FR2551285A1 - SHOOTING APPARATUS COMPRISING CHARGE COUPLING COMPONENT IMAGE ANALYZER - Google Patents

Abstract

THE INVENTION RELATES TO A PHOTOGRAPHIC APPARATUS INCLUDING A CHARGE COUPLING COMPONENT IMAGE ANALYZER. THE APPARATUS ESSENTIALLY INCLUDES AN ELECTROMETER 14-20 OUTPUT STAGE WITH A FLOATING ELEMENT14, A DIFFERENTIATOR CIRCUIT30 OF THE IMAGE RESPONSE SAMPLES IN RELATION TO TIME AND A SAMPLING AND HOLDING CIRCUIT40 WHICH RECOGNIZES THE SAMPLES AND THAT PRODUCES AN IMAGE RESPONSE WITHOUT SPACES. THE INVENTION APPLIES IN PARTICULAR TO A SEMICONDUCTOR TELEVISION CAMERA.

Description

The present invention relates to the reduction of noise accompanying the

  image response of a coupling component camera

  load, while avoiding double sampling. 5 In this text, the description of a component

  load transfer uses the following convention.

  A surface of a semiconductor substrate, on which gate electrodes of the component are disposed, is described as the "upper" surface of the component, even though the gate electrodes are actually oriented differently from the surface of the substrate. In addition, words such as "below" and "above" are used in the text according to

of this convention.

  In general, a floating diffusion output stage of a charge coupled component comprises a semiconductor metalisolent field effect transistor (hereinafter referred to as MISFET) comprising a gate electrode connected to the floating diffusion transistor. in common drain (or common source) configuration as an electrometer, in order to measure the potential on the floating diffusion This potential indicates the charge in a potential sink "below" the floating diffusion The potential measurements are made during respective signal-sampling intervals

  which are separated by rest intervals.

  During such a quiescent interval, the floating diffusion level is set by the MISFET effect to a reference potential at a quiescent drain. More particularly, the floating scattering is a virtual source in this MISFET and the MISFET effect occurs in response to the potential applied to a quiescent gate electrode 35 which is disposed between the floating diffusion and the quenching drain. It is common practice to interpose an additional gate electrode between the floating diffusion and the quiescent electrode and apply a direct potential to that extra gate electrode in order to avoid

  that potential responses to the quiescent pulses are induced in the floating diffusion.

  The quiescent operation of periodically setting the level of the floating diffusion at the potential of the quiescent drain is undesirably accompanied by a type of noise called "quiescent" noise. is potential variations left on the floating diffusion from one quiescent interval to another quiescent noise is a problem in charge transfer components including floating gate output stages, as well as Floating noise is the predominant noise in the super-video frequencies of the output signals of charge-coupled components such as charge-coupled component image analyzers. Noise resting is

  typically 8 d B above the noise in the MISFET electrometer stage following the floating diffusion.

  At lower video frequencies, the flicker (or 1 / f) noise predominates. The flicker noise occurs in the MISFET electrometer stage.

  It is known to provide the output signals of the image analyzer to a sampling and holding circuit. Such a circuit samples during the sampling intervals of the output signal of the image analyzer and then retains the samples. During Sampling Intervals The response of the sample and hold circuit has increased baseband (or first harmonic spectrum) content and reduced content in the higher harmonic spectrum compared to the analyzer output signal. It receives the coefficients of use of the image analyzer output signal and the mixed noise during the following amplification operations are made similar by the sampling and maintenance By

  Therefore, the signal-to-noise ratio is not significantly degraded during amplification. This procedure is hereinafter referred to as "simple sampling".

  The desire to reduce flicker and quiescent noises has led to the use of correlated double sampling. In this procedure, in each clock cycle of the charge coupled image analyzer output register, the signal on the floating diffusion is sampled a first time, when the load which depends on the quiescent noise but not the signal, is present in the induced potential well "below" the floating diffusion This signal is then sampled a second time when the load depending on both the quiescent noise and the signal is present on the floating scattering. Each pair of samples is then differentially combined to produce samples that depend essentially only on the signal. at rest

being deleted.

  Double correlation sampling becomes less feasible when the sampling rate of the load transfer component output stage increases. Pulse times become shorter and the pulse gap is decreased to the limit imposed by the Time Required for Load Balancing Under the Floating or Floating Gate Output When the clock frequency of the output register rises above a few megahertz, the double correlation sampling technique becomes difficult to use. It has been observed that although with double correlation sampling a noise reduction of 20 d B can be obtained at a clock rate of 100 kHz in an image analyzer system at a clock frequency of 1 M Hz, a reduction of more than

  3 to 6 d B is obtained only with difficulty.

