Method and system to enhance differential dynamic range and signal/noise in CMOS range finding systems using differential sensors

[0001] Priority is claimed from co-pending U.S. utility patent application Ser. No. 11/350,209 filed 8 Feb. 2006, soon to issue as U.S. Pat. No. 7,157,685 entitled Method and System to Enhance Differential Dynamic Range and Signal/Noise in CMOS Range Finding Systems Using Differential Sensors. Parent application Ser. No. 11/350,209 claimed priority from then co-pending U.S. provisional patent application Nos. 60/650,900 filed 8 Feb. 2005 entitled Improved Op Amp Assisted CMR, Ser. No. 60/650,901 filed 8 Feb. 2005 entitled High Active Light Dynamic Range for 3-D Imaging, and Ser. No. 60/650,902 filed 8 Feb. 2005 entitled A Postiori VGA Gain Setting. Further, parent application Ser. No. 11/350,209 was a continuation-in-part from then co-pending U.S. patent application Ser. No. 11/110,982 filed 19 Apr. 2005, entitled Method and System to Differentially Enhance Sensor Dynamic Range Using Enhanced Common Mode Reset, which application was a continuation in-part of then co-pending U.S. utility application entitled Method and System to Differentially Enhance Sensor Dynamic Range, application Ser. No. 10/823,415, filed 12 Apr. 2004, now U.S. Pat. No. 6,919,549 (2005) Method and System to Differentially Enhance Sensor Dynamic Range.

[0002] The invention relates generally to solid state optical range finding sensing systems, and more particularly to enhancing the dynamic range and signal/noise ratio in CMOS differential sensors used in such systems.

[0003] Systems that rely upon sensing optical energy to discern information are known in the art and have many applications. Exemplary applications might include an optical-based system to determine range between the system and a target object, or to identify and recognize features of a target object. Many such systems acquire two-dimensional or intensity-based information, and rely upon an intensity image of light reflected from a target object. Such luminosity-based systems can use ambient light falling upon the target object, or may actively generate light that is directed toward the target object.

[0004] Unfortunately, it is difficult to accurately determine distance solely from the amplitude and brightness of an intensity image. For example, in a range finding system, a highly reflecting target object that is father away from the system can produce a greater amplitude signal than a nearer target object that is less reflective. The result would be that the more distant, shiny, object is erroneously reported as being closer to the system than the closer, duller, object. In a range finding system used to control robot machinery in an industrial setting, such errors may be intolerable for reasons of safety to nearby human operators. If such a system is used to identify and recognize different target objects, an object might be misidentified. Simply stated, two-dimensional intensity-based systems are very prone to measurement error.

[0005] U.S. Pat. No. 6,323,942 to Bamji et al. (November 2001) entitled “CMOS-Compatible Three-Dimensional Image Sensor IC” describes a three-dimensional range finding system that can determine range distance without reliance upon luminosity-based data, the entire content of which patent is incorporated herein by this reference. As disclosed in the '942 patent, such a system generates a depth map that contains the distance Z from each pixel in a CMOS-compatible sensor array to a corresponding location on a target object.

[0006] FIG. 1A is a block diagram of a three-dimensional range finding system 10 as exemplified by the '942 patent. Such systems determine distance Z between the system and locations on target object 20 by determining the amount of time for a light pulse to be emitted by the system, to reflect off the target object, and be detected by the system. Such systems commonly are referred to as time-of-flight or TOF systems. System 10 may be fabricated upon a single IC 30, requires no moving parts, and relatively few off-chip components, primarily a source of optical energy 40, e.g., a light emitting diode (LED) or laser source, and associated optics 50. If desired, laser source 40 might be bonded onto the common substrate upon which IC 30 is fabricated.

[0007] System 10 includes an array 60 of pixel detectors 70, each of which has dedicated circuitry 80 for processing detection charge output by the associated detector. At times herein, the terms “detector”, “photodiode detector” (because of its somewhat equivalent function), “photodetector”, “pixel” and “pixel detector” may be used interchangeably. More rigorously, the term “photodetector” may be reserved for the single-ended or more preferably differential photodetectors, e.g., the semiconductor devices that output detection current in response to incoming detected optical energy. In the spirit of such more rigorous definition, “pixel” or “pixel detector” would refer to the dedicated electronics associated with each single-ended or differential photodetector. In other usages, “pixel” may refer to the combination of a photodetector and it dedicated electronics. Using this terminology, array 60 might include 100×100 photodetectors 70, and 100×100 associated detector processing circuits or pixels 80, although other configurations may be used. IC 30 preferably also includes a microprocessor or microcontroller unit 90, RAM and ROM memory, collectively 100, a high-speed distributable clock 110, and various computing and input/output (I/O) circuitry 120. System 10 includes analog-to-digital conversion functions, and for purposes of the present invention, let it be understood that such functions are subsumed within I/O circuitry 120 as are some video gain functions. System 10 preferably further includes a lens 130 to focus light reflected from target object 20 upon pixels 70 in array 60. Controller unit 90 may carry out distance-to-object and object velocity calculations and can output such calculations as DATA, for use by a companion device, if desired. As seen in FIG. 1A, substantially all of system 10 may be fabricated upon CMOS IC 30, which enables shorter signal paths, and reduced processing and delay times. Also shown in FIG. 1A is ambient light that is present in the environment in which system 10 and target object 20 are found. As described herein, high levels of ambient light relative to levels of light from energy source 40 can be detrimental to reliable operation of system 10.

[0008] In brief, microprocessor 90 can calculate the roundtrip time for optical energy from source 40 to travel to target object 20 and be reflected back to a pixel 70 within array 60. This time-of-flight (TOF) is given by the following relationship:
Z=C·t/2 where C is velocity of light.  eq. (1)

[0009] Thus, without reliance upon luminosity information, system 10 can calculate that Z1=C·t1/2, Z2=C·t2/2, Z2=C·t3/2, and so on. The correct Z distances are obtained, even if more distant regions of target object 20 happen to be more reflective than nearer regions of the target object.

[0010] The ability of system 10 to determine proper TOF distances Z can be impacted when the magnitude of ambient light is large relative to the magnitude of reflected light from source 40. What occurs is that the various pixels 70 respond to incoming optical energy that represents the real signal to be measured (e.g., active energy originating from source 40 and reflected by target object 20), and also respond to ambient light. The depth resolution of each pixel, i.e., the accuracy of the distance measurement, is determined by the system signal-to-noise ratio (S/N). Even if ambient light could be measured and subtracted from the total signal, its noise component (e.g., shot noise) would still degrade system performance. Further, the presence of ambient light can have even more severe consequences by causing the pixel detector to saturate.

[0011] A differential pixel photodetector is a detector that receives two input parameters and responds to their difference. With reference to TOF type systems, the active optical energy emitted by the system contributes to both a differential mode signal and a common mode signal, while ambient light contributes only to the common mode signal. Differential pixel detectors can exhibit higher signal-to-noise ratio than single-ended pixel detectors. However the presence of strong ambient light, sunlight perhaps, can degrade the performance of differential pixel detectors.

[0012] Differential pixel photodetectors will now be described with reference to U.S. Pat. No. 6,580,496 to Bamji et al. (June 2003) entitled “Systems for CMOS-Compatible Three-Dimensional Image Sensing Using Quantum Efficiency Modulation”. The '496 patent describes the use of quantum efficiency modulation techniques and differential detectors suitable for a three-dimensional range finding systems. The quantum efficiency of the substrate upon which differential CMOS sensors were fabricated was modulated synchronously with the active optical energy emitted from an energy source. Relative phase (Φ) shift between the transmitted light signals and signals reflected from the target object was examined to acquire distance z. Detection of the reflected light signals over multiple locations in the pixel array resulted in measurement signals referred to as depth images.

[0013] FIG. 1B depicts a system 100 such as described in the '496 parent, in which an oscillator 115 is controllable by microprocessor 160 to emit high frequency (perhaps 200 MHz) component periodic signals, ideally representable as A·cos(ωt). Emitter 120 transmitted optical energy having low average and peak power in the tens of mW range, which emitted signals permitted use of inexpensive light sources and simpler, narrower bandwidth (e.g., a few hundred KHz) pixel photodiode detectors (or simply, photodetectors) 140′. System 100, most of which may be implemented upon a CMOS IC 30′ will also include an array 130′ of differential pixel photodetectors 70 and associated dedicated electronics 80. It will be appreciated that optical energy impinging upon array 130′ includes a fraction of the emitted optical energy that is reflected by a target object 20, which reflected energy is modulated, and also includes undesired ambient light, which is not modulated. Unless otherwise noted, elements in FIG. 1B with like reference numerals to elements in FIG. 1A may be understood to refer to similar or identical elements.

[0014] In system 100′ there will be a phase shift Φ due to the time-of-flight (TOF) required for energy transmitted by emitter 120 (S₁=cos(ωt)) to traverse distance z to target object 20, and the return energy detected by a photo detector 140′ in array 130′, S₂=A·cos(ωt+Φ), where A represents brightness of the detected reflected signal and may be measured separately using the same return signal that is received by the pixel detector. FIGS. 1C and 1D depict the relationship between phase shift Φ and time-of-flight, again assuming for ease of description a sinusoidal waveform. The period for the waveforms of FIGS. 1C and 1D is T=2π/ω.

[0015] The phase shift Φ due to time-of-flight is:
Φ=2·ω·z/C=2·(2πf)·z/C

[0016] where C is the speed of light 300,000 Km/sec. Thus, distance z from energy emitter (and from detector array) to the target object is given by:
z=Φ·C/2ω=Φ·C/{2·(2πf)}

[0017] Various techniques for acquiring and processing three dimensional imaging have been developed by assignee herein Canesta, Inc. of Sunnyvale, Calif. For example, U.S. Pat. No. 6,906,793 (2005) to Bamji et al. describes Methods and Devices for Charge Management for Three-Dimensional Sensing, U.S. Pat. No. 6,522,395 (2003) to Bamji et al. discloses Noise Reduction Techniques Suitable for Three-Dimensional Information Acquirable with CMOS-Compatible Image Sensor ICs; and U.S. Pat. No. 6,512,838 to Rafii et al. (2003) discloses Methods for Enabling Performance and Data Acquired from Three-Dimensional Image Systems. But it still remains a challenge to provide a TOF system with differential pixel photodetectors that are protected from saturation, including saturation from differential mode signals, while enhancing signal/noise ratios.

[0018] It is useful at this juncture to review prior art implementations for differential pixel photodetectors. Such review will provide a better understanding of the challenges presented in protecting differential pixel photodetectors against saturation, while trying to enhance signal/noise ratios. In the '496 patent, differential detectors responded to amplitude of incoming optical energy and to phase of such energy relative to energy output by emitter 40. A comparison of FIG. 1C and 1D indicates the nature of the shift in phase (Φ).

[0019] Referring now to FIG. 2A, the singular term “pixel” is sometimes used collectively to refer to a pair of differential photodetectors, for example first and second photodiode detectors D_A and D_B as well as at least a portion of their dedicated electronics. With this understanding, what is shown in FIG. 2A is a pair 70 of pixel photodetectors, hundred(s) of which can comprise an array 130′, as suggested by FIG. 1B. Incoming optical energy falling upon a pixel detector 70 generates an extremely small amount of photocurrent (or photocharge), typically on the order of picoamps (10⁻¹² amps). Such detection current signals are too small in magnitude to be measured directly. Pixel detectors can function in a direct integration mode in which optical energy induced photocurrent is integrated. Integration can result using an integration capacitor, where the final capacitor charge or voltage is readout at the end of an integration interval. A capacitor C_x has finite maximum charge capacity Q_max defined by:
Q_max=C_x·V_swing  eq. (2)
where C_x is the total capacitance and V_swing is the maximum voltage swing across the capacitor. A pixel photodetector is said to be in saturation when the total charge integrated on the capacitor exceeds the maximum charge capacity, in which case no useful information can be readout from that pixel photodetector.

