JPS5923675A - Removing method of noise in mos type image sensor - Google Patents

Abstract

PURPOSE:To eliminate noise caused by variance of sensors between electrodes and discordance of scanning pulses etc. by switching an MOS switch connected in series to individual sensor element of an MOS type image sensor twice consecutively, and reading the difference between signals of the first time and the second time. CONSTITUTION:A pre-amplifier 13 that amplifies signals from a signal line and a driving circuit 17 that drives the MOS switch of the sensor are provided in the noise detecting circuit of a one-dimensional or two-dimensional MOS image sensor. Reset pulse is applied from the circuit 17 to the amplifier 13, amplified by the amplifier 13 and integrated output is applied to sample holding circuits A14 and B15. Each of sampling pulses H, I is given from the circuit 17 to circuits A14, B15 respectively, and the signal due to the switching of the first time is held by the circuit A14 and the signal due to the switching of the second time is held by the circuit B15. The difference between the first and second signals is read by a differential amplifier 16, and noise caused by scattering of the sensor and discordance of scanning pulses is eliminated perfectly.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a noise removal method for an image sensor that performs switching using a one-dimensional or two-dimensional MO8$ transistor.

The biggest problem with so-called MOS type image pickup devices using one-dimensional or two-dimensional MOS switches is improvement of the S/N ratio. In general, MOS image sensors have much more noise than image sensors that use CCDs, and an image sensor that does not have countermeasures against noise will have noise (103 to 104 times that of No. 6). Fixed pattern noise due to drive pulse feedthrough that occurs through parasitic capacitance between gate electrode and source or drain electrode, clock noise due to parasitic capacitance between clock wiring and horizontal output line in two-dimensional image sensor, random noise from vertical and horizontal switches The causes of fixed pattern noise, which is the main noise, will be explained using Figures 1(a) and (b).Figure 1(a) shows the circuit of the most basic MO8m sensor. It shows the configuration.

1 is an MO8i transistor for switching, 2 is a fumetodiode serving as a sensor, 3 is an electrode interrogation 1.4 is a readout resistor, and 5 is an air source.

MOS) Ranjista 1 game!・I hit, (gek 7jt
When the scan pulse is applied and the MOS transistor is switched off, a feed-through of the scan pulse occurs due to the gap between the electrodes, but a spike-like sound is heard on the line.

Normally, this noise is much larger than the signal, and π
This is caused by variations in the spacing between the poles and variations in the waveform and peak value of the scanning pulse.This situation is shown in Figure 1 (bl). He will become a monk.

In response to this, various noise reduction methods have been proposed in the past. Examples of typical noise reduction methods are shown below. Figure 2 (a
), (hl l1ill are two examples of sound reduction methods.

In Fig. 2 (at, (1)), 6 is a drive circuit consisting of a shift register, etc., 7 is a signal line, Fll: 1 tone line, 9 is a MOS) transistor, 10 is a claw 1-dimate, MOS transistor play It is assumed that the characteristics of adjacent elements in an integrated circuit are made in a completely integrated manner, and that the capacitances of adjacent elements are approximately equal. The idea is to apply a scanning pulse to the transistor again to generate noise, and then use a differential amplifier to extract only the signal component. In addition, in the circuit example shown in Fig. 2 (bl), the M
The OS capacitor is arranged equally and symmetrically with the MOS transistor, and the noise component is removed by the differential amplifier as described above. These methods can remove Kana υ noise, but no matter how integrated the circuit is, it is difficult to make the mutual characteristics uniform among all of the large number of bits, and even in ζ, the variation becomes a problem. In particular, since the switch array has one integration port, the signal is drawn in an integrated form. Furthermore, the process of resetting the charge necessary for the integrator is not shown because the explanation would be complicated. In the figure, the diagonal part indicated by A is the optical signal, and B is the noise component due to the variation in capacity.

Another noise suppression method is to integrate the pulse noise generated at the leading edge and trailing edge of the scanning pulse and remove it.

This is based on the fact that the tone components generated at the rise and fall are equal. Figure 3 (b) shows the integrated output. In this figure as well, the rillet process is omitted. As shown in the figure, the initial signal appears as the sum of the noise (T3 part) and the optical signal (human part) due to the noise at the rising edge of the scanning pulse, but if the rising edge and the rising edge of the scanning pulse If they are completely equal on the downward slope, then the downward slope (
It is impressive that C71 cancels out the rising noise and only the optical signal remains after scanning. In an example of application of this method, in a two-dimensional MO8 type Inoji sensor, approximately 15 + IL)
There are reports that it can suppress fixed pattern noise of one degree.

