference to the accompanying drawings, in the what:

- figure 1 shows a block diagram of a pixel circuit according to a possible embodiment of the invention,

Figures 2 and 3 show simplified diagrams of exemplary embodiments of transient detector circuits for detecting changes in the photoreceptor signal;

Figures 4 and 5 show simplified diagrams of exemplary embodiments of the exposure measurement circuit for measuring the photoreceptor signal;

Figures 6 and 7 show simplified diagrams of exemplary embodiments of photoreceptor circuits;

figure 8 shows a simplified diagram of a gain stage for amplifying changes in the photoreceptor signal before its exploitation by the transient detector circuit and the exposure measurement circuit;

figure 9 shows an image sensor comprising a plurality of pixel circuits according to the invention.

Figure 10 shows an alternative block diagram of a pixel circuit according to a possible embodiment of the invention, connected to an analog-to-digital converter, external to the pixel.

In all figures, like reference characters refer to the same elements.

Detailed description of the invention

In figure 1, a simplified diagram of a pixel circuit according to a possible embodiment is shown. The pixel circuit comprises a front stage circuit 1, a transient detector circuit 2, and an exposure measurement circuit 3. The front stage circuit 1 comprises a single PD photodiode and has one output 4. The PD photodiode converts light incident on a photocurrent Iph determined by exposure to light from said single PD photodiode. The front stage circuit 1 also comprises a photoreceptor circuit 5 and a gain stage 6 for generating a photoreceptor signal derived from the exposure to light of said single PD photodiode. The photoreceptor signal is delivered at the output 4 of the front stage circuit 1 to be exploited by both the transient detector circuit 2 and the exposure measurement circuit 3.

The transient detector circuit 2 is configured to detect a change in the photoreceptor signal delivered at the output 4 of the front stage circuit. The transient detector circuit 2 continuously monitors the photoreceptor signal for changes, and responds with a signal that identifies a fractional increase or decrease in the photoreceptor signal that exceeds adjustable thresholds.

The exposure measurement circuit 3 is configured to measure the photoreceptor signal delivered at the output 4 of the front stage circuit 1. The exposure measurement cycle starts if the transient detector circuit 2 has detected a change in the signal. photoreceptor, although it can also be initiated via an externally applied control signal independent of any detection of a change.

Transient detector circuit

US 7,728,269 discloses a transient detector circuit that can be used in some embodiments of the invention. In the following, principles of a transient detector circuit of the aforementioned type are explained.

In Fig. 2, a simplified diagram of the transient detector circuit 2 for detecting changes in the photoreceptor signal is shown. The input signal Vfront at the input 20 of the transient detector circuit 2 is the photoreceptor signal Vfront at the output 4 of the front stage circuit 1. The transient detector circuit 2 comprises a first capacitor Ci. One of the terminals of said first capacitor Ci is connected to the input 20 of the transient detector circuit 2, that is, it is connected to the output 4 of the front stage circuit 1. The other terminal of the first capacitor Ci is connected to an amplifier A1, to a second capacitor C2 and to a reset switch Srs, said amplifier A1, second capacitor C2 and reset switch Srs being arranged in parallel, and connected, at one end, to the first capacitor Ci and, by the

ES 2 811 152 T3 other end, to a common node Diff. Capacities and gain are selected to define a self-balancing, self-timing, capacitor-switched amplifier. Two voltage comparators 7, 8 detect, respectively, up and down changes of the voltage VDf at the common node Diff. The voltage comparators 7, 8 have the common node Diff at their inputs, and their outputs are connected to a logic circuit 9.

The capacitively coupled inverting amplifier A1 amplifies changes in the photoreceptor signal, and these are manifested as a deviation from a defined voltage level (operating point after a reset event) at the Diff node. If the signal at the Diff common node crosses certain adjustable threshold levels, this event is detected by one of two voltage comparators 7, 8, which sends a signal to logic circuit 9, and logic circuit 9 triggers a request signal ( Vreq, rel + or Vreq, rel-, depending on the direction of the detected change).

Upon receipt of the request signal, Vreq, rel + or Vreq, rel-, and the associated pixel data retrieval (see below), an external data receiver (not shown) returns an acknowledgment signal Vack, relay which is converted to a reset signal RST by logic block 9. The reset signal RST controls the reset switch Srs, and said activation closes the reset switch Srs. Consequently, the input node of amplifier A1 is short-circuited with its output and the operating point of amplifier A1 is reset. Subsequently, the request signal is deactivated and the circuit is ready to detect a new change event.