[0020] A differential pixel photodetector (e.g., detectors 70 in FIG. 1B) may be represented as shown generically in FIG. 2A, in which modulation circuitry has been omitted for simplicity. Each pixel photodetector 70 has a differential structure with two perhaps identical reset and readout circuit components denoted A and B. Components A and B may be considered as part of the pixel photodetector 70 or as part of the pixel's associated circuitry 80. For ease of depictions, the photodetector pair comprising each differential pixel 70 is shown as photodiodes D_A and D_B, but other detector structures could be used instead, for example photogate structures. Capacitors C_A and C_B are shown in parallel with diodes D_A and D_B and represent detector parasitic capacitance and/or dedicated fixed value capacitors.

[0021] Referring briefly to FIG. 1B, within system 100 microprocessor 160 commands generator 115 to cause optical energy source 120 to emit pulses of light that are directed by lens 50 toward target object 20. Some of this optical energy will be reflected back towards system 100 and will be focused by lens 135 onto pixel photodetectors 70 within array 130. Incoming photon energy falling upon a detector 70 will cause photodetector pair D_A and D_B to generate a small amount of detection signal current that can be directly integrated by capacitors C_A and C_B. Before the start of integration, microprocessor 90, which may (but need not be) implemented on IC chip 30, will cause photodetectors D_A and D_B and their respective capacitors C_A and C_B to be reset to a reference voltage V_ref. For the components shown in FIG. 2A, reset is caused by raising a reset signal Φ_reset (see FIG. 2B). During the integration time, photocurrent generated by detectors D_A and D_B respectively discharge associated capacitors C_A, C_B, as shown in FIG. 2B. During the integration time, the voltage seen at nodes S_A, S_B will decrease as a function of the photocurrent generated by the associated photodiode D_A, D_B. The magnitude of the photodiode-generated photocurrent will be a function of the amount of light energy received by the respective pixel 70 in array 60 in that the amount of light received by the pixel determines the final voltage on nodes S_A and S_B.

[0022] Readout circuitry is provided for circuit A and B, comprising transistors T_follower and T_read. At the end of the integration time, which will be a function of the repetition rate of the optical pulses emitted from optical energy source 40, microprocessor 90 causes a readout signal Φ_read to go high. This enables the voltages on nodes S_A and S_B to be read-out of array 60, e.g., through a bitline. In the exemplary configuration of FIG. 2A, if the voltage on node S_A or S_B drops below a certain level denoted here as saturation voltage V_sat, the readout circuit cannot perform the reading operation properly. Therefore the dynamic range of such known differential pixel configuration shown in FIG. 2A is (V_ref−V_sat), as depicted in FIG. 2B. While the waveforms in FIG. 2B depict a diminishing potential at nodes S_A, S_B as a function of photocurrent, one could instead configure the detector circuitry to charge rather than discharge a reference node potential.

[0023] But in addition to generating photocurrent in response to optical energy or active light (from emitter 40) reflected by target object 20, pixel 70 will also generate photocurrent in response to ambient light that is also integrated by capacitors C_A, C_B, thus affecting the potential at nodes S_A, S_B. FIG. 2B depicts two examples, showing the effect of relatively low magnitude ambient light and relatively high magnitude of ambient light. In range finding applications, the difference (A_final−B_final) generally contains range information, and common mode is of lesser importance. As shown in FIG. 2B, relatively weak ambient light does not cause the pixel to saturate, and at the end of integration time, the final voltages read-out from the pixel are above V_sat. But relatively strong ambient light discharges the associated capacitor potential rapidly, which saturates the pixel. Due to the saturation condition, the pixel does not output any useful result in that the differential voltage, which contained range information, is now zero. Thus, a very real problem with prior differential pixel detectors is that the dynamic range of the pixel is not sufficient to handle strong ambient light.

[0024] Thus, whereas CMOS sensors used in systems to acquire images generally rely upon strong levels of ambient light, CMOS sensors used in time-of-flight systems seek to reduce the effects of ambient light. As seen in FIG. 2B, the magnitude of ambient light can overwhelm detection of reflected optical energy, saturating the detectors. Image acquisition systems and time-of-flight systems that must function in environments exposed to strong ambient light or minimal ambient light may require a sensor dynamic range exceeding about 100 dB. In time-of-flight and similar applications in which ambient light is unnecessary, the detection effects of ambient light can be substantially reduced electronically.

[0025] There is a need for a method and topology by which the dynamic range of a differential pixel detector can be enhanced without sacrificing a substantial portion of the desired differential signal. Preferably saturation of the differential pixel detector should be substantially eliminated, even from high magnitudes of the desired differential signal. Further, signal/noise ratio for the detection signal path should be enhanced. These goals preferably should be met using additional circuitry that can function with existing detector circuitry and that can be implemented to fit within the perhaps 50 μm×50 μm area of a pixel differential photodetector.

[0026] Embodiments of the present invention provide such methods and circuit topologies.

[0027] Effective differential dynamic range in a differential pixel photodetector was increased in embodiments of priority application Ser. No. 10/823,415, now U.S. Pat. No. 6,919,549 by avoiding common mode contribution saturation. Photocurrent generated by each photodetector pair was directly capacitor integrated over an integration time T such that resultant capacitor voltage was proportional to detected optical energy including ambient light. Within time T, before either integrated capacitor voltage reached V_sat for the photodetector, at least one of the capacitors was reset to a voltage V_ref such that the desired differential detector signal was still determinable. After reset, capacitor voltages were again allowed to change as a function of photocurrent, and were reset before either capacitor voltage reached V_sat.

[0028] In another embodiment of what is now the '549 patent, the common mode component of the integrated capacitor voltages was reset periodically to prevent either photodiode detector from saturating. However the differential component of the integrated capacitor voltages was preserved. The result was to extend effective differential dynamic range of the differential sensor in the presence of ambient light by avoiding the effects of saturation.

[0029] In parent co-pending application Ser. No. 11/110,982 filed 19 Apr. 2005, further improvements are described for differential pixels, e.g., the circuitry associated with each pixel differential photodetector. More specifically, the '982 application describes pixels with improved common mode rejection, and improved retention of the desired differential signal component, accompanied by acceptably low KT/C noise due to capacitance. In one embodiment, signals to the pixel photodetector differential inputs are coupled as input to an operational amplifier and to a differential signal capacitor, configured as an integrator. During integration, charge from the detectors is accumulated in their integration capacitors. During a charge dump operation, at least half the total differential detection signal charge from the integration capacitors is read into at least one differential signal capacitor. Such charge dumping periodically transfers the differential detector signal into the differential signal capacitor(s) for storage. Thus during common mode resets, which usually follow dump operations, the differential detection signal on the integration capacitor is essentially zero, thus reducing the chance of differential charge loss. The desired result is enhanced common mode rejection with relatively little loss in differential detection signal.

[0030] The present invention provides still further improved differential pixels, especially with respect to avoiding saturating even with relatively large amplitude differential signals, and to enhancing signal/noise ratio for the detection signal path. In some embodiments, a fixed compensating offset (ΔV) is added to the differential signal capacitor voltage whenever magnitude of the differential signal exceeds a predetermined maximum or minimum value. The offset ΔV is negative if the differential signal capacitor voltage has become too positive, and the offset ΔV is positive if the capacitor voltage has become too negative. In some embodiments, the accumulated charge voltage on the differential signal capacitor is checked synchronously, at which time ΔV is added, if needed. A count is kept of the number (N) of ΔV offsets that had to be added, and effective differential signal capacitor voltage is actual output voltage across the capacitor (Vo)+N·ΔV. In other embodiments, reset of the integration capacitor voltage is asynchronous, and occurs whenever the voltage exceeds a predetermined maximum or minimum threshold. Again a count of the number (N) of resets is kept, and effective differential signal capacitor voltage is Vo+N·ΔV. These embodiments preserve the desired differential signal and prevent saturation of the differential pixel even when the differential signal is large in amplitude. If desired, the compensating signal ΔV could of course be ΔQ, where Q is charge. Saturation due to common mode signal is prevented, preferably using embodiments of the above-referenced co-pending application Ser. No. 11/110,982. If desired, magnitude of the compensating offsets could of course be varied, i.e., a positive Δ compensating offset need not be identical in magnitude to a negative Δ compensating offset, where the offset may be voltage or charge. Additionally, in a sequence of Δ compensating offsets, the magnitude of each offset could, if desired, be different. In such embodiment, the sum of the compensating offsets would be recorded.

[0031] Further embodiments of the present invention enhance differential pixel signal/noise ratio by dynamically maximizing gain on the operational amplifier associated with each differential pixel. Because such operational amplifiers are found early in the detection signal paths, high amplifier gain (A_G) reduces effective noise contribution downstream in the signal path by 1/Ag. Gain of each such amplifier is variably controlled to adjust A_G individually for each pixel as a function of its present signal value. Within the array of differential pixels, each amplifier is first operated at maximum A_G, and integration capacitor values are readout and stored in a row buffer. A_G for each amplifier in the row is then incrementally decreased, and the row buffer is updated only for those amplifiers whose associated integration capacitor is not presently saturated. The above process is repeated until the value in the row buffer corresponds to the highest non-saturating gain for each amplifier associated with the row. The row buffer also records the value of the highest non-saturating gain for each amplifier associated with the pixels in that row. At this juncture the row buffer is readout, and the process is repeated for the next row in the array, and so on continuously. In this fashion amplifier values of A_G are individually maximized, commensurate with avoiding overload or saturation of components downstream in the signal path. The desired result is enhanced signal/noise ratio. Alternative embodiments can, of course, increment rather than decrement amplifier gain or even randomly or otherwise scan or vary A_G and cause the row buffer to latch the non-saturated gain value for each amplifier associated with pixels in a row.

[0032] Embodiments of the present invention use much circuitry already in place with respect to enhancing common mode rejection and retaining desired differential signal components. As a result, implementing the present invention can be carried out within the form factor of pixel differential photodetectors.

[0033] Other features and advantages of the invention will appear from the following description in which the preferred embodiments have been set forth in detail, in conjunction with the accompanying drawings.

[0072] Before describing the present invention, whose description commences with FIG. 14, it is useful to first describe embodiments of U.S. Pat. No. 6,919,549, from which priority is claimed. Description of the '549 patent will be made with reference to FIGS. 3A-10B. Next a description of priority application Ser. No. 11/110,982 will be made with reference to FIGS. 11, 12, and 13. An understanding of the operation of pixel differential photodetectors and the evolution of their associated electronic circuitry (sometimes referred to herein as differential pixel) will now be described with reference to FIGS. 3A-13, to enable a better understanding of the present invention.

[0073] FIG. 3A, taken from the '549 patent, depicts one-half of differential pixel detector 7C′, where it is understood that TOF system 100 shown in FIG. 1B might now employ an array 60′ of rows and columns of differential pixel detectors 70′ as will now be described, rather than pixel detectors 70. In FIG. 3A, only one of the two pixels is shown for ease of illustration, namely photodetector D_A (denoted PD D_A). Associated with each photodetector in the pixel is a capacitor, C_A being associated with D_A, where C_A can be the capacitance inherent with D_A, and/or a discrete capacitor. In a conventional configuration, in the presence of a signal, typically ambient light, the voltage across C_A would decrease until a saturation voltage V_sat was attained, at which point an output signal from D_A would be meaningless.

[0074] But as shown by the waveforms in FIG. 3B, the voltage at node S_A, e.g., the voltage across C_A and across D_A, is prevented from exceeding V_sat by resetting the node voltage to a fixed reference V_ref whenever V_sat is attained. Each differential pixel detector 70′ includes two photodiodes and two capacitors, and each capacitor-photodiode node is independently reset to V_ref, as soon as the voltage across either capacitor reaches V_sat.

[0075] In FIG. 3A, a comparator 140 compares the voltage signal from photodiode D_A present at node S_A to V_ref. As soon as the S_A potential reaches V_ref, comparator 140 changes state, going from low-to-high for the configuration shown. Thus when VS_A>V_sat, the output from comparator 140 turns-on a reset transistor T_reset coupled between node S_A and V_ref. The potential VS_A at node S_A is reset by being pulled from V_sat to V_ref. The desired result is that overall dynamic range of pixel detector 70′ is increased.