  U.S. Patent No. 4,330.

  753 discloses a method for obtaining information signals which it characterizes as relatively free of noise, at the output stage of a charge transfer device. According to this method, the output signal of the electromete stage with regular sampling passes through a bandpass filter to separate the sidebands from a double sideband amplitude modulation, on each side of a harmonic of the clock frequency of the electrometer stage.

  Separate side effects are then detected in synchronism using a switching demodulator operating on the harmonic of this clock frequency.

  The baseband spectrum of synchronously detected amplitude-modulated sidebands is separated from the associated harmonic spectrum. The baseband spectrum is used as the output signal of the charge-coupled device rather than the spectrum spectrum. base of the output signal of the image analyzer that is suppressed by the passband filtering

before synchronous detection.

  This method is effective for suppressing flicker noise at the electrometer stage, which noise is mainly in the baseband. The quiescence noise is neglected in this patent, but, as noted above, quiescent noise is a primary source of noise in a solid-state image analyzer with a quench stage. floating-gate or floating-diffusion output The quiescent noise has a wide band and extends over the harmonic spectrum of the video samples supplied at the output of the solid-state image analyzer. Standby noise makes a major contribution to noise even though a synchronous detection of the sidebands surrounding the clock frequency harmonic is used to recover the image signal from the analyzer's output samples. heard in the course of

  this description that the noise of resting does not concern the simple crossing of the pulses of resting whose reduction relates to the aforementioned patent.

  It has been found that if the output signal of the charge-coupled image analyzer undergoes high-pass filtering or differentiation before sampling and sustaining, the sample and hold circuit converts the spectrum components of The harmonic of the analyzer output signal into baseband spectrum components which are in the frequency domain of a synchronous detection. Therefore, the low frequency content of the image is recovered despite the filtering.

  pass-high of the output signal of the analyzer.

  In addition, the hold operation responds to the peaks of the harmonic spectral content of the output signal of the image analyzer and not to the average absolute value. Therefore, the conversion efficiency of the presently presented procedure is significantly higher than in the synchronous detection operation described in the aforementioned patent This process uses a directly followed sampling without maintaining a low-pass filtering, for a regular detection response. WF Kosonocky and JE Carnes of RCA Corporation's David Sarnoff Research Center describe the floating diffusion amplifier in their article 35 entitled "Basic Concepts of Charge-Coupled Devices" published in September 1975 in RCA Review, Volume 36, pp. 566-593. The article suggests the floating diffusion quiescent at the potential barrier established by a grid that is biased with a direct potential and interposed between floating diffusion and

  a grid functioning as a quiescent grid.

  In other words, the floating diffusion is quiesced at a channel potential in the charge transfer channel in which the floating diffusion is arranged rather than at the grain potential at the end of the charge transfer channel. restoring the floating diffusion has for the most part been eliminated as impracticable by those skilled in the art because this solution introduces pronounced low frequency distortion into the modulation transfer function. Tasks on the back edges of light regions in dark regions appear in television images based on video samples from charge-coupled image analyzers having floating diffusion output stages resumed at rest at channel voltages rather than at drain voltages when output samples from the output stages are processed conventionally and when n sample and hold output circuit is used

  to remove the clock passage.

  Low-frequency distortion in the Kosonocky and Carnes quiescent procedure has been shown to harm only the baseband portion of the output coupled frequency analyzer frequency spectrum of the charge-coupled image analyzer. If the base band (or at least the lower frequencies in which distortion occurs) is removed before sampling and holding to retrieve a space-free signal, the distortion due to the Kosonocky procedure

  and Carnes does not appear in the output response of the sampling and maintenance.

  It has also been found that the quiescent noise can also be suppressed if the quiescent pulses are suitably timed even if the floating scattering is brought to rest at the quiescent drain potential after each quench sample. image element has been clocked to be "below" the scattering for the electrometer measurement In addition, this quiescent can be done without requiring more than one sampling during each scanning interval

of a picture element.

  It should be noted that after quiescence, there is a gap that precedes the introduction of the next charge packet under the floating element of the electrometeque load coupled output stage. rest in this previous interval is the same as the level of noise put to rest during the next interval after the introduction of a charge and before the next resting This phenomenon is exploited in the double

correlation sampling.