The request signal Vreq, rel + or Vreq, rel- is also used to generate the control signal Vres, abs sent to the exposure measurement circuit 3 in order to start an absolute exposure measurement, making this measurement is conditioned to the prior detection of a change in the illuminance of the pixels, signaled by the transient circuit detector 2. Alternatively, the entire pixel array (1 or 2 dimensions) can be started to run an exposure metering simultaneously on all pixels by an externally applied control signal.

The request signals Vreq, rel + or Vreq, rel- are sent to a bus arbiter (not shown) which initiates and controls the transmission of data packets. In this way, changes in the illumination of the photodiode are detected and, as a consequence, the matrix direction of the respective pixel is transmitted with low latency via an asynchronous data bus (not shown), whereby the coordinates are signaled in the space and (inherently) in time of the detected change. The direction of change (increase or decrease) for each event is determined by the comparator, of the two, that detects the event.

Figure 3 illustrates an improvement of the transient detector circuit 2, whereby a two-stage amplifier is used instead of the capacitively coupled single inverting amplifier whereby the highest sensitivity to time contrast can be achieved. Such a structure is disclosed in the article by C. Posch, D. Matolin and R. Wohlgenannt, “A Two-Stage Capacitive-Feedback Differencing Amplifier for Temporal Contrast IR Sensors,” Analog Integrated Circuits and Signal Processing Journal, vol., 64, no. 1, pp. 45 to 54, 2010.

The two-stage topology corresponding to common source inverting stages with a single terminal, and with capacitive feedback, operating in the subthreshold region and separated by a follower buffer Asf, allows a significant increase in the gain of the amplifier per unit area and leads to reduced charge injection noise (as will be explained later), consequently improving the sensitivity to time contrast of the transient detector circuit.

The first stage has a first capacitor C1 connected by one of its terminals to the input 20 of the transient detector circuit 2.

The other terminal of the first capacitor C1 is connected to a first amplifier A1, a second capacitor C2 and a first reset switch Srs-i, said first amplifier A1, second capacitor C2 and first reset switch Srs being arranged in parallel, and connected at one end to the first capacitor C1 and, at the other end, to a first node Diff-i. In this way, the second capacitor C2 is charged by means of the photoreceptor signal at the output 4 of the front stage circuit 1.

The follower buffer Asf separates the two stages. It is connected, at one end, to the first node Diff1 of the first stage and, at the other end, to a terminal of a third capacitor C3 belonging to the second stage. The other terminal of the third capacitor C3 is connected to a second amplifier A2, to a fourth capacitor C4 and to a second reset switch Srs2, said second amplifier A2, fourth capacitor C4 and second reset switch Srs2 being arranged in parallel, and connected, at one end to the third capacitor C3 and, at the other end, to a second node Diff2. The voltage comparators 7, 8 are connected to the second stage through the second node Diff2. The two voltage comparators 7, 8 are threshold detectors arranged to detect if a voltage on the fourth capacitor C4 exceeds threshold values, and if so, a signal is sent to the control logic module 9, and the logic circuit 9, as described above, it activates a request signal (Vreq, rel + or Vreq, rel-, depending on the direction of the detected change).

ES 2 811 152 T3

As both amplifier stages have similar gain, the load injection in the first stage through the first reset switch Srsi has a greater impact on the amplified signal in the second Diff node.₂ than a charge injection in the second stage through the second reset switch Srs2 To eliminate the effect of the charge injection of the first reset switch Srsi, it is sufficient to ensure that the second stage is activated sufficiently after the first stage . This is achieved by a suitable delay of the switching of the second reset switch Srs2 with respect to the first reset switch Srsi.

Thus, a reset control circuit RCC is provided, which receives the reset signal RST from logic circuit 9 and outputs a first reset signal RSTi controlling the first reset switch Srsi and a second reset signal RST2. which controls the second reset switch Srs2.Thus, The first and second reset signals can be controlled to eliminate load injection due to the first reset switch Srsi.

Exposure measurement

A measurement cycle of the exposure to light of a photodiode by means of the exposure measurement circuit 3 is usually initiated by the detection, by the transient detector circuit 2, of a change in the photoreceptor signal derived from the intensity of the incident light on the PD photodiode. The light exposure of the photodiode PD is measured by determining the time for a voltage on a discharge capacitor C_s of the exposure measurement circuit 3 reaches at least a reference voltage. Before the exposure metering cycle, the metering capacitor Cs of the exposure metering circuit 3 is charged with a voltage corresponding to the photoreceptor signal.