[0076] As shown in FIG. 3A, output from comparator 140 (node P) is also input to a counter 150 that essentially will count the number of resets that occur for the detector. It is understood that as FIG. 3A depicts half of a differential pixel detector, there will be two comparators, two counters, and two sets of switching transistors for each differential pixel detector 70′. The photodiode signal at node S_A is coupled via a high input impedance voltage follower transistor T_follower, whose output is read via a bitline when a Φ_readA signal goes high (for the configuration shown). An additional row selection transistor T_readD is coupled between the output from counter 150 and the bitline signal, and is turned on when a Φ_readD signal goes high (for the configuration shown). Note that a feedback path exists between the comparator output and the gate for reset transistor T_reset. Those skilled in the art will appreciate that means other than the above described solid state switches, comparators, counters, etc., may be used.

[0077] Referring briefly to FIG. 1B, optical energy source 120 typically outputs a pulse train of optical energy, which energy may be modulated, for example according to the '496 patent. The pulse train will have a period between adjacent output pulses. Within differential detector 70′, the maximum period of integration is made less than the period between adjacent pulses of optical energy emitted by source 120 Referring back to FIGS. 3A and 3B, in the present of strong ambient light, sunlight perhaps, during integration the voltage at node S_A across C_A continues to decrease in magnitude until V_sat is reached. At that moment, comparator 140 changes states, emitting a short output pulse that is present at node P. This pulse turns-on reset transistor T_reset for a short time, causing C_A to be again reset to voltage V_ref. Such reset is self-triggering, and can occur multiple times during the integration interval. The total number of such resets is recorded by counter 150, there being one counter for each of the two photodiode detectors in a differential pixel detector 70′.

[0078] At the end of the integration time, the counter value (n) and the final voltage V_final on capacitor C_A are read-out separately by turning-on T_readD and T_readA, respectively. FIG. 3A is conceptual in that while counter 150 is shown being read-out as though its counter n were an analog value, in practice the digital counter will be read-out with a bus. The signal waveforms for node S_A and the comparator output at node P are shown in FIG. 3B. Note that the effective voltage swing on node S_A is V_swing=n(V_ref−V_sat)+V_final, which is n times larger than the maximum voltage swing (V_ref−V_sat) of known differential sensors. Thus using the self-resetting configuration of FIG. 3A, capacity is extended by n times, where n is the number of self-resets occurring during integration. The resultant extended maximum charge capacity for the photodiode enables the pixel sensor to detect differential mode signals even in the present of very strong ambient light.

[0079] While FIG. 3A has been described with respect to use of a counter 150 that operates digitally, the role of counter 150 can instead be implemented in analog fashion. FIG. 3C depicts such an implementation, in which an analog charge pump products an analog voltage value proportional to n. Such an analog circuit can be implemented using small area on an IC, e.g., IC 30′ in FIG. 1B, where conventional detectors 70 are replaced by detectors 70′ according to the '549 patent.

[0080] In FIG. 3C, a current source changes voltage across a capacitor C_r, where each time a reset pulse (of fixed duration) is generated by comparator 140, the current source is turned on. Thus for each comparator reset pulse, a fixed amount of charge is injected into capacitor C_r, altering the voltage across the capacitor by ΔV_r. At the end of the integration time, the voltage on capacitor Cr changes by an amount equal to ΔV_r times the number of reset pulses n that occurred during integration. It is possible to determine n by n=(V_ref−V_r)/ΔV_r. Note that capacitor C_r is also initialized to V_ref when the photodetector is initialized. If desired, an initial voltage other than V_ref could be used for capacitor C_r.

[0081] In FIG. 3C, photodetector D_A can be reset by using the non-inverting input of comparator 140, which input normally is set to V_sat. But this non-inverting input can be used to perform an initial (frame) reset before integration. For example, during the initial reset period this input can be switched to V_DD, which will cause the comparator to output a pulse at node P that resets T_reset and thus resets photodetector D_A and its associated capacitor C_A. Thereafter the non-inverting node of comparator 140 can be returned to V_sat to remain at that potential until the next (frame) reset. By judiciously making voltage V_sat low, all photodetectors are simultaneously reset, thus removing the need for a separate reset signal.

[0082] Turning now to FIGS. 4A and 4B, an embodiment of a differential comparator is described in which a differential time to saturation counter is employed. In the block diagram of FIG. 4A a single detector PD D_A is shown, although it is understood that a complete differential pixel detector 70′ will comprise two detector diodes (or the like), two comparators, a counter, and associated reset and read-out transistors. In FIG. 4A, while counter 150′ is shown implemented with analog components, a counter could instead be implemented to function digitally.

[0083] At start of integration, counter 150′ starts counting to measure the time since the last reset to V_ref. In the following discussion it will be assumed that the magnitude of incoming ambient light does not change substantially during the integration time. Two scenarios will be considered: ambient light is strong, and ambient light is not very strong.

[0084] If the ambient light is strong, each photodetector D_A and D_B (D_B not shown) will reach V_sat multiple times during one integration period. In this case, every time either. photodetector reaches V_sat, the photodetectors and counter 150′ are simultaneously reset. At the end of the integration period, each photodetector will be at a determinable voltage level, which level will likely be different for each detector. Further, at the end of the integration period, the counter will have stored the time (Δtf) since the last reset required for the photodetectors to attain these final voltage levels. Since each photodetector end-of-integration voltage level is known, as is the time Δtf, the slope of the voltage curves for each photodetector and the number of resets that should have occurred during integration can be determined; see FIG. 4B. Note at the right-hand side of FIG. 4B that the final photodiode voltages are (V_ref−ΔV_af) and (V_ref−ΔV_bf) for photodiodes D_A and D_B respectively. Subtracting these magnitudes from V_ref yields ΔV_af and ΔV_bf. The total swing can be calculated as follows:
V_swing-a=ΔV_af·T/Δtf  eq. (3)
V_swing-b=ΔV_bf·T/Δtf  eq. (4)
where T is the total integration time, which is known.

[0085] If the ambient light is not strong, at the end of the integration time T, the counter value will be equal to the integration time, and the voltages across the photodiodes D_A and D_B will represent all that the photocharge each pixel could collect during the whole integration time. In this case, no further calculation is needed to determine the total voltages, since it follows from T=Δtf that:
V_swing-a=ΔV_af  eq. (5)
V_swing-b=ΔV_bf  eq. (6)

[0086] Once the relevant times and photodiode voltages are read-out, an external computation unit, perhaps microprocessor 90 in system 10 (see FIG. 1) or pure logic circuitry can calculate to provide the differential signal.

[0087] As noted, in FIG. 4A an analog equivalent of a digital resettable counter 150′ is used, in which a charge pump can be used to measure Δtf. A charge pump capacitor C_r is reset by transistor Tc_rreset each time photodiode voltage reaches V_sat. Thus at the end of time T, the voltage on C_r is proportional to the time from the last reset to the end of integration. Assuming the constant current supplied to C_r is I_r, then the final voltage on C_r will be V_r=V_ref−I_r·Δtf/C_r and Δtf can be determined as Δtf=(V_ref−V_r)·C_r/I_r. In the above descriptions, two independent counters are present. However since only the counter causing the most recent reset is of use, both counters can be combined into a single counter. This single counter and the reset for both halves of the pixel detector could be controlled by the logical OR of both comparators.

[0088] As noted, capacitor C_r is initialized at the beginning of integration along with the photodetector. As with the self-reset method, the non-inverting input of comparator 140 may be switched to V_DD (rather than to V_sat during integration) to reset pixel 70′ (e.g., both photodetectors and C_r). By judiciously making voltage V_sat low, all photodiodes are simultaneously reset, thus removing the need for a separate reset signal.

[0089] Turning now to FIGS. 5A and 5B an embodiment is shown in which differential pixel detector 70′ includes a voltage controlled charge pump 160 to record voltage difference between the two differential photodetectors D_A and D_B before they are saturated by high common mode signal. Once the photodetector voltage difference has been recorded on a third capacitor C_r, photodetectors D_A and D_B and their associated capacitors (C_A, C_B) can be reset by the relevant reset transistors, and integration starts again.

[0090] In FIG. 5A, charge pump 160 includes two voltage controlled current sources I_s1, I_s2 and a capacitor C_r. Although the voltage difference between photodetectors D_A and D_B may be monitored continuously by differential amplifier 170, charge pump 160 is only turned-on for a fixed period of time. In this fashion, charge accumulated on capacitor C_r during each sample is proportional to the voltage difference. After each sample, the differential photodetector is reset and a new integration cycle starts. The sampling frequency preferably depends upon incoming light intensity, e.g., optical energy falling upon photodetectors D_A and D_B, which intensity can be estimated. The final charge on C_r is the summation of the samples and is proportional to the total voltage difference between the differential photodetectors. FIG. 5B depicts control signal and voltage waveforms at various nodes in the configuration of FIG. 5A.

[0091] At the end of integration, three voltage values are read-out from pixel 70′, namely V_a (voltage across D_A), V_b (voltage across D_B), and V_r (voltage across capacitor C_r). Given these quantities, one can calculate the resulting differential voltage V_swing-a−V_swing-b as follows. Looking at FIG. 5B, it is seen that ΔV_af=V_ref−V_a and that ΔV_bf=V_ref−V_b. The differential voltage is then given by V_swing-a−V_swing-b=ΔV_af−ΔV_bf+f(V_r), where f(V_r) is a linear function of V_r.

[0092] This linear function f(V_r) is obtained by writing V_r as V_r=V_ref+n·k·(V_a−V_b). As noted, V_ref is the initial voltage for capacitor C_r (e.g., the reset voltage for photodetectors D_A, D_B), n is the number of sample/reset cycles, and k is a constant determined from the circuit of FIG. 4A, and represents how much voltage change occurs on C_r for given a unit voltage change in (V_a−V_b). The amount contributing to V_swing-a−V_swing-b is n·(V_a−V_b), which is equal to n·(V_a−V_b)=(V_r−V_ref)/k=f(V_r). In summary the final differential voltage is calculated from known quantities, according to V_swing-a−V_swing-b=V_b−V_a+(V_r−V_ref)/k. Common mode voltage can also be estimated from ΔV_af and ΔV_bf since the time between the last reset and the end of integration is known.

[0093] It I s possible to automatically generate the Δ_sample or Δ_reset signals within each pixel 70′, by providing some additional circuitry. One can use two comparators to compare S_A, S_B node potentials with V_sat, and the logical OR can be taken of the comparator outputs to yield the Δ_sample signal. Thus as soon as either photodiode potential reaches V_sat, Δ_sample goes high. If desired, Δ_reset can be a delayed version of Δ_sample. Such a self-resetting configuration would use two comparators, an OR gate, and a delay element that could be a simple RC delay.

[0094] Turning now to FIG. 6A, an embodiment of a differential pixel detector 70′ is shown in which during reset operation capacitors acquire exactly the same charge in each half of the configuration. By adding exactly the same charge to each half of the configuration, common mode contribution is essentially removed and differential mode contribution is preserved. Such an approach offers several advantages. For example, extra resets do not affect the system operation, and the pixel detector may be reset even if it is not discharged. Further, capacitor or component mismatch has substantially no effect on the accuracy of the reset. In addition, it can be shown that common mode reset generates no KT/C noise in the differential domain. The only resulting KT/C contribution appears in common mode where it is unimportant.

[0095] For ease of illustration, FIG. 6A does not depict Q_A, which is the sum of the charge on the top plate of capacitors C_A and CD_A, or Q_B, which is the sum of the charge on in the top plate of capacitors C_B and CD_DB. In operation, the configuration of FIG. 6A preserves the differential quantity Q_A−Q_B during the common mode reset operation, although the common mode quantity, (Q_A+Q_B)/2, is changed at each reset. What occurs is that after a reset, the quantity (Q_A+Q_B)/2 is moved closer to some constant Q_reset0. Thus in contrast to other reset approaches, additional resets have no adverse impact in FIG. 6A as they simply move the operating point for (Q_A+Q_B)/2 even closer to Q_reset0.

