Charge Injection Device Readout

1. Field of the Invention

This invention relates in general to charge injection imaging devices (see U.S. Pat. No. 3,786,263) and more particularly to apparatus for reading out the information stored by such a device.

2. Description Relative to the Prior Art

Current thinking with respect to the readout of a charge injection device (CID) is to employ any of a number of techniques for sensing capacitance changes of the charge-collecting image sites or pixels. Techniques for sensing such capacitance changes have appeared in the literature, although on such technique is described below for purposes of facilitating understanding of the invention.

Two problems are associated with capacitance readout techniques: (1) capacitance differences from pixel to pixel, which may result from any of a number of structural sources within a CID, manifest themselves as pattern noise in the device output and (2) under certain illumination conditions, the visual image corresponding to the output of the CID is distorted intensity-wise. Although the first of the above-noted problems is relatively easy to appreciate, why the second problem should occur is another matter: An analysis is presented below to explain the source of the second problem.

1. Proceedings of 1975 International Conference on the Application of CCD's, Oct. 29 - 31 (San Diego, California) pages 351-260.

2. U.S. Pat. No. 3,949,245

The present invention provides a quasi-capacitive readout technique for use in combination with a CID imager, and which technique nullifies the second of the above-noted problems and, in so doing, contemporaneously obviates the first, or pattern noise, problem as well. In essence, the invention--while utilizing the capacitance change occurring at a pixel site as a measure of light received at such site--works to restore a quiescent capacitance (and voltage) level at such site, utilizing the discharge of a reference capacitor for such purpose. Thus the voltage change across the reference capacitor (which results from its discharge) is representative of the light received at the pixel site.

The invention will be described with reference to the figures within.

FIG. 1 is a plan view of a part of a basic CID structure.

FIGS. 2a through 2d are illustrations useful in describing a typical prior art readout technique over which the invention is an improvement.

FIG. 3 illustrates the timing associated with the readout technique of FIG. 2.

FIGS. 4a and 4b are drawings useful in illustrating the nature of a problem inherent in CID readout techniques similar to that of FIG. 2.

FIGS. 5a through 5d are drawings which serve to illustrate the cause of the problem inherent in prior art readout techniques.

FIG. 6 is a schematic diagram illustrating a presently preferred embodiment of the invention.

Reference should be had to FIG. 1 which shows four pixels of a conventional CID 10: Each pixel, as is known, comprises a row-addressable MOS capacitor and a column-addressable MOS capacitor. By applying suitable address signals from processing electronics 12 to, for example, X' and Y' busses, a pixel 14 of the CID may be interrogated to produce a readout signal corresponding to the light information received at such pixel. (In the addressing scheme of FIG. 1, it will be noted that the row bus is addressed; then the column bus is addressed--readout being "column readout" as evidenced by the double-headed arrows which connect the processing electronics 12 and the column busses).

To appreciate the workings of a typical pixel readout technique which is based on capacitance-change sensing, reference should be had to FIGS. 2a through 2d and the accompanying timing diagram of FIG. 3: Up to time T=1, light .lambda. causes a corresponding number of mobile minority carriers (holes) 16 to form in the body of the N-type semiconductor. Since the row gate X is at -20V, as shown, the minority carriers flow into the depletion region under the gate X. At time T=2, under control of the processing electronics 12, the row gate X potential goes to zero volts, causing the minority carriers 16 to shift to the depletion region under the Y gate, which is held at -10V by the processing electronics.

At time T=3, an electronic switch 18 within the processing electronics is opened, causing the potential of the Y gate to float. (To be noted is that the Y gate potential does not change at time T=3, this being because the Y gate, and the depletion region thereunder, constitute, in actuality, no more than a charged capacitor which has been charged, and then removed from its charging source).

At time T=4 (and with the float switch 18 still open), a positive-going pulse is applied, via a coupling capacitor 20 within the processing electronics, to the Y gate. This causes the stored minority carriers to be injected into the semiconductor and gathered by the P-type epitaxial layer. Then at time T=5 (and with the float switch 18 still open), the potential of the Y gate is compared with the pre-injection potential (-10V) by a difference circuit 22, the comparison resulting in a video signal voltage representative of light .lambda. received at the pixel.

The mechanism for the production of the video signal voltage may be appreciated from FIGS. 2b and 2d: In FIG. 2b, a relatively large capacitance, corresponding to the close proximity of the depletion level 28 to the Y gate, is charged to -10V. Since, as indicated in FIG. 2d, the Y gate is still floating after the charge 16 has been injected, Y gate electrons are trapped on the Y gate. Such being the case, fixed (immobile) holes appear in the semiconductor material, thereby to neutralize the electronic charge on the Y gate. The immobile holes so produced characterize a depletion region 28' which, to be noted, is not only greater in depth than the depletion level 28, but is also greater in depth than the original Y depletion region (28", FIG. 2a). The relatively small capacitance that corresponds to the depth of the FIG. 2d depletion region 28' exhibits a larger voltage than that which corresponds to the capacitance associated with FIG. 2b, "signal voltage" being variable inversely with capacitance when the charge level is fixed (because of the action of the float switch 18). Thus, the FIGS. 2b-to-2d capacitance change translates directly into a signal voltage change.