Exposure Metering Circuit - 1^to way of realization

Figure 4 shows a simplified diagram of an example of the exposure measurement circuit 3 for measuring the photoreceptor signal according to a possible embodiment of the invention. The exposure measurement circuit 3 comprises an input 30 connected to the output 4 of the front stage circuit 1 to receive the photoreceptor signal. A measuring capacitor Cs is connected via a first switch Ss to input 30. A unity gain follower buffer 13 may be provided at input 30. The first switch Ss is configured to turn the measuring capacitor Cs on or off with respect to input 30, and is controlled, for this purpose, by a control signal of measurement Vem · The control signal of measurement Vem is obtained by logic block 12 from the control signal V_res, abs sent by the logic module of the control 9 of the transient detector circuit 2 upon detection of a change in the photoreceptor signal.

The measuring capacitor Cs and the first switch Ss are connected through a common node S. The other terminal of the measuring capacitor Cs is connected to ground. Parallel to measuring capacitor C_s, a current source 10 is arranged in series with a second switch S2. The second switch S2 is also controlled by the measurement control signal Vem and is configured to control a discharge of the measurement capacitor C_s. When the second switch S2 does not pass (open state), the current source branch 10 is open, and therefore the measuring capacitor Cs cannot discharge. When the second switch S2 passes (closed state), the current source branch 10 is closed and therefore the measurement capacitor Cs can discharge through this branch. The second switch S2 is grounded, or it can be connected to any current sink. The first and second switches can be implemented as MOS transistors. It will be noted that, for illustrative purposes, the first switch Ss and the second switch S2 are both shown in an open state at the same time, although, during their operation, only one of them is open while the other is closed. The same applies for the switches in Figure 5.

The exposure measurement circuit 3 comprises a voltage comparator 11, which has a signal input connected to one of the terminals of the measurement capacitor Cs and a reference input connected to a reference voltage V_reF. The terminal of the measurement capacitor Cs connected to the measurement comparator 11 is the common node S to which the first switch Ss and the current source 10 are connected. The output of the voltage comparator 11 is fed to a logic circuit 12. The logic circuit 12 is responsible for controlling the state of the exposure metering cycle and the transmission (asynchronous) of the pixel signals, and therefore the result of the exposure metering, to an address encoder. and to a bus referee (not shown).

For the exposure measurement, the instantaneous voltage Vs is used at the common node S, to which the measuring capacitor Cs and the current source 10 are connected. An approximation of the instantaneous voltage Vs can be obtained as

K = ^ (^) + ^ 2

ES 2 811 152 T3 in which l_ph is the intensity of the photocurrent of the photodiode PD of the front stage circuit 1, and k-, yk₂ they are constant factors. The instantaneous value of the voltage Vs is logarithmically related to the instantaneous photocurrent lph, with which a measurement of the voltage Vs allows us to reconstruct the instantaneous photocurrent l_ph, and, therefore, obtain the level of exposure to light from the photodiode PD.

The constants ki and k₂ they depend on details of the circuit implementation as well as individual device parameters which may vary due to non-uniform parameters of the manufacturing process. Consequently, ki and k may₂ are not identical for individual pixel circuits on a matrix (leading to so-called fixed pattern noise, FPN). Preferably ki and k₂ they are determined for each pixel individually, and their influence on the exposure measurement results are eliminated by calibration. Such a calibration can be based, for example, on a homogeneous optical stimulation of the pixel array, or on a uniform stimulation of electrical signals.

Before the start of an exposure metering cycle, the first switch S is closed._s, so that the common node S is connected to the input 30 of the exposure measurement circuit 3. Thus, the voltage V_s at the common node S follows the voltage Vf_rOnt at output 4 of the front stage circuit 1. The voltage across the terminals of the measuring capacitor C_s also follows the voltage Vf_rOnt at the output 4 of the front stage circuit 1, and consequently depends on the light exposure of the photodiode PD.

After the transient detector circuit 2 has detected a relative change in illumination, the exposure measurement circuit 3 receives a measurement control signal V_res, abs, the first initiating an exposure metering cycle.

Upon activation of the control signal V_res, abs, the measurement control signal Vem opens the first switch S_s, thus disconnecting the measurement capacitor Cs with respect to the input 30 of the exposure measurement circuit 3. At that time, the measurement capacitor Cs is charged according to the instantaneous value of the voltage Vs of the common node before opening the first switch Ss. The second switch S₂ can be closed simultaneously by means of the same measurement control signal Vem, or shortly thereafter by means of another signal controlling the second switch S₂.

A reference voltage V is applied_refa a reference input of the voltage comparator 11, the reference voltage V being selected_ref so that the relation V is guaranteed_ref <Vs in all cases. The signal input of the voltage comparator 11 is connected to Vs. Due to the closing of the second switch S₂, the measurement capacitor Cs is discharged by means of a constant current Idee controlled by the current source 10. In this way, the voltage Vs at the signal input of the measurement comparator 11 is reduced, with a rate of decrease that depends of the capacitance of the measuring capacitor Cs and of the intensity of the constant current Idee imposed by the current source 10.