[0096] In normal operation switching transistors T_swA and T_swB are open, while transitors T_vrefA and T_vrefB, and T_disA and T_disB are closed. Initially photodetector diodes D_A and D_B are reset to V_ref via transistors T_resetA and T_resetB, but during integration transistors T_resetA and T_resetB remain open. As optical energy impinges on photodiode D_A it discharges its parasitic capacitor C_DA as well as capacitor C_A, while photodiode D_B discharges its parasitic capacitor C_DB as well as its capacitor C_B. Initial reset is achieved though transistors T_resetA and T_resetB, which initialize the circuit at potential V_ref, although other potential levels may instead be used.

[0097] During common mode reset, signal Φ_resetA remains low, while the other control signals operate as shown in FIG. 6B. During operation of pixel 70′, the following steps take place:

[0098] (1) First, capacitors C_A and C_B are decoupled from associated photodiodes D_A and D_B by bring the Φ_dis signal low, which opens discharge transistors T_disA and T_disB going low. This operation does not change the differential charge quantity Q_A−Q_B, and no KT/C noise is introduced on Q_A−Q_B.

[0099] (2) Next, when control signal Φ_norm goes low, the bottom plates of capacitors C_A and C_B are decoupled from V_ref by the opening of transistors T_refA and T_refB. The quantity Q_A−Q_B remains unaffected, even in terms of KT/C.

[0100] (3) When the control signal Φ_switch goes high, capacitors C_A and C_B redistribute their charge. Let Q_CA be the charge on capacitor C_A and let Q_CB be the charge on capacitor C_B. If capacitors C_A and C_B are now shorted together in parallel the total charge quantity Q_CA+Q_CB would be preserved. However since C_B is connected to C_A inverted, the quantity Q_CA−Q_CB is preserved during this operation. Since no switches are opened no KT/C arises from this step.

[0101] (4) When the control signal Φ_swA goes low; a KT/C uncertainty appears in the charge redistribution, but this KT/C appears as common mode on the quantities Q_CA and Q_CB. Any charge taken from Q_CA shows up exactly on Q_CB but with a minus sign. Thus after switches T_swA and T_swB open, Q_CA′=Q_CA+Noise_KTC, and −Q_CB′=−Q_CB′−Noise_KTC. Thus there is no KT/C noise on Q_CA′−Q_CB′=Q_CA−Q_CB+(Noise_KTC−Noise_KTC).

[0102] (5) Next, when control signal Φ_norm goes high again the differential charge Q_A−Q_B is unchanged.

[0103] (6) Finally, when control signal Φ_{dis goes} high, Q_A−Q_B is unchanged.

[0104] If desired some of the above steps may be combined for simplicity. For example steps 5 and 6 may occur simultaneously or even out of sequence. Steps 1,2 and 5,6 clearly do not affect Q_A−Q_B, and it was demonstrated that steps 3 and 4 do not affect Q_CA−Q_CB. Thus, steps 1 through 6 do not affect Q_A−Q_B. In operation, the generation of the various control signals can be handled by microprocessor 160 in system 10 in FIG. 1B, where it is understood that array will comprise differential pixels 70′.

[0105] Note that no assumption as to closeness of component matching was made in the above analysis, and the conclusions reached remain irrespective of the values of capacitors C_A, C_B, CD_A, and CD_B. Additionally the linearity of the capacitors does not affect performance, and the configuration shown in FIG. 6A will function with capacitors that are mismatched or even nonlinear.

[0106] Consider now the common mode voltage on photodiodes D_A and D_B after reset. Within reasonable bounds, the exact value of the common mode voltage is not critical. Although an analysis can be carried out for almost any capacitor values, for simplicity of explanation let it be assumed that C_A=C_B and C_DA=C_DB.

[0107] Since it is small, let KT/C reset noise be ignored. Thus after step 5 if V_CA (the voltage across capacitor C_A) and if V_CB (the voltage across capacitor C_B) have the relation V_CA=−V_CB, the voltage on the top plate of C_A is (V_ref+V_CA) and the voltage on the top plate of C_B is (V_ref−V_CA).

[0108] As noted, charge difference Q_A−Q_B is preserved in the configuration of FIG. 6A. Assuming that C_A=C_A and C_DA=C_DB it can be shown that the voltage difference V_DA−V_DB on the photodiodes is also preserved, and that is voltage is V_DA −V_DB=2V_CA=−2V_CB, after step 5. Thus following step 5, V_DA and V_DB must differ from the top plate voltages on capacitors C_A and C_B by only a constant K.
V_DA=V_ref+V_CA+K  eq. (7)
V_DB=V_ref+V_CB+K=V_ref−V_CA+K  eq. (8)
V_DA+V_DB=V_ref+V_CA+K+V_ref−V_CA+K=2(V_ref+K)  eq. (9)

[0109] After step 6 because of charge redistribution the new voltages are:

[0110] $\begin{array}{cc}{V}_{\mathrm{DA}}^{\prime }=\left[C_{\mathrm{DA}}·V_{\mathrm{DA}}+C_A·\left(V_{\mathrm{ref}}+V_{\mathrm{CA}}\right)\right]/\left(C_{\mathrm{DA}}+C_A\right)& \mathrm{eq}.\left(10\right)\\ =\left[\begin{array}{c}{C}_{\mathrm{DA}}·\left(V_{\mathrm{ref}}+V_{\mathrm{CA}}+K\right)+\\ C_A·\left(V_{\mathrm{ref}}+V_{\mathrm{CA}}\right)\end{array}\right]/\left(C_{\mathrm{DA}}+C_A\right)& \mathrm{eq}.\left(11\right)\\ =\left[\left(V_{\mathrm{ref}}+V_{\mathrm{CA}}\right)+C_{\mathrm{DA}}·K/\left(C_{\mathrm{DA}}+C_A\right)\right],\mathrm{and}& \mathrm{eq}.\left(12\right)\\ V_{\mathrm{DB}}^{\prime }=\left[C_{\mathrm{DB}}·V_{\mathrm{DB}}+C_B·\left(V_{\mathrm{ref}}+V_{\mathrm{CB}}\right)\right]/\left(C_{\mathrm{DB}}+C_B\right)& \mathrm{eq}.\left(13\right)\\ =\left[\begin{array}{c}{C}_{\mathrm{DB}}·\left(V_{\mathrm{ref}}+V_{\mathrm{CB}}+K\right)+\\ C_B·\left(V_{\mathrm{ref}}+V_{\mathrm{CB}}\right)\end{array}\right]/\left(C_{\mathrm{DB}}+C_B\right)& \mathrm{eq}.\left(14\right)\\ =\left[\left(V_{\mathrm{ref}}+V_{\mathrm{CB}}\right)+C_{\mathrm{DB}}·K/\left(C_{\mathrm{DB}}+C_B\right)\right]& \mathrm{eq}.\left(15\right)\\ =\left[\left(V_{\mathrm{ref}}-V_{\mathrm{CA}}\right)+C_{\mathrm{DA}}·K/\left(C_{\mathrm{DA}}+C_A\right)\right]& \mathrm{eq}.\left(16\right)\end{array}$
Thus the sum V_DA′+V_DB′ then becomes
V_DA′+V_DB′=[(V_ref+V_CA)+C_DA·K/(C_DA+C_A)]+[(V_ref−V_CA)+C_DA·K/(C_DA+C_A)]=2·[V_ref+K·C_DA/(C_DA+C_A)]  eq. (17)
Thus the sum V_DA′+V_DB′ is advantageously always closer to ²V_ref than to V_DA+V_DB. This demonstrates that with each reset in FIG. 6A, the common mode is brought closer to V_ref by K·[1−C_DA/(C_DA+C_A)]=K·C_A/(C_DA+C_A).

[0111] To recapitulate, for the embodiment of FIG. 6A, reset operation has the desired effect of centering the common mode about V_ref. Relevant waveforms for FIG. 6A are shown in FIG. 6C. As a consequence, a reset can be applied without consideration of over-saturating or under-saturating the common mode for the pixel configuration. Thus in normal operation, reset can occur as often as desired without concern as to ill effects resulting from over or under saturation of the common mode.

[0112] Transistors T_disA and T_disB can be used as global shutters, thereby improving resilience to ambient light by stopping the effects of all light impinging on the differential pixel when the shutter is turned off. When T_disA and T_disB are off, capacitors C_a and C_b are decoupled from photodetectors PD_DA and PD_DB and therefore stop integrating the signal from PD_DA and PD_DB. If the output of the pixel is chosen to be top plate of capacitors C_A and C_B then the output of the pixel will be frozen after T_disA and T_disB are turned-off, thereby providing the function of a global shutter.

[0113] FIG. 6D depicts another embodiment of a capacitor common mode reset configuration for pixel 70′, whose basic operation is as described for the configuration of FIG. 6A. However, in FIG. 6D, initialization of voltages V_DA and V_DB across photodiodes D_A, D_B respectively at the beginning of integration does not involve transistors T_resetA and T_resetB as was the case for the configuration of FIG. 6A. Instead, in FIG. 6D, reset is achieved by simultaneously turning-on transistors T_disA and T_disB with high control signals Φ_dis, turning-on transistors T_swA and T_swB with high control signal Φ_sw, and by turning-off transistors T_VrefA and T_VrefB with low control signal Φ_norm. This has the effect of resetting photodetectors PDD_A and PDD_B to V_ref. Note that transistors T_disA and T_disB may be used as global shutters in this configuration.

[0114] FIG. 6E depicts yet another embodiment for pixel 70′, wherein discharge transistors T_disA and T_disB are eliminated. Indeed these discharge transistors could also be removed from the configurations of FIG. 6A and FIG. 6D. While these alternative configurations reduced common mode, unfortunately detector performance is diminished. This degradation results as each reset reduces some of the differential mode signal, and after a usually small number of resets, the differential signal is lost as a function of C_DA/C_A and C_DB/C_B. Such embodiments may still find use in applications that do not require high precision, or where the number of resets is low, or where C_DA<<C_A and C_DB<<C_B.

[0115] Turning now to FIGS. 7A and 7B, a configuration and waveforms for a differential pixel detector 70′ is shown in which a charge integrator is used for differential signal integration and common mode reset. Integration is carried out by integrator 180 and integration capacitor C_int. During an initial frame reset, transistors controlled by Φ_reset, Φ_r, and Φ_int signals are all turned-on, and the voltages on photodetectors D_A and D_B are reset to V_ref.

[0116] During integration, transistors controlled by signal Φ_int are turned-on and transistors controlled by signals Φ_r, Φ_reset are turned-off. Thus during integration, light induced photocurrent discharges photodiode parasitic capacitance C_DA, C_DB, as well as capacitors C_A and C_B. As has been noted, the integrated signals contain both differential mode and common mode components. FIG. 7B depicts various control voltage waveforms used in the embodiment of FIG. 7A.

[0117] Next, in a common mode reset phase, control signal Φ_int goes low, causing transistors T_int to decouple C_A from C_DA and to decouple C_B from C_DB. Then control signal Φ_r goes high, turning-on transistors T_r and charge in both C_A and C_B transfers to the integration capacitor C_int. Note that polarities of the charge transferred onto C_int are opposite due to the arrangement of the T_r switches.

[0118] The integrated charge on C_int after the common mode reset can be expressed in terms of charge on C_DA, C_DB before the common mode reset as:
Q_int=Q_CDA−Q_CDB  eq. (18)

[0119] Therefore the common mode signal is cancelled while the differential signal is preserved, which is the desired result.

[0120] Preferably common mode reset is performed multiple times and is interleaved with the integration during the whole frame integration. During integration, the integrating operational amplifier 180 may be turned off to save power. The total number of common mode reset performed will depend on the intensity of ambient light. The final signal readout is the accumulated charge (hence voltage) on C_int.

[0121] While the charge integrator in the embodiment of FIG. 7A used a single-ended operational amplifier 180, a fully differential operational amplifier could be used, among other types of integration configurations.

[0122] In the embodiment of FIG. 7A, each differential pixel (e.g., each D_A and D_B photodiode pair) has its own integrator, e.g., 180. In an alternative embodiment, one can implement only integration capacitor C_int within each pixel, and share operational amplifier 180 as well as switches associated with connecting C_int to amplifier 180 among multiple pixels. This approach would require fewer transistors per pixel, allowing a higher fill factor to be achieved on the integrated circuit containing the detector system.