  Examining this phenomenon in a somewhat different way, it should be noted that the transition of the signal which is caused by the transition of a charge packet under the floating element can be superimposed on a resting noise level and This level of noise changes only from one picture element to another. It should also be noted that the time difference of the output signal of the charge-coupled image analyzer, according to FIG. The invention eliminates the response to dwell noise levels with respect to the responses of transitions that are caused by the introduction of charge packets under the floating element. Synchronous detection of the differential output signal can then be performed over periods of time. which extend over respective fixed parts of the decreasing responses of these transitions The resulting output signal

  has a good signal-to-noise ratio.

  The invention therefore relates to a reduced noise recording apparatus comprising a charge-coupled component image analyzer or the like, with a floating-element (eg floating or floating-gate) output electrometer stage. ) rested repeatedly, followed by

  a differentiating circuit and then a sampling and holding circuit.

  Other features and advantages of the invention will be better understood when reading

  the following description of an exemplary embodiment

  and with reference to the accompanying drawings, in which: FIG. 1 is a diagram of the signal recovery device according to the invention, used with a field-transfer type charge-coupled image analyzer, FIG. potential profile describing the quiescent floating quiescent at the quiescent drain potential, FIGS. 3, 4 and 5 are potential profiles describing examples of different ways in which the quiescence of a broadcast Floating at a channel potential rather than a quiescent drain potential can be done as an element of the invention, and FIG. 6 is a timing chart describing the preferred quiescent timeout delay.

  the floating diffusion at the draining drain potential according to FIG. 2.

  FIG. 1 thus represents the signal recovery device according to the invention, used

  with a solid state image analyzer.

  For example, the image analyzer is a charge-coupled analyzer 10 of the field-transfer type The charge-coupled analyzer 10 has an image register 11 (or A), a register of memory

25512 Z 85

  12 or (B) and a parallel input and serial output (13) register. The samples of the output signal are generated by charge packets transferred to the right end of the C 13 register using As an example, in this conversion stage, charge packets are regularly clockwise displaced from the output of register C 13 to 10. a potential well disposed under the floating diffusion 14 The charge value in each packet in

  the potential well is then determined by an electrometer. The electrometer comprises a cascaded connection of source-loaded metal-insulator-semiconductor field-effect transistors 15 and 16.

  Another MISFET 17 is connected to the MISFET 15 as a constant current generator source load. An off-chip resistor 28 constitutes a source load for the MISFET 16.

  load 10 appear across the resistor 28.

  A direct potential, OD, is applied to the drains of the MISFE Ts 15 and 16 to condition them in source-load operation. The source of the MISFET 16 is connected to the output signal terminal 27 of the image analyzer. charge coupling 10 and, by the load resistor 28 to a ground connection. Samples of the output signal appearing at the output terminal 27 of a charge coupled image analyzer 10 are applied to the input

  a voltage amplifier 29 low noise.

  The output of an amplifier 29 provides samples of the input signal to a differentiator 30.

  The amplifier 29 separates the input connection of the

  differentiator 30 of MISFET 16 at source load.

  The amplifier 29 preferably has a bandwidth sufficiently small to cut some of the higher video frequencies. With this cutoff, the differentiator 30: (a) responds to the transitions of the signal having a reduced slope by producing pulses with a larger and (b) does not introduce into the final video response excessive amplitude peaks that would otherwise be present as a result of the passage of the clock signal. The differentiated video response with respect to the time of the differentiator 30 is applied as an input signal to a low noise amplifier 35

and broadband.

  When each charge packet has been measured, the potential of the floating diffusion 14 is brought back to rest by the pulse. The pulse (r is applied to the quiescent gate 18 and is usually narrower than the clock pulse applied to the last clock gate (not shown) 20 of the register C 13 The pulse r appears in the temaps where the clock pulse appears. The quiescent gate 18 is arranged "above of a charge transfer channel 19 which extends along the C 13 register and beyond to include the floating diffusion 14 and the quenching drain 20. More particularly, the queuing gate 18 is disposed "above" the charge transfer channel 19 between the floating diffusion 14 and the quenching drain 20 and is preceded by a gate D-C 21 A direct potential RG is applied to the grid 21 The grid 21 is preferably a short grid

  to reduce the amount of charge below it.