When the voltage Vs at the signal input of the measurement comparator 11 reaches the reference voltage V_ref, the measurement comparator 11 switches, that is, its output changes, and the logic block 12 activates an end-of-measurement signal Vreq.abs The time between the active edges of the control signal V_req, abs and the end-of-measurement signal Vres.abs encodes the mean absolute pixel exposure measurement during this time, according to the relationship

Fs (/_pJ - Vref = '^ T where Idee indicates the intensity of the constant current imposed by the current source 10, C_s is the capacitance of the measuring capacitor Cs, l_ph is the intensity of the photocurrent of the photodiode PD, and T is the time required for the voltage V_s reach the reference voltage V_ref (or, in other words, the time between the active edges of the control signal V_res, abs and the end-of-measurement signal Vreq.abs) From this relationship, and due to the fact that

V_s= ^ (l_ph) + k₂, the intensity l_ph of the photocurrent of the PD photodiode and therefore the light exposure of the pixel.

Similar to the request signals (Vreq, rel + and Vreq.rel-) that are derived from change detection events, the exposure measurement request signals Vreq.abs are sent to a bus arbiter (not shown) that initiates and controls the transmission of data packets. In this way, the matrix address of the respective pixel is transmitted with low latency through an (asynchronous) data bus (not shown), thus signaling the coordinates in space and - inherently - in end-of-time. measurement, whereby the instantaneous gray level value of the pixel is effectively transmitted.

Alternatively, the control logic module 12 may contain a digital counter device that digitizes

ES 2 811 152 T3 directly the time between the activation of the control signal V_res, abs and V exposure metering request signals_req, abs · In this case, said transmitted data packet may contain, in addition to the matrix address of the pixel, the measured gray level digitized by the counter.

Upon deactivation, by the control logic module 12, of the control measurement control signal Vem, the first switch Ss is closed, and the second switch S2 is opened, so that the voltage V_s at the common node S you can resume tracking the photocurrent signal. The Vem signal is deactivated by the control logic module 12 when an external acknowledgment signal V is received._tock, abs · A new exposure metering cycle can be started as soon as it is started by the transient detector circuit 2 or by an external signal.

It will be seen that the injection of charge into the capacitor C_s occurs while opening the first switch S_s, which influences the signal voltage V_s. In order to minimize this charge injection, the measuring capacitor must have a capacity C_s high enough, and techniques for compensation can be used, such as dummy switches and balanced transistor switches or bottom-plate transistor switches.

Exposure Metering Circuit - 2^to way of realization

Fig. 5 shows a simplified diagram and an example of the exposure measurement circuit 3 for measuring the photoreceptor signal according to another possible embodiment of the invention. This embodiment is similar to the embodiment shown in Figure 4 and described above, except that the reference input of the measurement comparator 11 is connected to the reference switch S_ref instead of being directly connected to a reference voltage V_reF. The reference switch S_ref can be operated to connect the reference input of the measurement comparator 11 or to a first reference voltage V_ref, h or well at a second reference voltage V_ref, i. The first reference voltage V_ref, h is greater than the second reference voltage V_ref, i. Logic block 12 controls reference switch S_reF.

As in the embodiment represented in Figure 3, the activation of the measurement control signal Vem by the logic block 12 upon receipt of the control signal V_res, abs opens the first switch S_s, thus disconnecting the measurement capacitor Cs with respect to the exposure measurement input 30. The activation of the control signal V_res, abs also resets logic block 12 to an initial state Zo, and, via logic block 12, controls reference switch S_refso that the first reference voltage V is selected_ref, h for its application to the reference input of the voltage comparator 11.

As previously described, the second switch S2 is closed and the voltage Vs at the signal input of the measurement comparator 11 decreases, with a rate of decrease that depends on the capacity of the measurement capacitor Cs and the intensity of the voltage. constant current Idee imposed by the current source 10. When the voltage V_s reaches the first reference voltage V_refh, the measurement comparator 11 switches, that is, its output changes, and the first end-of-measurement signal V is activated_req, abshLogic block 12 changes to another state Zi, and reference switch S_ref is switched so that the second reference voltage V is selected_ref, i for its application to the reference input of the voltage comparator 11. The discharge of the measurement capacitor Cs continues, and after the time T_ref, the voltage V_s reaches the level of the second reference voltage V_refi, which activates the second end-of-measurement signal V_reF,_tobsi · With activation of the second end-of-measurement signal V_ref, absi, the logic circuit 12 changes to an idle state Z2, the first switch Ss closes, and the second switch S2 opens, so that the voltage V_s You can resume tracking the photocurrent signal. A new exposure measurement cycle can then be started as soon as it is initiated by the transient detector circuit 2 or by an external signal.