[0123] Turning now to FIG. 8A, an embodiment of a differential detector 170′ is shown in which the common mode reset circuitry compensates for potential mismatch between components such as mismatched detector area between D_A and D_B, mismatched tolerance between capacitors C_A and C_B, as well as mismatched transistor sizes.

[0124] FIG. 8B depicts control waveforms found in an alternate embodiment of FIG. 8A. In this alternate embodiment, as shown in FIG. 8B, the phase of the optical energy waveform from emitter 40 (LIGHT) alternates between 0° and 180° with respect to the phase of a signal used to modulate the photodetectors. Thus, rather than use fixed charge transfer polarity as in the embodiment of FIG. 7B, polarity between D_A and D_B is switched synchronously in time with modulation of the light emitted from system 10 towards target object 20. The accumulated charge on integration capacitor C_int at the end of frame integration is expressed as:

[0125] $\begin{array}{cc}{Q}_{\mathrm{int}}=\left(Q_{A,0}^1-Q_{B,0}^1\right)+\left(Q_{A,180}^1-Q_{B,180}^1\right)+\left(Q_{A,0}^2-Q_{B,0}^2\right)+\left(Q_{A,180}^2-Q_{B,180}^2\right)+\dots =\sum _{i=1}^{n/2}\left(Q_{A,0}^i-Q_{B,0}^i\right)+\sum _{i=1}^{n/2}\left(Q_{A,180}^i-Q_{B,180}^i\right)& \mathrm{eq}.\left(19\right)\end{array}$
where Q_A,0 represents the charge collected by detector D_A with respect to 0° light phase, Q_B,0 represents the charge collected by detector D_B with respect to 0° light phase, Q_A,180 represents the charge collected by detector D_A with respect to 180° light phase, and Q_B,180 represents the charge collected by detector D_B with respect to 180° light phase. As is apparent from the above equation, if the total number of common mode resets is n, then half of the final differential charge is collected from detector D_A and half of the charge is collected from detector D_B. Another advantage of this embodiment is that KT/C noise associated with the transistor switches becomes common mode; therefore such noise is cancelled out from the final differential signal value.

[0126] Most of the above-described embodiments accumulate the D_A−D_B charge difference in a capacitor, and periodically reset D_A and D_B to V_ref to avoid saturation. FIG. 9A depicts an alternative approach in which potentials V_a and V_b are increased by a fixed amount ΔV before these potentials drop below a certain level due to high-common mode light. This approach is somewhat analogous to the capacitor common mode reset embodiments that have been described. However the embodiment of FIG. 9A uses a separate circuit with an external current source 190.

[0127] In the embodiment of FIG. 9A, a periodic injection of a fixed amount of charge into detectors D_A and D_B occurs. The result is that while the differential (D_A−D_B) charge does not change, the common mode of D_A and D_B is refreshed (i.e., decreased) to prevent photodetector saturation. An external current source 190 is required, which current source may be a reference Voltage Controlled Current Source (VCCS) or perhaps a reference Constant Current Controlled Current source (CCCS), in which case the current source becomes a current mirror.

[0128] The embodiment of FIG. 9A and the embodiment of FIG. 9B demonstrate two approaches to periodically refreshing charge into detectors D_A and D_B. In FIG. 9A, current source 190 is always on, but switches Tsw responsive to an external signal Φ_XC are used to couple the constant current output by source 190 to nodes S_A and S_B. During integration Φ_XC is periodically turned-on for a brief period of time to charge-up nodes S_A and S_B, hundreds of nanoseconds perhaps.

[0129] In the embodiment of FIG. 9B, switches T_sw are eliminated and instead the input current or voltage to current source 190 is itself pulsed. In essence rather than import a signal Φ_XC, Φ_XC pulses are imported into current source 190 and result in current pulses of constant amplitude as shown.

[0130] FIG. 9C depicts waveforms for the configurations of FIGS. 9A and 9B. Note that advantageously the final differential voltage is simply (V_A−V_B) and that no other computation need be done. The rate at which Φ_XC or the reference input to current count 190 will depend upon the common mode ambient light. A higher rate would be called for in the presence of very strong ambient light to keep source nodes S_A and S_B from saturating.

[0131] As noted, current source 190 may be controlled using a voltage reference or a current reference. If a voltage reference is used, the voltage can be V_DD or V_ref, in which case only the Φ_XC signal would be needed to perform common-mode removal. For CCCS (or current mirrors) a variety of circuit configurations are structures are available, two of which are shown in FIGS. 9D and 9E. While the configuration of FIG. 9D has fewer components, its current output may suffer from nonlinearity caused by transistor channel length modulation. The configuration of FIG. 9E provides a cascoded current mirror that is insensitive to voltage at the current output nodes. The choice of current source configuration involves a tradeoff between circuit complexity and accuracy.

[0132] An important concern in implementing current mirrors is matching, in this case between output currents A and B. To minimize matching errors, transistors with large values of width and length should be used and should be carefully laid out. For example the orientation of transistors should be the same for currents A and B.

[0133] FIG. 10A depicts a shunt capacitor embodiment that periodically injects a certain amount of charge into photodetector D_A and D_B to compensate for the common mode. FIG. 10A depicts one-half of such a circuit, while FIG. 10B depicts control signal waveforms. A, preferably, very small capacitor C_charge is initially charged-up to a relatively high voltage. When charge signal Φ_charge goes low and reset signal Φ_reset goes high, C_charge and C_A are connected, and most of the charge will be transferred to C_A since its capacitance is much higher than C_charge. Since the same amount of charge is added into both halves of the configuration, the common mode signal is reset while the differential mode signal is preserved.

[0134] A description of the embodiments of parent application Ser. No. 11/110,982 will now be given with reference to FIGS. 11-13. In overview, new generations of CMOS sensors such as SmallCam, Pixim, can permit sensor operation at high dynamic range levels. For use in CMOS image sensing, ambient light is usually the parameter to be measured, and for which a high dynamic sensor range is needed. But for three-dimensional TOF sensing using a modulated optical energy source such as 120 in FIG. 1B, strong ambient light, a common mode signal, is detrimental. Unless the effects of ambient light are substantially reduced or eliminated, saturation of the different pixel photodetectors can result. In practice, TOF pixel photodetectors may have to operate within a large dynamic range of perhaps 100 dB.

[0135] As noted, the embodiments of FIGS. 3A-10B, taken from the '549 patent, disclose the use of common mode reset to reduce the effects of ambient light and dark current, both of which are common mode parameters. The challenge of course is to preserve all desired differential pixel detection signal values while resetting all common mode signal components to a fixed value. Preferably such results occur with little or no noise uncertainty from KT/C noise resulting from capacitors, save for parasitic capacitances.

[0136] In various aspects, embodiments of FIGS. 11-13 can improve upon the performance of embodiments described in the '549 patent, especially with regard to preserving more of the desired differential signal over each common mode reset. Referring now to FIG. 11, taken from application Ser. No. 11/110,982, a pixel differential photodetector 70′ with common mode reset circuitry is shown. In FIG. 11 (as well as FIG. 12), components other than differential detector 70′ may be collectively designated as dedicated per-pixel detector electronics 80′, preferably implemented on IC chip 30′. Both halves of the full differential pixel detector are shown in FIG. 11, and it is understood that pixel detector 70′ is preferably one of many detectors in an array 60′, such as shown in FIG. 1B. The configurations of FIG. 11 and FIG. 12 may be operated with timing waveforms such as shown in FIG. 6B for common mode reset operations.

[0137] In FIG. 11, clock signals, e.g., CLKA, CLKB, CLKC, are generated from a clock driver system, preferably implemented on IC 30′ (see FIG. 1B) and are coupled to detector 70′. It is understood that that other detector types using different clocking structures may be used. Further information as to clocking schemes may be found in U.S. Pat. No. 6,906,793 (2005) entitled Method and Devices for Charge Management for Three-Dimensional Sensing.

[0138] In FIG. 11, capacitors C_A and C_B are charge storage capacitors and typically are about 60 fF. Switch transistors T_refA, T_refB, responsive to a Φ_norm signal, couple a known reference potential Vref to capacitor bottom nodes R_A and R_B. Reset transistors TresA, TresB, couple or de-couple nodes R_A and R_B to capacitor upper nodes O_A and O_B, respectively. Reset signals Φ_sw selectively close resets transistors T_resetA and T_resetB, which cause nodes O_A and R_B or nodes O_B and R_A to be connected to each other. Shutter transistors T_shutA and T_shutB are responsive to shutter signals Φ_ds that when active couple the respective “A” and “B” outputs from differential detector 70′ to nodes O_A and O_B Transistors T_fA and T_fB are source followers that deliver the detector “A” or detector “B” portions of differential detector 70′ signals to BITLINE A or BITLINE B via read transistors T_rA and T_rB. (see FIG. 13).

[0139] Those skilled in the art will recognize that associated with nodes O_A, O_B, R_A, and R_B will be parasitic capacitance, not shown in FIG. 11. One undesired effect of parasitic capacitance at these nodes is that instead of providing 100% of the differential detector signal to BITLINE A and BITLINE B, in practice only about 97% of the differential detector signal is preserved after each common mode reset operation. As a result, after multiple common mode resets, a relatively smaller fraction of the differential signal is available. For example, after about thirty common mode resets, only about 40% of the differential signal will remain. This results follows from (0.97)³⁰. Although the final signal/noise ratio is better than if only 40% of the incoming optical energy could be detected and collected, the 3% signal loss nonetheless degrades the overall signal/noise ratio. The parasitic capacitance of the detector results in elimination of only about 70% of the common mode signal (e.g., ambient light component).

[0140] This unremedied differential signal degradation results from undesired charge sharing between the integration capacitors C_A and C_B and the parasitic capacitance associated with detector 70′. What occurs is that when shutter transistors T_shutA, T_shutB open during common mode reset, some charge is left in the detector capacitance and for this residual charge, common mode is not eliminated. This in turn requires more common mode reset cycles than would otherwise be necessary, which additional reset cycles result in further loss of the desired differential detector signal. Further the presence of residual common mode signal means that detector 70′ operates at a lower magnitude voltage within its dynamic range than would otherwise be the case. The configuration of FIG. 11 will introduce KT/C noise but only for the parasitic capacitances on nodes O_A, O_B, R_A, R_B. However since the parasitic capacitances are small (perhaps on the order of 1 fF), the KT/C noise is small and is generally less than the shot noise of the detector.

[0141] FIG. 12 presents an embodiment using additional common mode reset circuitry 200, that improves differential loss over the embodiment of FIG. 11, while enjoying the same relatively low KT/C noise characteristics. Within circuitry 200, operational amplifier 210 functions to remove differential detector signal charge from nodes O_A, O_B and to store the removed charge in an integration capacitor. With this functionality, the embodiment of FIG. 12 preserves the stored removed charge without substantial loss due to subsequent common mode reset cycles.

[0142] In practice, at least a substantial fraction (e.g., ≧50%) of the stored charge is removed for isolated storage on each so-called dump cycle. It will be appreciated that the system could function with less than 50% stored charge removal per cycle, however overall more differential charge will be lost during common mode reset operations. The result from saving at least a substantial fraction of the differential charge is improved dynamic range, improved retention of the desired differential detector signal, and improved common mode rejection. Components shown in FIG. 12 other than differential detector 70′ may be designated as electronics 80′, preferably dedicated to each pixel detector 70, and CMOS-implemented on the common IC 30′.

[0143] During an integration period T, operation of the embodiment exemplified by FIG. 12 involves a number n of common mode reset operations, and a number x of dumps (transfers-out) of the differential charge from capacitors C_A, C_B associated with each differential pixel detector into an active integration capacitor C220. Once dumped, the differential charge is stored in capacitor C220 and is not affected by subsequent common mode resets. More than one integration capacitor may be used, and within an integration period T, the number of dumps x may be less than or equal to or even greater than n. However in a preferred mode of operation, there will be a dump or transfer-out of differential charge before a common mode reset.