  The gate 21 is used to prevent the pulses r applied to the quiescent gate 18 from being electrostatically coupled with the scattering.

floating 14.

  A clock generator 25 is shown in FIG. 1 comprising respective groups of three-phase clock signals for a register A 11, a register B 12 and a register C 13 as is common in an analyzer. Field-transfer type charge-coupled images Other well-known clock systems using a 2, 4 or a virtual phase clock control could be used. The clock generator 25 generates the pulses. described above for the application 10 to the gate electrode 18 of the output stage to

floating broadcast.

  Clock generator 25 also produces OS pulses at a repetition rate equal to the clock frequency of register C 13 during serial line reading. These binary pulses are applied by line 26 to a synchronous detector. Bones are used as a carrier to control the times when the signal applied to the synchronous detector 40 from the output of the low noise, wideband amplifier 35

  is sampled in the synchronous detection process.

  The amplifier 35 provides a voltage gain which increases the signal level so that the associated noise is greater than that introduced by the synchronous detection which follows; the noise attributed to the amplifier 30 is negligible, compared to the noise 1 / f produced in the image analyzer 10 to

load coupling.

  The differentiator 30 is shown in FIG. 1 as a simple high-pass filter RCZ comprising a capacitor 31 in series and a shunt resistor 32. The time constant RC 1 'is chosen to suppress at least as much of the band frequency spectrum. the output samples 35 of the image analyzer 10 that accompanied by the flickering noise or (1 / f) significantly larger compared to the background thermal noise This suppression of the frequency spectrum in baseband is manifested in the output response of the differentiator supplied to the synchronous detector 40 The time constant t is the inverse of a frequency of angle fc expressed in radians per second, the amplitude components at this frequency fc being suppressed by 3 d B by the RC filter The frequency fc can be considered as a limit frequency between a substantial suppression and a substantial non-suppression of the frequencies present in the input signal of the differentiator but present only selectively in its

output signal.

  A 430 picofarad capacitor 31 and a 75 ohm resistor 32 were used in a signal recovery system with a 7.5 MHz C 13 register clock frequency. The high pass filter RC had a time constant of 35. nanoseconds, giving an angle frequency of 5 MHz. The upper frequencies of the baseband spectrum of the output samples of the image analyzer 10 from the differentiator 30 are comkinated with the first harmonic spectrum demodulated in the signal. output of the synchronous detector 40 to obtain the high-frequency video peaks The baseband signal residues and the signal of the first demodulated harmonic spectrum are correlated and summed algebraically while the noise components of the respective bands are not therefore, the advantages over the signal-to-noise ratio are increased with

  this form of high frequency video peak.

  A switching demodulator followed by a low pass filter could be used instead of the synchronous detector 40 But these switching demodulators perform average detection, producing a demodulator output signal in which the recovered baseband is accompanied by a strong spectrum. Harmonic It is preferable to use a synchronous detection which is a peak nature detection to reduce the intensity of the harmonic spectral residues from the detection compared to the recovered baseband spectra. A sample and hold circuit works here as

such a synchronous detector.

  FIG. 1 shows a sampling and holding circuit 40 comprising a MISFET 41 and a capacitor 42. The MISFET channel 41, when conducting, transmits a respective sample to capacitor 42 which holds this sample. The gate of MISFET 41 receives line 26 the pulses Os which are supplied at a frequency equal to the clock frequency of the register C The channel of the MISFET 41 is made conductive by each pulse XS The MISFET 41 functions as a transmission gate of a type in which the Control signals (pulses s of line 26) are not applied to a measurement applicable to the conductive channel. The output circuit of this form of synchronous detector is not balanced with respect to the input signals supplied to the conducting channel. baseband spectrum applied to the input of this synchronous detector 25 appear at its output, allowing the passage of the higher frequency components e of the baseband spectrum This produces high peaks

video signal frequency.

  The output signal detected from the sampling and holding circuit 40 (unlike a signal taken directly from a switching demodulator) is a usable video signal and need not be filtered further than by the video amplifier cutoff. The detected output signal from the detector 40 is applied to a buffer amplifier 50 and thence to a regulation filter 51. The filter 51 is preferably a low-pass filter which removes clock frequency residues so as to provide a low noise video signal which is also free of image detail distortions This low noise video signal is usually applied to a video processing amplifier (not shown) in which timing and correction pulses are inputted to video processing amplifiers (not shown). moments in coordination with timing

of the clock generator 25.