The light exposure of the photodiode PD is measured by determining and comparing time spans corresponding to the time for the voltage across the discharge capacitor Cs to reach the first and second reference voltages. During a measurement cycle, a relationship between the different values ​​is:

Idee Ye /, Λ Ye / JY Tref

The first reference voltage V_ref, h and the second reference voltage V_ref, i are external voltages provided to each pixel in the matrix. Consequently, the voltage difference V_ref, h - V_ref, i is the same for each pixel in a matrix. As previously, where T is the time space for the decreasing voltage V_s reach the first reference voltage V_refh, the following relationship still holds:

Vs ^ - V_ref,_h = '- ^ T

Therefore,

ES 2 811 152 T3

Vref, h Vref, l Tref

It will be seen that the exact values ​​of the current Idee imposed by the current source 10 and the capacitance C_sof the measuring capacitor are canceled, and do not influence the determination of the intensity of the photocurrent l_phthat can be deduced. Since the Idee intensity and the capacity C_s may differ from one pixel to another, and have a tendency to be affected by the variation of the parameters of the manufacturing process, a determination of the intensity of the photocurrent is more reliable l_ph independent of these values.

Similar to the first embodiment of the exposure measurement circuit, both signals Vreq.absh and Vreq.absl are sent to a bus arbiter (not shown) that initiates and controls the transmission of data packets. In this way, the matrix address of the respective pixel is transmitted with low latency through an (asynchronous) data bus (not shown), thus signaling the coordinates in space and (inherently) in the time of the first and of the second threshold voltage crosses (signifying the beginning and end, respectively, of an exposure measurement), thereby effectively transmitting the instantaneous value of the pixel gray level.

Again, similar to the first embodiment of the exposure metering circuit, the control logic module 12 may contain a digital counter device that directly digitizes the time between (activation of) the Vreq.absh and Vreq.absl signals. . In this case, only Vreq.absl is sent to the bus arbiter (not shown) which initiates and controls the transmission of data packets, together with the result of said digitization. Thus, the transmitted data packet may contain, in addition to the pixel matrix address, the measured gray level digitized by the counter.

Alternative exposure metering circuit

In an alternative embodiment illustrated by Figure 10, the exposure measurement circuit 3 is constituted by a sample-and-hold circuit 100 that samples the output 4 of the front stage circuit 1 at the moment of activation, by part of the transient detector circuit 2, of the control signal V_res, abs · The output 101 of the sample-and-hold circuit 100 is connected to a voltage analog-to-digital converter (ADC), external to the pixel, 102. One ADC 102 can be provided for each pixel matrix or one ADC 102 for each column of pixels. The output 103 of the ADC 102 is connected to a data bus (not shown).

Similar to previous embodiments, activation of the sample-and-hold circuit 100 is controlled by the control signal V_res, abs of the transient detector circuit 2. After completion of a sampling operation of the instantaneous voltage level at output 4 of the front stage circuit upon receipt of the active control signal V_res, abs, the sampled voltage is sent to the ADC via output 101 for analog-to-digital conversion. After the analog-to-digital conversion is complete, the conversion result along with the pixel matrix address is transmitted to a bus arbiter (not shown) which initiates and controls the transmission of data packets. In this way, the matrix address of the respective pixel and its instantaneous gray level are transmitted with low latency through a (asynchronous) data bus (not shown).

Photoreceptor circuit - prior art embodiment

The photoreceptor circuit 5 of the front stage circuit 1 can be, for example, the one described in the patent US 7,728,269, which is represented in figure 6. This circuit comprises an output 50 to deliver the photoreceptor signal derived from the exposure. in the light of the single photodiode PD, said photoreceptor signal being constituted by a forward voltage Vfront. It also comprises a first photoreceptor transistor Mp1 having a drain and a gate, the gate of said first photoreceptor Mp1 being connected to said output 50. A second Mp2 photoreceptor transistor has its gate connected to the PD photodiode and its source is grounded (that is, connected to a low supply voltage) while its drain is connected to the source of a third Mp3 photoreceptor transistor, whose gate is biased by a bias voltage Vbias.cas. The drain of the third photoreceptor transistor Mp3 is connected to the output 50, therefore also to the gate of the first photoreceptor transistor Mp1. The drain of the third Mp3 photoreceptor transistor is also connected to the drain of a fourth Mp4 photoreceiver transistor, the gate of which is biased by a bias voltage Vbias.pr and whose source is connected to a high supply voltage Vdd. The first Mp1 photoreceptor transistor, the second Mp2 photoreceptor transistor, and the third Mp3 photoreceptor transistor are N-type MOSFETs, while the fourth Mp4 photoreceptor transistor is a P-type MOSFET.