[0144] During common mode reset operation, the differential detector signal charge is first read transferred into the integration capacitor C229 by turning-on dump transistors T_dA, T_dA′. So doing dumps charge from capacitor C_A node O_A and from capacitor C_B node O_B respectively into the non-inverting and inverting inputs of operational amplifier 210. Shutter transistors T_shutA and T_shutB remain open, which allows even the differential detector charge to be transferred. Subsequent common mode resets will have no effect on this safely stored-away differential detector and capacitor C_A and C_B charge. Next, shutter transistors T_shutA and T_shutB and dump transistors T_dA, T_dA′ are opened, and common mode reset is performed.

[0145] Whereas the embodiment of FIG. 11 preserved perhaps 97% of the pixel differential photodetector signal charge, the embodiment of FIG. 12 can retain as much as 99.5% of this charge, a substantial improvement. Furthermore, the 0.5% or so charge loss that occurs will be substantially independent of the number n of common mode rejection cycles. The KT/C noise characteristics and removal of common mode by common mode reset for the embodiment of FIG. 12 will be similar to that of the configuration of FIG. 11.

[0146] Amplifier 210 provides a single-ended output signal (AMP OUT) that could be used to directly drive a bitline (BITLINE B) without use of a source follower such as T_fB in FIG. 12. Nonlinear effects of the source follower for bitline A are small because the voltage magnitude will be close to Vref3. Also depending upon how bitline readout is implemented, BITLINE A may be coupled directly to Vref3 instead of via a source follower for reasons of simplicity. Inclusion of a source follower introduces nonlinear effects, especially in the presence of a large differential signal when source followers T_fA and T_fB would be operating at different gate input voltages. However in practice, considerations of operational amplifier size, operating power and stability may dictate the inclusion of source followers, as shown in FIG. 12.

[0147] Referring to FIG. 12, components in sub-system 230 are optional. However these components allow closed-loop gain of operational amplifier 210 to be varied by adding C240 to increase effective integration capacitance. Integration capacitance can be increased by judiciously enabling transistor switches in the feedback loop via control signals VGA1, VGA2, VGA3 and VGA4. This flexibility can advantageously vary amplifier 210 closed loop gain, and can be used to improve distance measurement resolution δZ, while still avoiding amplifier saturation.

[0148] At the end of an integration period, the total accumulated charge in integration capacitor 220 (perhaps 40 fF) may be read-out in several ways. When the READ signal to the gate of transistor T_rA is high (for the configuration shown), the signal present at the inverting input of operational amplifier 210 will be read-out to BITLINE A. Preferably simultaneously, the READ signal also turns-on transistor T_rB such that BITLINE B reads-out the AMP OUT signal. What results is a differential signal across BITLINE A and BITLINE B that represents the correct voltage value stored on integration capacitor C220.

[0149] An alternate read-out method will now be described. Consider now the signal at the non-inverting input of operational amplifier 210. A high signal C_Vref3 turns-on transistor T_ref3, which couples a known reference voltage V_REF3 to the non-inverting input of operational amplifier 210. As a result, a high READ signal to transistor T_rB .reads-out the signal on BITLINE B. If necessary, BITLINE A may be read-out simultaneously to reduce the effects of noise on V_REF3. The result is achieved by creating a differential value at the input of operational amplifier 210, wherein one of the differential values is V_REF3.

[0150] It will be appreciated that some mismatch may exist between the values of storage capacitors C_A, C_B as well as between node parasitic capacitance, which mismatches can affect the final signal value of the first-described read-out method. It is understood that there will be parasitic capacitance at the non-inverting input of operational amplifier 210. Relative to a capacitor reset value of Vref, the AMP. The charge (with respect to a capacitor reset value of Vref) on this parasitic capacitance is substantially eliminated when the node is coupled to V_ref3. This is what occurs in the second read-out method, but unfortunately a charge error is created whenever the initial amplifier input was not precisely V_ref3. However effects of mismatch using the first read-out method and effects of charge error using the second read-out method can both be reduced by first bringing the voltage levels at both operational amplifier 210 inputs substantially to the initial reset value. The desired result can be accomplished by performing a series of common mode reset and charge dump operations before beginning the read-out sequence.

[0151] Alternately by reading from Bitline A a single ended value (denoted SBA), the error resulting from mismatch between Vref and voltages on the inputs to operational amplifier 210 can be compensated for mathematically. This is because for both read-out methods, there is a mathematical formula between the error and SBA. This mathematical formula is a function of SBA and integration capacitor C220 , and either the capacitance mismatches (for the first read-out method) or the non-inverting operational amplifier positive terminal capacitance (for the second read-out method). Note that for the second read-out method the value SBA must be read-out before V_ref3 is connected.

[0152] A combination of both read-out methods can be used, as follows. First the voltage on the two operational amplifier inputs is brought close to the reset voltage V_ref3. Then SBA is read-out using either read-out method, and the remaining error is mathematically calibrated out. For economy of implementation, it is preferable to acquire S_B A with relatively low accuracy. Thus in practice, SBA is read-out before the voltage on the operational amplifier inputs is brought close to reference voltage V_ref3 via repeated common mode reset dump operations.

[0153] Before this series of common mode reset dump operations, magnitude of SBA relative to the reset voltage V_ref3 will be fairly large, perhaps in the tens or hundreds of mV range. But after the series of common mode reset dump operations, this residual S_B A voltage will be on the order of perhaps a few mV. Furthermore, this residual voltage will be a known fraction of the original SBA voltage before the series of common mode dump operations. Because this fraction is known a priori, by quantizing the larger quantity (magnitude of SBA before the reset operations), the smaller quantity (magnitude of SBA after the reset operations) can be known more accurately. The fraction can be determined empirically, or can be modeled taking into account relative values of C_A, C_B, and parasitic capacitance present at the non-inverting input of operational amplifier 210.

[0154] The addition of transistor switches connected to the DUMP B signal allows the differential detector system shown in FIG. 12 to function symmetrically with respect to “A” and “B” detector components. As a result, at some times the “A” and “B” components of differential detector 70′ will be coupled to the non-inverting input and inverting inputs, respectively, of operational amplifier 210, and at other times the capacitor couplings will be reversed. Within an integration period T, there may be several integration time slices defined. After each sub-integration time slice, one might decide to carry out a dump operation, a common mode reset operation or both. After each integration time slice, the roles of “A” and “B” within the differential detector 70′ may be alternated. This alternation can result from interchanging the clock signals for “A” and for “B”, or changing by 180° the phase of optical energy emitted by 120 (see FIG. 1B), which has the same effect. With such alteration, a differential signals is produced at the output of detector 70′ that, to the first order, is the inverse of the differential signal at the previous sub-integration time slice.

[0155] Note that the DUMP B-related transistor switches couple operational amplifier 210 with the non-inverting and inverting input terminals switched with respect to DUMP A. As a result, the signal that accumulates on integration capacitor C229 accumulates in additive fashion. This feature advantageously substantially reduces many errors associated with offsets and the like. Advantageously, this aspect reduces reliance upon 0°, 180° cancellation in different detection signal captures. This improvement follows because both 0° and 180° phases are used within a common capture (e.g., at a close time interval perhaps on the order of a mS or so) to cancel errors.

[0156] Note too that additional functionality results from the presence of operational amplifier 210. For example, this amplifier may be used for two purposes: to enhance common mode reset as noted above, and for pixel detector analog-to-digital conversation using techniques well known in the art. Other secondary uses of the operational amplifier can include dynamic range enhancement, 0°, 180° cancellation, 0°, 90° capture, and so on.

[0157] Some exemplary parameters for the embodiment of FIG. 12 not stated above will now be given. Capacitor C240 is nominally about half the value of integration capacitor 240, e.g., about 20 fF, where storage capacitors C_A and C_B are each about 60 fF. Exemplary fabrication data for transistors T_fA, T_fB are about 0.5 μ/0.356μ, transistors T_rA, T_rB are about 1.5 μ/0.6μ, transistors T_refA, T_refB, T_resetA, T_resetB are about 0.42 μ/0.6μ, transistors T_shutA, T_shutB are about 0.42 μ/0.6μ, and the four transistors associated with capacitors C220, C240 are each about 2 μ/0.6μ.

[0158] FIG. 13 depicts various oscilloscope amplitude versus time traces for AMP IN, AMP OUT and DUMP A waveforms for a received detector light phase such that the AMP OUT signal increases with time. The integration period T in FIG. 13 is approximately 18 ms. The uppermost waveform is the AMP A or BITLINE B signal, which represents the accumulated differential between charge on capacitor C_A and capacitor C_B during the integration time T. It is seen that the AMP OUT signal approximates a stair-step waveform that increases every time DUMP A is turned on. The resetting of AMP IN and AMP OUT to the reference voltage preceding each reset occurring at events Φ_F, is shown superimposed on the DUMP A reset signals. It is understood that when Φ_F is active, high in this example, active reset signals are also present at Φ_SW, Φ_norm, and so forth. In FIG. 13, waveforms just preceding events Φ_F are read actions, for which the non-inverting operational amplifier input is coupled to Vref3. For ease of illustration, magnitude of Vref3 is intentionally shown as being different than magnitude of the reset voltage.

[0159] To recap before describing the present invention, embodiments of FIGS. 3A-10B provided pixel differential photodetectors with enhanced common mode rejection characteristics, using various common mode reset methods and configurations to electronically reduce the maleffects of ambient light. The embodiments of FIG. 11 and FIG. 12 provided substantial improvement in retention of the desired differential charge in pixel differential photodetectors. But in some applications, large magnitudes of the desired differential detection signal may saturate the pixel differential photodetectors. An additional challenge is how to enhance signal/noise ratio for a TOF system employing pixel differential photodetectors, preferably while using all of the detectable incoming optical energy, so as to preserve system depth resolution. Embodiments of the present invention that addresses these issues using relatively little additional circuitry will now be described with reference to FIG. 14-FIG. 17C. Indeed, embodiments of the present invention may be implemented within the perhaps 50 μm×50 μm area of a pixel.

[0160] FIG. 14 depicts an embodiment of a differential pixel or pixel electronics 80′″ with improved differential dynamic range and signal/noise ratio, according to embodiments of the present invention. Portions of differential pixel 80′″ are similar to what has been earlier described. For example, in addition to implementing the present invention, differential pixel 80′″ also provides common mode reset with restoration of common mode potential at the input of operational amplifier 210. Elements in FIG. 14 bearing element numbers similar to element numbers used elsewhere herein may be understood to refer to elements that are similar to the earlier described elements. Thus detectors 70′ preferably is a differential photodetector, perhaps such as described with reference to FIGS. 3C-12. Elements of pixel electronics 80″ may also be similar to elements within electronics 80′, as described earlier herein. For example components within block 200′ in some respects may be similar to components within block 200 in FIG. 12, although components within pixel electronics 80″ can advantageously serve several functions. Detector 70′ in shown in FIG. 14 as receiving a number of VBIAS and clock signals, but it should be understood that more or fewer such bias and clock signals can instead be used. If desired, additional description regarding differential photodetectors may be found in U.S. Pat. No. 6,906,793 (2005) Methods and Devices for Charge Management for Three-Dimensional Sensing, assigned to Canesta, Inc.