  The phases of the pulses provided by the line 26 from the clock generator 25 to the gate of the MISFET 41 are arranged so that the channel of this transistor is conductive in the decreasing portions of the peak responses of the differentiator 30 to certain level transitions. These transitions can be attributed to the introduction of charge packets under the floating diffusion. These transitions are not accompanied by resting noise. But there is a noise of setting. residual rest of the decreasing portions of the differential responses of the differentiator 30 to the transitions of the output signal of the image analyzer 10 to the front and back flanks of the quiescent pulses Or '25 Therefore, these last transitions contain components These noise components result respectively from the previous picture element and the picture element bear as will be explained in more detail later. The synchronous detection of the decreasing portions of the peak responses at the trailing edges of the signal 4 r can be reduced by providing a shorter time constant t in the differentiator 30 but this arrangement reduces the importance of the horizontal peaks obtainable by the passage

  residual baseband spectrum.

  It may be easier to avoid synchronous detection of the transitions at the leading and trailing edges of the gold pulses with special attention to the timing with which the gold quiescent pulses are applied to the quiescent gate electrode. 18 of the output stage of the floating diffusion image analyzer The complexity of this quiescent procedure will be

  explained later using the potential profile 10 of Figure 2 and the timing diagram of Figure 6.

  Each of the potential profiles of Figures 2, 3, 4 and 5 shows in its upper part stylized representations of the structures encountered.

  by charging packets moving from left to right in the charge transfer channel.

  On these profiles, the more positive potentials are represented downwards. The diagram assumes an empty potential well under the floating diffusion. For reasons of simplicity, the effects of leakage field are neglected. is not essential for

the understanding of the invention. The potential profile of Figure 2 describes the quiescence of the

  Floating diffusion 14 at 25 quiescent drain potential RD which is applied to the quiescent drain 1 oc is the phase of the register C clock applied during the line read at the last lattice 61 controlled by the register clock C 13 The latter comprises a final gate 62 according to the clock grid 61 and to which is applied continuous BP The potential BP establishes a barrier height to block the passage of charges from a potential well below the gate of FIG. clock 61, to a potential well under the floating diffusion 14 except intervals or Qc pulses lower the gate 61 below the positive potential The gate 21 receives a potential RG applied to it so that the potential of channel under grid 21 is also

  positive or more positive than the drain potential RD.

  The leakage field effects strongly affect the real potential of the channel 21, which is normally very short to reduce the charge sharing with the floating diffusion. O r lies between: (a) during the charge measurement time, a voltage sufficiently negative to establish an insurmountable barrier to the passage of charges from the floating diffusion 14 to the quenching drain diffusion 20 and (b) during the quiescent time, a sufficiently positive voltage to allow the level, 15 under the floating diffusion 14 drain the potential RD as shown Therefore, the floating diffusion 14 is brought back to the potential of

drain of rest RD.

  The time constant RC of the charge transfer part in which the floating diffusion 14 is arranged is short in this quiescent system. This is so because the part C of the time constant is the small capacitance of the substrate of the floating-point diffusion 14 and that the R part is the weak resistance presented by the channel of the cascode field effect transistor effect (which extends between the floating diffusion 14 and the quenching drain 20 when the positive impulse Gold is applied to the quenching gate 18) The quiescent of the final value of the superimposed thermal noise variations on the quenching drain potential RD occurs because this time constant is too short to integrate the spurious variations This is a sampling and holding process that extends the duration of the last value of the thermal noise variation to the next quiescent interval.

  thus creating a sound of resting.

  A method was found whereby the quiescent noise occurring in the operation of RCA Corporation's 403-column charge-coupled image analyzer could be reduced from a level of 100 electrons or about one level less than 35 electrons or about. A noise of this last level can be attributed to the MISFE Ts 15, 16 and 17 This process will be explained to

using Figure 6.

  In the time diagram of FIG. 6, (a) is the clock signal applied to the last clock grid 61 of the C 13 register. This last clock gate electrode, in current practice, is immediately followed by a gate electrode 62 The gate 62 is biased with a continuous potential and is followed by the floating diffusion 14 As an example, a three-phase clock will be assumed. The clock cycle duration is 133 nanoseconds in an analyzer 403 column image-type image-coupled load-coupled image system which provides a NTSC television signal component When the field voltage applied to the last clock grid electrode 61 of the C-register 13 becomes negative, 25 a sample load packet of an element

  of images is transferred under the floating diffusion 14.