However, this method also requires the assumption that the rising and falling noises must be equal, and if this is not satisfied, it will not be possible to remove the noise. This assumption breaks down when low frequency noise enters the signal line, when there is an on-leak in the MOS transistor, or when the CMO
This is the case with a switch that performs complicated back gate processing, such as an S analog switch. This is particularly noticeable in the case of CMOS analog switches that include address decoders.

It is an object of the present invention to eliminate such drawbacks of the conventional method, and to eliminate unevenness of noise caused by variations in the amount between electrodes and the rising and falling edges of scanning pulses. An object of the present invention is to provide a method for removing noise in an MO8 type image center.

According to the present invention, in an image sensor using these MOS switches, a signal detection circuit having one or more circuits having a signal holding ability is used, and each sensor element is switched twice in succession, and the first switching A method for removing noise in an MO8 type image hinger is obtained, which is characterized in that the read signal or both signals are held in a circuit having the ability to hold the signals, and noise is removed by taking the difference between the two signals.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of a method of removing noise in a Muku08 type image seven-band array according to the present invention and a detection circuit for realizing the method will be described in detail with reference to the drawings.

FIG. 4 shows an embodiment of a detection circuit that implements the noise removal method in the M 08 i11 image sensor of the present invention, and FIG. 5 shows the timing of scanning pulses when the detection circuit of FIG. 4 is used. l! -1H waveform is shown.

In FIG. 4, reference numeral 13 is an amplifier for amplifying the input signal to some extent, and may also include a signal-to-vj converter circuit such as an integrator. Since it is easier to understand the signal by integrating it,
For convenience of explanation, it is assumed here that 81 dividers are included. 14
．． 15 is a ripple hold circuit with the same performance, and the one that holds the signal obtained from the first switching is the sample hold circuit 14, and the one that holds the signal obtained from the second switching is the sample 15. It is called hold circuit B. 16 is a differential amplifier that takes the difference between the outputs of two sample and hold circuits; 17 is a MOS switch that drives a sample pulse to the sample and hold circuit; and, if the preamplifier 13 includes an integrator, provides a reset pulse to the integrator. This is a drive circuit that gives Further, in the figure, E represents the output signal of the preamplifier, F and G represent the output signals of the sample and hold circuit A1B, respectively, and H and 111 represent the sample pulses to the sample and hold circuits jl&A and B, respectively. HlI is EXF in Figure 5
, GlHlI. In Figure 5, A, I3
-4 represents an optical signal and noise, respectively, and in this figure, 4 shows a state in which the optical signal is also changing.

This detection port interlocking operation will be explained below along with the signal flow.

First, a switching pulse is sent to the MOS switch of each bit, which switches twice in conjunction with each other, as shown in FIG. As a result, when the signal on the signal line is amplified by the preamplifier 13 including an integrator, the output signal becomes as shown in FIG. 5B.

As can be seen from the figure, the first signal that is checked is a combination of an optical signal and noise, and the second signal is only the noise immediately after the charge generated by light is read out. This noise is generated by switching the same MOS switch using a scanning pulse generated by the same drive system, so it is equal in amount to the noise generated by the first switching.

Now, when such a signal is output to the preamplifier 13, the sample and hold circuit A1B outputs the signal shown in FIG. When a sample pulse is given at the timing shown in I, the sample hold circuit I3 holds the signal obtained by the first switching, and the sample hold circuit I3 holds the signal obtained by the second switching. .This is the signal shown in Figure 5, ?,0.

Finally, the difference between these signals is obtained by the differential amplifier 16, and the result is shown as an output signal.

As is clear from the figure, at the timing indicated by JXK, the output signal is only an optical signal. This is because, as mentioned above, the noise components of the signal obtained in the first switching and the signal A obtained in the second switching are equal, and noise can be removed by taking the difference between the two.

The above is the operating principle of the noise removal method of the present invention. According to this method, even if there is variation in the noise of each bit, as is clear from B in Figure 5 and the waveform of the output signal, the scanning Noise can be removed even if the magnitude of the noise caused by the rise and fall of the pulse is not equal. Moreover, the detection circuit is not so complicated as compared to the conventional one, with only one sample and hold circuit added.