[0161] As noted, one aspect of the present invention further improves differential pixels, especially with respect to avoiding saturating even with relatively large amplitude differential signals, while another aspect of the present invention enhances signal/noise ratio for the detection signal path. In this first aspect, embodiments of the present invention add a fixed compensating offset (ΔV) to the differential signal voltage on capacitor C_DSC whenever magnitude of the differential signal exceeds a predetermined maximum or minimum value. (If desired, the fixed compensating offset signal could of course be ΔQ, where Q is charge.) With reference to FIG. 14, within electronics 80″, circuitry 300 is used to implement the insertion, as required, of the fixed compensating offset (ΔV) into differential signal capacitor C_DSC to avoid differential pixel saturation, even from relatively large amplitude differential signals. As such, offset ΔV is negative if the voltage on C_DSC has become too positive, and the offset ΔV is positive if the voltage on C_DSC has become too negative. In some embodiments, for example as shown in FIG. 15A, the accumulated charge voltage on the differential signal capacitor is checked synchronously, at which time ΔV is added, if needed. A count is kept of the number (N) of ΔV offsets that had to be added, and effective differential signal capacitor voltage is actual output voltage across the capacitor (Vo)+N·ΔV. For ease of illustration, FIG. 15A shows equal time intervals, however synchronous embodiments using unequal time intervals may also be employed. In other embodiments, as exemplified by FIG. 15B, reset of the integration capacitor voltage is asynchronous, and occurs whenever the voltage exceeds a predetermined maximum or minimum threshold. Again a count of the number (N) of resets is kept, and effective differential signal capacitor voltage is Vout+K N·ΔV, where K is a constant. If compensation is implemented using charge, then the effective differential signal capacitor voltage will be proportional to Vout+K N·ΔQ, where the constant K represents a units conversion from charge to voltage. These embodiments preserve the desired differential signal and prevent saturation of the differential pixel even when the differential signal is large in amplitude. Saturation due to common mode signal is prevented, preferably using embodiments of the above-referenced co-pending application Ser. No. 11/110,982.

[0162] Further, within electronics 80′″, a second aspect of the present invention is implemented using a portion of circuitry 200′ to dynamically vary the gain A_G of pixel amplifier 270 to enhance detection signal/noise ratio by using a highest possible gain that still avoids saturation of pixel electronics 80′″. A high amplifier gain (A_G) advantageously reduces effective noise contribution downstream in the signal path by 1/A_G. Gain of each such amplifier is variably controlled to adjust A_G individually for each pixel as a function of its present signal value. Within the array of differential pixels, each amplifier is first operated at maximum A_G, and integration capacitor values are readout and stored in a row buffer 65 (see FIG. 17A). A_G for each amplifier in the row is then incrementally decreased, and the row buffer is updated only for those amplifiers whose associated integration capacitor is not presently saturated. The above process is repeated until the value in the row buffer corresponds to the highest non-saturating gain for each amplifier associated with the row. The row buffer also records the value of the highest non-saturating gain for each amplifier associated with the row. At this juncture row buffer 65 is readout, and the process is repeated for the next row in the array, and so on continuously. In this fashion amplifier values of A_G are individually maximized, commensurate with avoiding overload or saturation of components downstream in the signal path. The desired result is enhanced signal/noise ratio. Alternative embodiments can, of course, increment rather than decrement amplifier gain, and cause the row buffer to latch the non-saturated gain value for each amplifier associated with a row.

[0163] Further details as to the increment/decrement ΔV aspect of the present invention to avoid pixel saturation due to large magnitudes of differential signal will now be given with reference to FIGS. 14, 15A, 15B, and 16. Referring to FIG. 15A, in accommodating for large differential dynamic range, for ease of explanation let the differential charge first be converted to a single ended value and be collected in a differential signal capacitor C_DSC (see FIG. 14, circuit block 230′). It is understood that the relationship between the detected differential photocurrent i, the capacitor C_DSC, and the resultant voltage is given by i=C_DSC δV/δt.

[0164] As shown in FIG. 15A, at periodic intervals, e.g., at t1, t2, t3, . . . , magnitude of the charge voltage developed on capacitor C_DSC is checked synchronously. If at the moment of check the voltage on C_DSC exceeds a threshold, V_high or V_low, then a compensating fixed amount of charge (denoted ΔV) is added to capacitor C_DSC as a compensating offset. Thus if the accumulated voltage on C_DSC becomes too positive, e.g., V>V_high, then an offset of −ΔV is added to capacitor C_DSC, and if V becomes to negative, V<V_low, then an offset of +ΔV is added to capacitor C_DSC.

[0165] For example, in FIG. 15A at time t1, V>V_high and a negative offset ΔV is added to the capacitor voltage. At time t2, magnitude of the capacitor voltage does not exceed V_high or V_low and no offset is added. However at time t3, the voltage is again too high and a negative offset ΔV is again introduced, and so on. The number (N) of resets is counted and at a given time, the effective voltage (V_effective), had no resetting occurred, is equal to V_out+nΔV. For the example of FIG. 15A, there were three resets (n=3), the V_effective=V_out+KNΔV, =V_out+3ΔV, where K is a constant that may equal one. A diagram similar to FIG. 15A can of course be drawn for capacitor C_DSC acquiring a negative charge, in which case a positive offset +ΔV would be added whenever the capacitor voltage goes below V_low. If the effective capacitor saturation voltage is very high, an offset larger than V_high but preferably not larger than (V_high−V_low) may be used to reduce the number of offsets N.

[0166] FIG. 15B depicts an alternative embodiment, again using the example of a capacitor C_DSC acquiring a positive charge, in which the voltage on C_DSC is reset asynchronously, whenever V>V_high. In this example, each reset adds −ΔV to the capacitor voltage, which returns the capacitor voltage to V_low. Again the number N of resets is counted, and the effective capacitor voltage is given by V_effective=V_out+NΔV, or since n=4 in this example, V_effective=V_out+4ΔV. If the effective capacitor saturation voltage is very high, a negative reset offset, preferably not lower than V_low (the low saturation voltage) may be used to reduce the number of resets N. Again a similar diagram may be drawn for the case of a capacitor C_DSC accumulating a negative voltage.

[0167] The choice of implementing synchronous or asynchronous reset depends upon many factors. Generally, an asynchronous reset is more complex as each pixel must constantly monitor its differential signal capacitor voltage, and self generate control signals required to adjust the C_DSC voltage. Further, these operations must be performed accurately in the presence of noise, as the other pixels are integrating and hence the modulation clocks are running. Further, if the reset count is not accumulated inside the pixel, the occurrence of resets may need to be communicated asynchronously, a difficult task when all pixels simultaneously reset.

[0168] On the other hand a synchronous implementation requires more frequent resets as the pixels must be reset well before they saturate. Further, it must be ensured that the pixels have sufficient remaining margin such that they do not saturate before the next C_DSC voltage check, which may not occur for a while. Also in synchronous implementations, each ΔV reset adjustment must be smaller as the C_DSC voltage may be relatively far from saturation.

[0169] FIG. 16 depicts exemplary pseudocode used to an embodiment of the first aspect of the present invention, the ΔV potential to compensate for large differential signal magnitudes. More specifically, the pseudocode of FIG. 16 provides detail as to implementing the synchronous reset embodiment of FIG. 15A, using circuitry shown in FIG. 14. Separate pseudocode to implement an asynchronous embodiment such as shown in FIG. 15B is not given in that implementation steps will be self-evident to those skilled in the relevant art in view of the description given with respect to FIG. 16. The exemplary algorithm of FIG. 16 may be stored as software 350 in memory 100 of a TOF system 400, for execution by microprocessor 90 (see FIG. 17A). At the circuit level, execution of this algorithm preferably implements the following method steps.

[0170] (1) relevant circuitry in FIG. 14 is reset by simultaneously turning on resetF (sometimes denoted Φ_F), resetA, resetB, refcon_int, shutterA, shutterB, dumpA, dumpB;

[0171] (2) resetF, resetA, resetB (and perhaps DumpA dumpB) are opened, modulation clocks CLKA, CLKB, CLKC are turned on, and integration begins. For the configuration shown, refcon_int and reset) must be high during integration. The high state can be forced by setting offset_control=0, HILO=1, which forces Y=0, whereby if refcon_bar=0 (e.g., complement of refcon, denoted REFCON in FIG. 14), refcon_int=1;

[0172] (3) modulation clocks are turned off;

[0173] (4) dump switches DumpA are closed and the differential charge is moved to capacitor C_DSC, whose value is on the order of perhaps 40 fF. Dump switches are then reopened;

[0174] (5) shutter switches ShutterA ShutterB are open circuit, and then ResetAB switches are closed to restore common mode at the input of operational amplifier 270, as described earlier herein;

[0175] (6) a test is performed on voltage V_x across differential signal capacitor C_DSC to see whether the voltage is becoming too large, perhaps >500 mv. Assume the logic trip point of the threshold test gate 320 is 2V. V_x test is achieved by turning on VGA1 (and/or VGA2), turning off VGA3 (and/or VGA4), and setting V_ref3 to 2.5 V. (This statement ignores, for the sake of simplicity, effects of capacitance at the (AMP_IN) inverting input of amplifier 270.) If the input of the threshold test Gate T is then V_test<500 mv then V_x>500 mv. During this test offset_control=1. Thus the output of gate 320 is high if and only if V_x>500 mv.

[0176] (7) if V_x>500 mv, a fixed charge ΔV is dumped into capacitor C_DSC to reduce V_x and avoid the saturation that would occur if V_x increased further. The decision is available at node Y: HILO=1 if the test was for V_x high, and HILO=0 if the test was for V_x being low. In this exemplary case, HILO=1 and Y=0 if and only if V_x>500 mv.

[0177] (8) during this time, reference levels (V_ref1−V_ref2) are set such that [C]·ΔV is approximately [C_DSC in parallel with C′_DSC]·(V_ref1−V_ref2), where C′_DSC is about 20 fF and, as noted, C_DSC is about 40 fF. The resultant charge [60 fF]·(V_ref1−V_ref2) is potentially dumped into integration capacitor C When recon_bar=0, recon_int=1, if and only if V_x>500 mv. If recon_int is high, then a differential charge 40FF*(V_ref−V_ref2) is created between the two capacitors. In this case an exemplary value for (V_ref−V_ref2) is perhaps −900 mv. If desired, a single capacitor C_DSC may be used in this embodiment, or if more capacitors are present, a single capacitor may be used, if desired.

[0178] (9) steps similar to (6), (7), (8) are repeated for testing if the voltage across capacitor X is too low (e.g. V_x<−500 mv). Note that during this operation an exemplary value for V_ref1−V_ref2 is perhaps +900 mv

[0179] (10) At this juncture the differential voltage in the capacitors is preferably 0 if no offset occurred, −900 mv if a negative offset occurred, and +900 mv if a positive offset occurred. At a time, e.g., when refcon_bar=1, this corresponding differential charge may be dumped onto differential signal capacitor C_DSC by turning off refcon_bar and turning on dump switches dumpA and dumpB;

[0180] (11) A count of the number of times a charge of either polarity of 60FF·(V_ref1−V_ref2) is dumped onto this capacitor is maintained. The total amount of differential charge generated by the detectors is then N·Q+V_x where Q_x is the final charge in the capacitor and N is the number of times the offset dump occurred.

[0181] Various methods for maintaining this offset count N may be implemented. The count N may be stored within the pixel to be subsequently read-out, or can be read-out from each pixel following each offset test operation. If the positive or negative sign of the offset ΔV (or if charge is used, ΔQ) is known, then only the magnitude of the count need be stored. Alternately positive numbers can be used for positive offsets, and negative numbers can be used for negative offsets.

[0182] A simple solution is of course to implement an up/down analog or digital counter within the pixel, e.g., electronics 80′″. The count should increment if the hi test is successfully performed (i.e. recon_int=1) and the count should decrement if the low test is successfully performed (i.e. recon_int=1).

[0183] A digital counter may be implemented in many ways, including without limitation providing a pseudo random sequence type counter. An analog counter may be also be implemented in many ways. For example, a short pulsed current source could be used to charge or discharge a capacitor with each change in N, where final capacitor voltage is indicative of how many charging/discharging pulses have been received. Alternately an analog counter may be implemented by dumping charge from a second capacitor onto a first capacitor, where dumped charge is positive or negative, depending upon whether the count was up or down. In such implementation, the final charge on the first capacitor would be indicative of the number of times the dump occurred, and whether the dump was up or down. Note that depending on the details of a particular counter implementation, a linear correspondence between the count N and the final analog voltage need not exist.

[0184] Rather than count offsets within the pixel, one may instead count outside the pixel, for example, by inferring the reset count N. In such an implementation, the preferably differential pixel value is read-out at the end of the first integration period, or after the first few integration periods. The resultant low resolution value provides an estimate of the final value of the pixel. In essence it provides an interval of possible values [V_final^low, V_final^high]. For example, if there are M integration intervals and the pixel reading after the first integration period is V₁ with a margin of error ε, then the interval of possible final values could be [V_final^low, V_final^high]=[M*(V₁−ε), M*(V₁+ε)]. The final value for the pixel is V_effective=N·[V_offset+V_out]. N must be selected such that V_effective∈[V_final^low, V_final^high]. If (V_final^high−V_final^low)<V_offset then N is unique and can be readily determined.