The output voltage Vfront has a logarithmic relationship with respect to the photocurrent l_ph:

Vfront ⁿMpA-V_t.ln t'MpI lph ¡Ο, ΜρΙ

ES 2 811 152 T3 in which

- πμ_ρι is the subthreshold slope factor of the first photoreceptor transistor Mp1,

- V_t is the thermal voltage,

- Lm_pi is the channel length of the first photoreceptor transistor Mp1,

- Wm_pi is the channel width of the first photoreceptor transistor Mp1,

Io, m_pi is the subthreshold saturation current of the first photoreceptor transistor Mp1, and

- Vd is the reverse voltage on the PD photodiode.

A change in photocurrent l_ph of a first value l_phia a second value l_Ph2 causes a AVf change_rOnt of the output voltage Vf_rOnt according to:

Vfront. (Iph2 = n_Mpl.V_t.ln -— \ Jphl

Due to the feedback from the output 50 to the input of the amplifier constituted by the Mp2, Mp3 and Mp4 transistors, the bandwidth of the photoreceptor element is significantly increased compared to logarithmic photoreceptor circuits without feedback. The frequency at 3 dB, which corresponds to the half power point, is roughly calculated as₌^{1 1} 'p * ^{1 1} ’^3dB InC »'V_t~ 2nC_UplfiV_TV ' ^cMpl, to with

- CMpi.a is the capacity between the source of the first photoreceptor transistor Mp1 and the output of the amplifier constituted by the second, third and fourth photoreceptor transistors Mp2, Mp3 and Mp4, which is also the output 50 of the photoreceptor circuit 5 ,

- Cd is the binding capacity of the photodiode PD,

- V_t is the thermal voltage, and

- v is the small signal gain of the amplifier made up of the second, third and fourth photoreceptor transistors Mp2, Mp3 and Mp4.

It will be seen that the capacity CM_Pi, a depends mainly on the gate-source overlap capacity of the first photoreceptor transistor Mp1, which is proportional to the width of its channel. For typical large values ​​of the small signal gain v, the 3 dB frequency of the circuit, compared to a configuration without feedback, is no longer dominated by the PD photodiode capacitance, but by the much smaller gate-overlap capacitance. source of the first photoreceptor transistor Mp1.

Improved photoreceptor circuitry

Figure 7 shows another circuit showing significant improvements with respect to gain and bandwidth. The improved circuit is also a logarithmic, time-continuous photoreceptor circuit with feedback to improve bandwidth, but, compared to the photoreceptor circuit of Fig. 6, a fifth additional Mp5 photoreceptor transistor has been arranged between the photodiode. PD and the first photoreceptor transistor Mp1.

Accordingly, the improved photoreceptor circuit 5 of Figure 7 comprises an output 50 for delivering the photoreceptor signal derived from exposure to light from said single photodiode, said photoreceptor signal being constituted by a voltage Vfront. It comprises a first photoreceptor transistor Mp1 having a drain and a gate, the gate of said first photoreceptor Mp1 being connected to said output 50. It also comprises a fifth Mp5 photoreceptor transistor having a drain, a source and a gate, the source of said fifth Mp5 photoreceptor transistor being connected to the single PD photodiode and the gate of the fifth Mp5 photoreceptor transistor being biased by a bias voltage Vbias.d. The first photoreceptor transistor Mp1 and the fifth photoreceptor transistor Mp5 have a common source.

The other photoreceptor transistors are arranged in a similar manner to the circuit of Figure 6, constituting an amplifier. A second photoreceptor transistor Mp2 has its gate connected to photodiode PD and to the drain of the fifth photoreceptor transistor MP5. Its source is grounded (that is, connected

ES 2 811 152 T3 at low supply voltage) while its drain is connected to the source of a third MP3 photoreceptor transistor, the gate of which is biased by a bias voltage Vbias.cas. The drain of the third photoreceptor transistor Mp3 is connected to the output 50, therefore also to the gate of the first photoreceptor transistor Mp1. The drain of the third Mp3 photoreceptor transistor is also connected to the drain of a fourth Mp4 photoreceiver transistor, the gate of which is biased by a bias voltage Vbias.pr and whose source is connected to a high supply voltage Vdd. The first Mp1 photoreceptor transistor, the second Mp2 photoreceptor transistor, and the third Mp3 photoreceptor transistor are N-type MOSFETs, while the fifth Mp5 photoreceptor transistor and the fourth Mp4 photoreceptor transistor are P-type MOSFETs.