  This transfer is represented in waveform (a) of FIG. 6 by an arrow on the voltage transition. Waveform (b) of FIG. 6 shows the timing of the positive quiescent pulse applied to quiescent. An initial impulse r_ 1 is represented on the waveform (b) followed by the pulses Or-2 and Or3; these pulses 35 are in a series of pulses $ r '. Each resting pulse Gold must normally be applied some time before the negative transition of the form.

1 8 2 2551285

  (a) when the double correlated sampling is not used But, as shown in the waveform (b), the quiescent pulses Gold used here are applied well before the negative transition of a respective clock pulse Each clock pulse has a duration t which is not much shorter than r, or time constant RC This timer allows the differential response of the differentiator 30 (at the transitions produced at the front flanks). and back of pulses or) to be sufficiently complete at the moment of appearance of an OS pulse (to make the MISFET channel 41 conductive in the sampling of the peak response of the differentiator 30 to the associated transistion with the introduction of a charge packet under the floating diffusion 14) This early onset of the quiescent pulses Gold is used in operation with double sampling

  also correlated, but for reasons that are different from those of the invention.

  The waveform (c) of FIG. 6 shows the resulting potential variation on the floating diffusion. The same potential variation, possibly with a different level of DC bias, occurs at the source electrode of the MISFET 16 and at the same time. terminals of the load resistor 28 of the image analyzer 10 The range of variations R attributable to the quiescent sounds appears in unfixed portions of the waveform (c) between the quiescent pulses Cr1 and # r-2 of

the waveform (b).

  A range of variations R ', also attributable to the quiescent noise, occurs in portions of the waveform (c) during the time between the quiescent pulses Or-2 and r-3 of the d-shape. (b) The ranges of variations R and R 'have similar amplitudes but the amplitude variations of these ranges R and R' are not in correlation The introduction of a charge packet under the floating diffusion 14 following the impulse

  Or-2 creates a sample height 5 on which is superimposed a noise of resting in the range R '.

  The waveform (d) of FIG. 6 is an idealized representation of the voltage samples appearing at the output of the amplifier 35 according to the differentiator 30 The clock noise is not considered The differentiator 30 removes the content of Low frequency in its response, this response being analyzed by superimposing the responses to transitions Differentiating the amplitude range of the transition from the waveform (c) to the leading edge of the quiescent pulse Or-2 accompanied by the resting noise of the R range amplified by the voltage gain G of the amplifier 35 causes an initial value of GR in the range of the resting noise appearing in the waveform ( d) during the Or-2 'pulse. This initial value of GR decreases exponentially until GR exp 1 (d-11) during the duration d of the pulse r-2'. The differentiation of the transition of the waveform (c) on the rear flank of the pulse er-2 which is accompanied by the noise of resting the range R 'and which is amplified by the voltage gain G of the amplifier 35, increases GR exp -l (d V -1) of the noise of This increase gives an initial value N for the exponential decay range of the total quiescent noise in the period between the pulses 6 r-2 and Or-3 '. The sum of these noise components of at rest without correlation -2 is such that on average N is equal to G (R'2 + R 2 exp 2 (d 1)) 1/2 on the trailing edge of the pulse qr-2 'At a moment from a time t later, when a charge packet is introduced under the floating diffusion 14, the quiescent noise is reduced to a value N (exp 1 (t 1)) At this time, the response of the voltage amplifier 35 at the differentiated sample height at a value G d The signal-to-noise ratio at this time has a value (R, 2 + R 2 exp-2 (dz -1)) - 1/2 exp (t 1) Since the sample GA and the quiescent noise N (exp (t T -1)) decreases exponentially, this signal-to-noise ratio remains practically constant when

the sample Gb decreases.

  Waveform (e) of FIG. 6 shows an example of a phase of two sampling pulses 15 (e-1 t s_ 2) of the pulse series S 5 which is applied by line 26 to the electrode grid

  N-channel MISFET 41 and which are used for switching samples in the synchronous detector 40.