[0185] Alternately the count N (number of times each pixel was offset) can be maintained outside the pixel array 60″(see FIG. 17A). At the end of each hi or low offset test, each pixel can output a 1 if the test was positive or 0 otherwise. The pixel array is quickly scanned and for each pixel whose output was 1 the corresponding count is incremented or decremented depending upon whether the test was for a high value or low value. In a preferred embodiment, at the end of both hi and low tests, a single value is produced that preferably can take on one of three meanings up-count, down-count, and no-count. Such embodiment is advantageous in that these values can be read from the array during the next integration period, e.g., a time that lasts for at least several 100 us. By contrast the high and low test may occur a few microseconds apart and thus, but for this aspect of the present invention, it can be difficult to read the entire array in such a short period.

[0186] Referring to block 310 at the bottom of FIG. 14, a compact tri-state circuit to accomplish the above is shown. This simple two transistor circuit outputs one of three values to indicate to the outside world whether the associated pixel detection signal required +Δ offset, −Δ offset, or no compensating offset at all. In a preferred embodiment, the output is first pre-charged at a value V_ref between ground and V_DD−V_t. If a low (−Δ) offset occurs then Refcon_int=1 with Offset_control=1 and HiLO=0 will be sampled onto the output out. Similarly if a high offset (+Δ) occurs then Refcon_int=1 with Offset_control=1 and HiLO=1 will be sampled as V_dd−=V_t onto the output out. Thus, the output of this simple tri-state device is V_ref if no offset occurred, is 0 if −Δ occurred, and is V_DD−V_t if +ΔV (high offset) occurred. Given that there are many hundreds of pixels within a pixel array, e.g., array 60″ in system 400 in FIG. 17A, it is required that relatively little area of IC 30′ be required to implement each tri-state device 310. The dynamic pass-through use of transistors to implement tri-state device 310 is preferred to an implementation using standard logic. A standard logic implementation would simply require too much area of IC 30′ (see FIG. 17A) to be practical, given that many hundreds of such tri-state devices must be implemented.

[0187] For ease of illustration and explanation, the foregoing embodiments were described primarily in terms of a constant fixed Δ compensation offset. In practice, however, there is no requirement that magnitude of a positive Δ compensation (+Δ_COMP) offset be identical to magnitude of a negative Δ compensation offset (−Δ_COMP). Thus if compensation is with voltage, if positive Δ compensation offset is ΔV₁, and negative Δ compensation offset is ΔV₂, it is understood that ΔV¹ need not be equal in absolute magnitude to absolute magnitude of ΔV₂. Similarly if compensation is with charge, if positive Δ compensation offset (+Δ_COMP) is ΔQ₁, and negative Δ compensation offset (−Δ_COMP) is ΔQ₂, it is understood that ΔQ₁ need not be equal in absolute magnitude to absolute magnitude of ΔQ₂. In these embodiments, it still suffices to count number N of resets.

[0188] In the various embodiments hitherto described, it has been assumed that in a sequence of compensation offsets that magnitude of each offset would be substantially equal. In such embodiments, the number N of resets could simply be counted for a sequence since each offset was substantially identical. However one could implement an embodiment in which magnitude of each offset in a sequence of compensation offsets was not necessarily equal. In such unequal compensation offset embodiments, one would have to record the sum total (i.e., magnitude and sign) of all compensating offsets.

[0189] As noted earlier, a second aspect of the present invention provides enhancement of signal/noise in the pixel detection path, by maximizing gain A_G of each amplifier 270 (see FIG. 14). During readout from a signal path, amplification (or gain) can reduce downstream noise by a factor of 1/A_G. Thus, embodiments of the present invention implement amplifier 270 as a variable gain amplifier (VGA).

[0190] Looking at TOF system 400 in FIG. 17A, the value of a pixel detector 70′ is read though an analog and digital signal path. To reduce effective noise (including analog-to-digital quantization noise) it is beneficial to keep the gain as high as possible as early as possible in the signal path. If desired, multiple variable gain amplifiers could be provided in the signal path, each amplifier maximizing the signal strength on segments of the path downstream. Understandably additional information corresponding to the A_G gain settings for each such VGA amplifier needs to be known to decode the output signal value of the signal path.

[0191] However, setting the VGA gain too high can cause saturation in the signal path downstream of the amplifier, and setting the VGA gain too low exposes the system to more effective noise downstream of the amplifier. VGA. The optimal setting for VGAs in a signal path depends on the present signal value and also the noise characteristics of the path. Hence, setting VGA gains beforehand (a priori) may result in a sub-optimal setting. However, if the VGA gain settings are set after the signal is known (a postiori) a more optimal setting can be determined and used.

[0192] In a preferred embodiment, a VGA such as amplifier 270 in FIG. 14 is incorporated into each pixel 80′″ in sensor array 60″. The VGA gain A_G is set individually for each pixel based on the present differential signal value at that pixel. According to the present invention, the VGA preferably is shared with other functionality in the pixel, for example, implementing improved CMR and/or high active dynamic range, as described earlier herein. A preferred implementation is shown in FIG. 14, in which there are two VGA feedback capacitors, C_DSC and C′_DSC, which capacitors advantageously also enhance detection ability despite the presence of large magnitude differential signals, as described earlier herein. The use of these two VGA capacitors enables three possible feedback capacitance values (discounting a null capacitor case).

[0193] In normal operation of system 400 (see FIG. 17A, part of which system is shown in FIG. 14), i.e., during integration, CMR, etc. the two VGA capacitors are coupled in parallel with an effective value of 60 fF, where an exemplary value for C_DSC is 40 fF, and an exemplary value for C′_DSC is 20 fF. Signal charge is stored in both capacitors, as they are parallel-coupled, but during readout, the gain of VGA 270 may be changed.

[0194] The feedback around amplifier 270 may be tuned. For example by using the following transistor gate signals VGA3=0, VGA1=1, VGA2=0, VGA4=1, charge from C′DSC Is dumped into the larger value capacitor C_DSC, thereby multiplying the VGA gain for amplifier 270 by (40+20)/40=3/2, where 40 and 20 refer respectively to the fF size of C_DSC, D′_DSC. Similarly, an alternate tuning results using the following transistor gate signals VGA3=1, VGA1=0, VGA2=1, VGA4=0, which result in charge from C_DSC (40 fF) being dumped into C′_DSC (20 fF). This in turn multiplies the VGA gain of amplifier 270 by (40+20)/20=3.

[0195] Thus during readout, the amplifier gain multiplier can be set to 1×, 1.5×, or 3× for the exemplary configuration of FIG. 14. Note that changing the A_G gain setting does not affect the total signal charge in the system, even if the output saturates. Thus even if the system saturates, setting A_G to 3× does not affect the pixel reading if gain A_G is later set to 1×. Note that other gain settings, including additional gain settings choices can be implemented by providing more than two capacitors for the amplifier feedback loop.

[0196] According to the present invention, during the readout phase each row of pixels in array 60″is read out sequentially. First the row is copied into row buffer 65, preferably provided at the bottom of array 60″. Next, each column of row buffer 65 is sent to an analog-to-digital converter that may be regarded as part of I/O circuitry 120 (see FIG. 17A).

[0197] Row buffer 65 preferably is a one dimensional array of row buffer elements, having an element for each row in array 60″. To enable a-postiori VGA gain according to this aspect of the present invention, each row buffer element is augmented so as to detect saturation and reject rather than store an incoming pixel value if saturated. First each VGA amplifier 270 in the array row currently being read is set to the highest gain value, 3× in the above-described example. Those row buffer elements that receive a non saturated signal from their corresponding pixels will latch the pixel signal values and record the gain setting for which the value has been latched (3×, in this example). Next the VGA for the same row is set to a reduced gain value of 1.5× (according to the above example). Those row buffer elements that have not already latched a value and that now receive a non-saturated signal from their corresponding pixels will now latch the pixel signal values and record the gain setting at which the value was latched at (2×, in this example). Finally the VGA amplifiers for the same row are set to a gain value of 1× (for the present example). Those row buffer elements that have not already latched a value, latch the pixel signal values and record the gain setting at which the value has been latched (1× in this example).

[0198] At the end of this process, each element in the row buffer array has latched a value and has recorded the gain setting at which the value was latched. Also, each element in the row buffer has latched the highest gain setting achievable without saturation (except for those elements saturated at gain 1×). By sending the latched signal value and the gain setting at which the signal value was latched, a correct value of the pixel signal can be determined with high accuracy and low noise.

[0199] Alternative embodiments can, of course, increment rather than decrement amplifier gain A_G or even randomly or otherwise scan or vary A_G and cause the row buffer to latch the non-saturated gain value for each amplifier associated with pixels in a row.

[0200] The first and second aspects of the present invention preferably are combinable into an overall system. For example, FIG. 17A depicts a time-of-flight system 400 that preferably emits modulated optical energy. Some of this emitted modulated optical energy is reflected by a target object 20′ and is detected by differential pixel photodiodes 70′ and their associated electronics 80′″, in an array 60″, typically fabricated on a CMOS IC 30′. As noted herein, photodiodes 70′ may also be sensitive to common mode optical energy, e.g., from ambient light source 310. Among other functions, system 400 can determine range distance z to the target object. System 400 preferably includes the present invention within electronics 80′″, and thus exhibits enhanced ability to cope with large magnitudes of differential detection signals, and further exhibits enhanced signal/noise ratios in the pixel detection path.

[0201] In the configuration of FIG. 17A, target object 20′ is a so-called virtual input device, here the optically projected image of a computer-type keyboard. As a user “types” on the “keys” of the virtual keyboard, system 400 can determine which virtual key was “contacted” and when time of contact occurred. Electronics within system 400 can then output relevant scan code to a companion device 500, perhaps a PDA, a cell telephone, a kiosk, a computer, etc. Advantageously, system 400 can function well, even in the presence of strong ambient light. Further details as to such implementations of TOF systems may be found in U.S. Pat. No. 6,710,770 (2004), U.S. Pat. No. 6,690,354 (2004), and U.S. Pat. No. 6,614,422 (2003), assigned to Canesta, Inc. of Sunnyvale, Calif.

[0202] FIGS. 17B and 17C depicts yet another application of system 400, such as shown in FIG. 17A. For ease of illustration, system 400 is shown deployed looking rearward on the rear portion of a motor vehicle 455. System 400 acquires z depth data, among other information, and can display information 470 on a monitor 460 within the vehicle, perhaps a heads-up-display. Output from system 400 can provide the operator of vehicle 455 with visual (and acoustic, if desired) warning as to dangerously close proximity of objects 20 within the vehicle's path. As such, system 400 can augment the vehicle operator's ability to drive safely, despite blind zones. System 400 could instead be deployed within vehicle 455, perhaps aimed towards the front passenger seat. As such system 400 could determine the size and disposition of a target object in the front passenger seat and communicate such information to another system within the vehicle. Such other system might be the control system for deployment of the emergency air bag. Thus if system 400 determines that a child is the front seat passenger, such information might be used to override deployment of the air bag, or perhaps cause deployment at less than full force, so as to reduce air bag injury to the child. Obviously many applications can be found for system 400, especially in environments where the present invention can help detect useful depth data, despite less than ideal environmental conditions.

[0203] Various embodiments of the present invention have been described in the context of enhancing performance of differential pixel detectors, commonly used in time-of-flight systems. However it is to be understood that embodiments of the present invention may be used in other applications that may involve differential detection signals that may have high amplitude, and whose signal path may require enhanced signal/noise characteristics. Further, those skilled in the art will appreciate that various described embodiments or portions thereof may be implemented using components other than the specific semiconductor switches, amplifiers, comparators, integrators, counters, etc. described herein.

[0204] Modifications and variations may be made to the disclosed embodiments without departing from the subject and spirit of the present invention as defined by the following claims.