Relative to the gain increase, the output voltage Vf_rOnt of the circuit of Figure 6 continues to depend logarithmically on the intensity of photocurrent l_ph:

,,, (lph. \ ,,, / ^ Mp5 lph \, ,,

Vfront ^ Mpl · ^ t · I .... IT n ^ p ^. Vj-. In í 1 + ^ bias, d yV'Mpl 'O.MplJ \^vvMpS 'O.MpSj with

- πμ_Ρ5 is the subthreshold slope factor of the fifth photoreceptor transistor Mp5,

- Lm_P5 is the channel length of the first photoreceptor transistor Mp5,

- Wm_ps is the channel width of the first photoreceptor transistor Mp5, lo, M_P5 is the subthreshold saturation current of the first photoreceptor transistor Mp5, and

- Vb¡as, d the bias voltage applied to the gate of the fifth photoreceptor transistor Mp5.

A change in photocurrent l_ph of a first value l_phia a second value l_Ph2 causes a AVf change_rOnt of the output voltage Vf_rOnt according to:

..., ilph2 \,. ilph2 \

AVfront ^ Mpi · 14 · I j T ríMps · Lt · In (7) \ 'phl / \' phl /

Assuming that πμ_ρι “πμ_Ρ5, can be simplified to:

ÚV_front «2.n_Mpl.V_c.Zn (^) \ 'ph.l /

Consequently, the gain is doubled relative to the photoreceptor circuit of Figure 6.

The increase in gain achieved by the photoreceptor circuit of figure 7 makes it possible to detect changes in current intensity ΔΙ_ρsmaller κ of photocurrent l_ph since, in response to a certain change ΔΙ_ρκ of the intensity of the photocurrent l_ph, the resulting voltage change is increased by a doubled gain. Furthermore, since the resulting voltage change increases before the input of a successive amplifier, such as the capacitor-switched differentiation amplifier of the transient detector circuit 3, the gain of said successive amplifier can be lower while achieving the same sensitivity. general to temporal contrast (i.e. relative change), which leads to a significant reduction in the size of CMOS devices, especially the size of the capacitors of a capacitor-switched amplifier.

With reference to the bandwidth, the frequency at 3 dB, which corresponds to the half power point, is roughly calculated as f ~ 1 i^ph

J3dB 'ir p' ϊ / ^ MpS.DS ^vt in which Cm_ps, ds is the drain-source coupling capacity of the fifth Mp5 transistor. In general, this capacity is significantly less than the door-source overlap capacity Cm_pi of the first photoreceptor transistor Mp1 that was used to determine the frequency at 3 dB in the circuit of Figure 6. Therefore, the bandwidth increases according to the relation Cm_P5, ds / Cm_pi.

Due to the increase in the bandwidth of the photoreceptor circuit 5, the response delay of the transient detector circuit 2 is significantly reduced, and the temporal resolution of the pixel circuit and therefore of the sensor device is improved.

Alternatively, the gate of the fifth photoreceptor transistor Mp5 can be connected to its source instead of being driven by a bias voltage Vbias.d. However, in this configuration, although the voltage gain continues to double, there is no increase in bandwidth.

ES 2 811 152 T3

Gain stage

Figure 8 shows a simplified diagram of an example of a preamplifier gain stage 6 for amplifying the Vfront signal at the output of the photoreceptor circuit 5, instead of a conventional follower buffer.

The preamplifier gain stage 6 comprises an input connected to the output of the photoreceptor circuit 5 to receive the photoreceptor signal Vfront, and an output connected both to the input of the transient detector circuit 2 and to the input of the exposure measurement circuit 3, to deliver the amplified photoreceptor signal Vamp. The preamplifier gain stage 6 comprises a first gain transistor Mg1 having a drain, a source and a gate. The gate of the first gain transistor Mg1 is connected to the input of the gain stage, that is, to the output of the photoreceptor circuit 5. The source of the first gain transistor Mg1 is connected to a reference bias voltage Vbias.ref and the drain of said first gain transistor Mg1 is connected to the output of the gain stage. The first gain transistor Mg1 is an N-channel type MOS transistor.

The preamplifier gain stage 6 also comprises a plurality of Mg2, Mgk gain transistors in series, each of these Mg2, Mgk gain transistors presenting a drain, a source and a gate, and in such a way that each of these transistors of Mg2 gain, Mgk in series has its drain connected to its gate (transistors connected as diode). One of these series gain transistors has its drain connected to the drain of the first Mg1 gain transistor and is referred to as Mg2. Therefore, this Mg2 gain transistor has its gate connected to the gain stage 6 output. The series gain transistors are P-channel type MOS transistors.