  Although circuit 40 is considered to be of a sampling and holding nature, it is more accurately described as being of a tracking and holding nature, the last value of each sample being kept in the following sample period. In this context, the amplitude of the output response of the sampling and holding circuit 40 increases if the sampling is performed with a shorter pulse, sooner after the negative transition of G A. Since the resonance noise for the preceding picture element and the resting noise component for the present picture element are not correlated and tend to have on average the same amplitude, the shortening of as possible for a given value of d + t tends to reduce the quiescent noise If d is 33 nanoseconds and t, 40 nanoseconds, (compared to the time constant t of 35 nanoseconds of the filter 30), the quiescent noise is reduced in a ratio of three compared to quitting just before the resignation of a charge under the floating diffusion 14 With regard to the operation of the specially-described circuit, using the RCA Corporation's 403-column load-coupled image analyzer, this reduces the noise of

  set to rest below the 35 electron 10 amplifier noise.

  As indicated above, the quiescent noise is produced when quiescent floating quencher 14 is set at quiescent drain 20 due to the short time constant RC of the voltage. appearing in the broadcast

  floating 14, allowing the rest to thermal noise.

  Alternatively, the quiescent noise can be lowered by taking steps to lengthen the time constant RC on the floating diffusion 14 This can be done by restoring the floating diffusion, not the potential of the quiescent drain. This latter forms the integration with respect to the time of the charge under the floating diffusion 14 This integration lengthens the effective time constant RC associated with the floating diffusion. Other modes of resting using these

  principles will be described in this description.

  Differentiator circuits other than the high-pass circuit RC of the resistor 31 and the capacitor 32 can of course be used.

  for the implementation of the invention.

  Extending the decreasing response to the leading edges of the quiescent pulses to overlap the response to the measurement of the following picture element sample has a pronounced effect on the video peaks (in the scanning direction of lines) of the output signal from the analyzer of images detected in synchronism This will be explained to

using Figure 6 (d).

  The loss of the DC baseline reference during the passage of the output signal from the image analyzer 10 by the differentiator 30 cooperates with the loss of DC component recovery, except during the line return to produce the Next effect The exponential decay response at the leading edges of each quiescent pulse has a tail with an amplitude component that depends on the value of

  the image sample admitted under floating diffusion 14 before the quiescent pulse.

  The lower frequency components of this previous image sample are similar to those of the next image sample admitted under floating scattering 14 after this quiescent pulse. Therefore, the exponential decay positive response at the front edges of the quiescent pulse counteracts the exponentially decreasing negative response of the next image sample This opposite relationship reduces the lower frequency modulation of each harmonic of the register C clock frequency in the signal This opposition relationship and the higher frequency components of the two image samples are dissimilar so that the cancellation effect is less pronounced. This increases the amplitude of the sidebands of the image analyzer. higher frequency of modulation

  harmonics of the clock frequency of the register C relative to the amplitude of their sidebands 35 of lower modulation frequencies.

  The factor by which the output signal of the synchronous detector 40 is accentuated at each frequency

2-3 2551285

  is substantially constant despite variations in amplitude of the quiescent pulses or variations in the difference between the amplitudes of the successive samples of picture elements The emphasis value depends mainly on the degree to which the tail of the positive response exponentially decreasing in contrast to successive image samples is reduced in its overlap on the negative response to exponential decay of the last sample of images The value of the tail amplitude reduction depends on the time elapsed between the leading edge of In other words, the tail amplitude reduction value depends on the sum (d + t) of the duration d of the quiescent pulses and the reception of the next image sample load under the floating diffusion. the resting pulse and the interval t between the trailing edge of the pulse

  resting and charging this admission.

  The duration of the quiescent pulse can be chosen to obtain a flat video response at the output of the synchronous detector 40 (or in this case to obtain a cutoff or an accentuated video response if desired). The interval t can be set to a degree which also affects acceleration. As noted above, this setting is limited by the desire to avoid the noise of resting. The values d, t and can be chosen from video response above the peaks, the video response can be adjusted to

  will by using a variable resistor interposed in series with the MISFET channel 41 to introduce a variable degree of cutoff.

  The use of the differentiator 30 before the synchronous detector 40 solves another problem appearing in the charge coupled image analyzers During the parallel transfer of the charge packets from the register B 12 to the register C 13, a transient change occurs in the polarization of the substrate This transient causes a level shift in the response to the image analyzer The level shift is manifested as an initial brightness excess with pronounced decay in the first few video samples taken from the register C 13 in each line This excess brightness on one image edge decreases at a speed of image elements to another sufficiently low to be completely suppressed in the response of the differentiator 30 A similar effect that appears at the top of the image results from

  transfer from register A to register B during the vertical return interval This event is also reduced by the differentiator 30.