Finally, some embodiments of detection circuits that implement the noise removal method in an image sensor of the present invention will be shown.

FIGS. 6(a) and 6(b) show a 2f& class detection circuit.

In FIG. 6(b), 18 and 19 are two equivalent or peak hold circuits. Figure 6 (al is a detection circuit in which one of the sample and hold circuits is removed from the detection circuit in Figure 4 and 1. If it does not appear, there is no need to hold the second switching.

When we used this detection circuit to drive a 960-bit one-dimensional image sensor by pressing the CMOS analog switch with a switch, the operation was good, and no signal fluctuations were observed due to the omission of one sample-and-hold circuit. .

Next, FIG. 6(1)) uses a peak bold circuit as the signal storage fft circuit, and the signal output from the preamplifier 13 may be sharp such as a differential pulse.
There is no need to attach a special divider to the preamplifier section. The signal lines still entering the peak bold circuits 18 and 19 from the drive h circuit J7 are for resetting the peak hold circuits. The operation is almost the same as that using a sample and hold circuit.

Next, FIG. 7 shows a digital detection circuit using a converter. In the figure, 20 converts the analog output of the preamplifier 13 into a digital -1-ru△/1) converter 21. 22 is a register that digitally holds the signals from the first and second switching, respectively, and 23 is the contents of the register 21. This is a reducer that subtracts the contents of 22 registers from . Further, 21 to 23 can be configured by software by using a CPU. The operation is almost the same as the previous example, with the only difference being the analog gain. Each of the signals read out twice consecutively from each sensor element is A/D converted, held in a register and output, the difference is taken, and all noise is removed.

As explained above, when the method for removing noise in a MOS type image sensor of the present invention is used, noise due to variations in electrode interrogation Jλ and noise mismatch occurring at the rise and fall of the scanning pulse, etc., which were present in the conventional method, can be eliminated. One-dimensional or two-dimensional M
Useful for Os type image sensors.

[Brief explanation of the drawing]

Figure 1 tj, mu 408 Basics of image sensor 11- (1
t1 signal and its signal waveform, Figure 2 is an example of the conventional M 1'11 tone removal method, l-Q 3 times is the conventional 1 noise removal (=/ko l'! 円i [I output signal Waveforms Figure 2114 is an example of a detection circuit that implements the Mll:W removal method of the present invention↓9. Figure 15 is the signal ?+f type of each part of the detection circuit shown in Figure 4, '64E G l 1. Fig. 7 shows an embodiment of a detection circuit that realizes the noise removal method of the present invention. ...Reading resistor 5...Power supply 6...Drive circuit 7...Signal line 8...P□(
G sound line 9...λ40S) Ranjisk 10...Futa dimade et...Differential amplifier +2...Noy ×
MOS capacitor for Gancell 13...Preamplifier
+4.15...Sample hold circuit +6 Differential amplifier 17...j pivot circuit 1B, 19...Peak bold circuit 20...Include/I) Convergence 21.22
Regio 1 Diagram (θ)

Claims (1)

[Claims] 1. In a 1-dimensional or 2-dimensional MOS, W image sensor, the MO8 switch connected in series to each seven element is switched twice in succession, and the readout is carried out at the first time (no. Both of the signals read out for the first time and the second time are held by a circuit having signal holding ability.
MO characterized in that the difference between the second signal is used as the read signal.
How to remove noise from an 8-inch image sensor. 2. If there is one circuit with a signal holding fIu force, use two sample and hold circuits, and calculate the difference between the first signal held in one sample and hold circuit and the signal that does not pass through the sample and hold circuit, or two. i'l' iT! A method for removing noise in an MO8 type image sensor according to item 1. 3. The circuit with signal holding ability uses two peak bold circuits, and the peak of the signal due to the first switching and the peak of the signal due to the second switching (the peak of No. 17 is bolded with the corresponding peak bold circuit, and the differential Read the difference with an amplifier, = MP = 04, a circuit with signal holding ability is A/1] using a converter and a register, 2
The signals resulting from consecutive switching times are stored in registers, and the difference is calculated using CI).
ii 11 Requested Scope 1st Item 1111 Method for removing snoring caused by MO8 type image center.