The gain stage 6 described is a common source amplifier, the first gain transistor Mg1 being an input transistor N-MOS and so that the series of gain transistors P-MOS connected as diode Mg2, Mgk are a load. Such a structure with a load connected as a diode has a gain that does not depend considerably on dimensions, so that the influence of mismatch related to dimensions is reduced, which improves the performance of fixed pattern noise (FPN ) of the pixel array.

At least two gain transistors are provided in series connected as diode Mg2, Mgk, that is k = 3. Preferably, there are more diode-connected gain transistors arranged in series. In preferred embodiments, three or four diode-connected gain transistors are connected in series. The maximum number of such diode-connected gain transistors that are arranged in series is determined by the excursion of the input voltage, that is, the expected upper level of the input voltage Vfront such that the gate-source voltage over the series of Transistors connected as diode Mg2 to Mgk are not limited by the resulting output voltage Vamp.

If the gain stage circuit 6 is operated in the subthreshold region, and assuming the same dimensions for the series gain transistors connected as diode Mg2 to Mgk, the output voltage Vamp is calculated as follows:

„,, z, z. _T,. (^ Mg2 /o.Evil \ Ά z,.

Vamp - Vqd ~ (Jt ~ V) .n_Mg2.V_t. \ n I— - - I - (/ c - 1). (Vfront^— Vbías.re /) \^vvMg2 ^LMgl '0, Mg2 / ⁿMgl

In which

- Vdd is the high supply voltage,

- πμ₉ι and πμ₉two are the subthreshold slope factors of the gain transistors Mg1 and Mg2, respectively,

- V_t is the thermal voltage,

- Lm₉i and Lm₉two are the channel lengths of the first gain transistor Mg1 and the second gain transistors Mg2, respectively,

- WMgi and W_Mgtwo are the channel widths of the first gain transistor Mg1 and the second gain transistors Mg2, respectively, lo.Mgi and lo, M_gtwo are the subthreshold saturation currents of the first gain transistor Mg1 and the second gain transistors Mg2, respectively.

A change in input voltage AVf_rOnt results in AV output voltage change_A.Mp from:

ES 2 811 152 T3

Δν_απιρ = - (k - l) ^ ÚV_frontⁿMgl

The gain thus provided by the gain stage 6 makes it possible to detect smaller changes in current intensity ΔΙρκ of the photocurrent l_ph since, in response to a certain change ΔΙ_ρκ of the intensity of the photocurrent l_ph, the resulting voltage change is increased by a gain of (k-1) times the original. Furthermore, since the resulting voltage change increases before the input of a successive amplifier, such as the capacitor-switched differentiation amplifier of the transient detector circuit 3, the gain of said successive amplifier can be lower while reaching the same sensitivity. full contrast, which leads to a significant reduction in the size of CMOS devices, especially the size of the capacitors of a capacitor-switched amplifier.

By using a gain stage 6 according to Fig. 8, connected to the output 50 of a photoreceptor circuit 5 as shown in Figs. 6 or 7, a limitation of the bandwidth depending on the photocurrent can be achieved. The gain stage then provides an automatic noise reduction, controlled by photocurrent, by self-adjusting its bandwidth. The frequency at 3 dB, which corresponds to the half power point, is proportional to the current l_Mgi flowing through the first gain transistor Mg1:

so that the current l_Mgia across the first gain transistor Mg1 depends on the voltage difference Vfront - Vbias.ret · With the first gain transistor Mg1 operating in the subthreshold region, the frequency at 3 dB is proportional to:

ΙΛ / Vfront Vbias.ref f ^ dB ~ 7 ^ W e '-'Mgl

With transistors of the same size for the first photoreceptor transistor Mg1 of the photoreceptor circuit 5 as in Figures 6 and 7 and for the first gain transistor Mg1 of the gain stage 6, and with an equal bias reference voltage Vbias.ret to the inverse voltage Vd on the photodiode PD, it follows that the frequency at 3 dB is proportional to the photocurrent l_ph:

fzdB ~ Iph

Since the reverse voltage Vd on the photodiode PD is approximately independent of the photocurrent l_ph, the bias reference voltage Vbias.ret can be obtained globally for all the pixels of a matrix using a dummy circuit with a photodiode covered in a photoreceptor circuit such as that of figure 6 or 7, using the voltage of output of said photoreceptor circuit as bias reference voltage Vbias.ret ·

In this way, the gain stage circuit 6 can substitute for a conventional source follower to effectively decouple the sensitive front stage circuit 1 from subsequent circuits, that is, the transient detector circuit 2 and the transient measurement circuit. exposure 3, and photocurrent dependent bandwidth control and additional signal amplification can be achieved.

Figure 9 shows an image sensor 90 comprising a plurality of pixel circuits 91 in accordance with possible embodiments of the invention as described above. The pixel circuits 91 shown are arranged in a matrix.