  The other resting modes already mentioned will now be considered. The difference between the quiescence of the floating diffusion 14 at the quiescent drain potential and the quiescent at the channel potential will be explained using

  potential profiles of Figures 3, 4 and 5.

  FIG. 3 is a potential profile diagram describing the quiescence of the floating diffusion 14 at the channel potential. The channel potential is established by the most positive excursion.

  O r 'which is not as positive as RD.

  The most positive excursion has a lowered barrier height which is overcome by the charge carriers in the floating diffusion region 14 and under the gate 21, until a potential level more positive than the height. This potential is less positive than RD The quiescence is as high as the barrier with a slight shift due to the current of the black which circulates under the effect of the thermal excitation of the carriers of load The variation of the offset due to this thermal excitation is a main cause of the low frequency peak of the baseband of the modulation transfer function when the quiescence is at potential

of the canal.

  FIG. 4 is a potential profile diagram of a preferred way of restoring the floating diffusion 14 to a channel potential. The quiescent gate 18 is controlled in a range containing the quiescent drain potential. The positive vr excursion is not the channel potential at which the floating diffusion 14 is reduced. On the contrary, the continuous potential RG applied to the gate 21, this potential being easily filtered to eliminate the noise, is made more positive that the draining drain attention RD A potential barrier 63 is erected under the door 21 and the floating diffusion 14 is brought back to the barrier potential 63 with a slight positive shift This shift occurs because the charge flow does not occur. only until the barrier potential can no longer be overcome. FIG. 5 is a potential profile diagram of the resting at the barrier potential of the floating diffusion 14 under the gate 21 when another DC gate 64 is interposed between the queuing gate 18 and the drain This other gate 64 is in the charge coupled image analyzers currently manufactured by RCA Corporation. This other gate 64 is internally connected to the gate 21 of its devices. the bottom of the quenching grid 18 when the latter is no longer positively pulsed is towards the quenching drain 20 This is so because the potential barrier under the grid 21 tends to be less easily overcome than the potential barrier under the gate 64 due to the leakage field from the quenching drain 20 reducing the barrier height on the gate 64. The invention has been dec in view of a field-transfer type charge coupled image analyzer, but it can also be applied to other types of charge-coupled image analyzer such as transfer type

  between lines or line transfer.

  Of course, various modifications may be made by those skilled in the art to the embodiments described and illustrated by way of non-limiting examples without departing from the scope of the invention.

Claims (4)

  A camera comprising a charge-coupled component image analyzer for producing a series of spaced-apart image response samples at a clock-fixed frequency, characterized in that it comprises an electrometer output stage (14-20) comprising a floating element for providing said series of spaced image response samples, said floating element (14) being periodically returned to rest in response to pulses (at) at said clock frequency a differentiating circuit (30) for differentiating said image response samples with respect to time, and for suppressing certain frequencies in the differentiated image samples produced, said certain frequencies being lower than an angle frequency and containing substantially all of the energy of a flicker noise, and a sample and hold circuit (30) to which the samples are supplied differentiated image response from said differentiating circuit, for sampling and holding samples supplied in response to sampling pulses (os) at said clock rate to produce a response. images without spaces.   Apparatus according to claim 1, characterized in that said floating element (14) is reset to a channel potential of a charge transfer channel (19) of said output stage to reduce the noise associated with the setting. rest. An apparatus according to claim 1 or 2, characterized in that quiescent pulse (OR) applications precede the respective times at which charge packets are introduced under said floating element (14) at intervals which do not occur. are not significantly shorter than the inverse of   angle frequency in radians per unit of time.   4 Apparatus according to any one of   Claims 1 to 3, characterized in that the   The angle frequency of said differentiator circuit (30) is selected from the baseband frequency spectrum of said image response samples to produce a video emphasis of the response to   the image of said sampling and holding circuit (30).   Apparatus according to any one of claims 1 to 4, characterized in that said differentiator circuit consists of a high-pass filter   comprising a series capacitor (31), and a shunt resistance (32).   6 Apparatus according to any one of   Claims 1 to 5, characterized in that said sampling and holding circuit comprises a   insulated gate field effect transistor responsive to the clock frequency sampling pulses (Os) at its gate for receiving samples of said differentiator circuit for said sample and hold circuit, and a capacitor (42) used to maintain the response to the sampled image admitted to the output of said   sampling and holding circuit.