Assuming one were to use a CID capacitance readout scheme as disclosed in connection with FIGS. 2a through 2d, faithful reproduction of a scene as depicted in FIG. 4a would occur so long as the light images falling on the imaging face of the CID were not too bright. Were the light images to be brighter than a threshold level, the CID, per FIGS. 2a through 2d, would be productive of an intensity-distorted display as depicted in FIG. 4b. (The intensity-distortion problem has heretofore been sidestepped, principally--it is believed--because it was not fully appreciated). While one solution to such intensity-distortion is to employ a filter, or the like, over the imaging face of the CID, thereby reducing the light level reaching the CID face, such a technique leaves much to be desired since it sacrifices the sensitivity of the CID to assure against exceeding the above-noted brightness threshold. By analysis, the present invention identifies the cause of the intensity-distortion problem and provides apparatus which--while still basically employing capacitive readout--nullifies the influence of such problem cause:

FIGS. 5a through 5d depict 2 pixels, say 14 and 40 of FIG. 1, wherein the pixel 14 shall be assumed to be addressed, for example, as indicated in connection with FIGS. 2a through 2d and FIG. 3. Assume further that the pixel 40 has been over-illuminated (time T=1), the depletion region under its X" gate so overfilling with minority carriers that some of such minority carriers reside beneath the Y' gate of the pixel 40. At time T=2 (FIG. 5b) the voltage on the X' bus associated with the addressed pixel 14 is changed from -20V to zero volts, causing the minority carriers of the pixel 14 to shift to beneath the Y' gate thereof. (Since the potential on the X" gate of the pixel 40 remains at -20V at time T=2, the storage of its minority carriers does not change from that depicted in FIG. 5a). At time T=3, the potential of the Y' gate of the pixel 14 is floated.

In response to a positive "injection" pulse (time T=4) the minority carriers beneath the Y' gate of the pixel 14 are injected (FIG. 5c), the minority carriers of the pixel 40 shifting to reside, at least moreso, under the X" gate thereof. After the "injection" pulse (time T=5), the situation depicted in FIG. 5d obtains. As was the case in connection with FIG. 2d, the Y' gate capacitance of the pixel 14 at this time has decreased as a result of the charge-injection--but such charge-injection has had no effect on the storage of minority carriers beneath the Y' gate of the over-illuminated pixel 40. Since the capacitance change which is measured per the above-described technique is, in actuality, the capacitance change occurring in a "column" of capacitors connected in parallel, i.e., the "Y' capacitor" of the pixel 14 is in parallel with the "Y' capacitor" of the pixel 40, the Y' column capacitance of FIG. 5d is greater than when the pixel 40 is not over-illuminated. In other words, in FIG. 5d, the capacitance corresponding to the depletion level 40', when summed with the capacitance corresponding to the depletion level 14' produces a greater column capacitance than when the capacitance corresponding to the depletion level 40" is summed with the capacitance corresponding to the depletion level 14'. Because such column capacitance is greater than it ought to be in this instance, all pixels associated with the Y' column which are not over-illuminated, will, when addressed, produce a readout voltage which is lower than what would be representative of the light falling thereon. Attendantly, the display of FIG. 4b results.

Having identified the source of the intensity-distortion problem, a new problem obtains, viz. how to provide capacitive-based readout without having to read column capacitance.

The invention--while utilizing the mechanism of pixel capacitance-changing within a CID for purposes of determining the level of light received--does not directly convert, say, a column capacitance to a signal voltage. Rather, the invention teaches, in the readout of a CID pixel, the supply of charge to the pixel being read, such charge restoring the capacitance and voltage of the pixel readout gate (and of the column, since all pixels in the column are connected in parallel) to its pre-light-integration level. The supply of charge is from a pre-charged capacitor, the voltage change thereof resulting from its discharge being a video signal that is totally independent of column capacitance; and attendantly a display therefrom will not be intensity-distorted as a result of the over-illumination of column pixels.

Reference should now be had to FIG. 6: A schematically depicted CID 50 is row-and-column addressable by processing electronics 52 and 54 as per the discussion relating to FIGS. 2a through 2d. Gates 52--which may be on-chip MOS switches--are respectively connected to each column bus and are adapted to open when their respective columns are addressed, thereby effectively insulating the columns from each other. (Gate-operation is, for example, designed to occur while "measuring", which is a set-time after occurrence of the "injection" pulse). The output circuits of the gates 52 are connected to one input A of an operational amplifier 54 having a reference voltage V.sub.R applied to its other input B. The reference voltage V.sub.R equals the column potential -10 volts) discussed above in connection with FIG. 2. A feedback capacitor 56 is connected between the output and input B of the amplifier 54; and such capacitor is periodically charged to the reference voltage V.sub.R through a gate 58. That is, the gate 58 periodically removes the reference voltage source from across the feedback capacitor 56 under action of a signal ("measure", time T=5 per FIG. 2d) calling for a pixel to be read.

In reading the minority charge accumulation, say, of the pixel X.sub.1 Y.sub.1 (FIG. 6), the gate 58 allows the capacitor 56 to charge (between times T=0 and T=5) to the reference voltage V.sub.R. Up to time T=5, the amplifier 54 has virtually no output since its A and B inputs are identical. At time T=5, however, the gate 58 is inhibited; and the gate 52.sub.1 opened. Assuming the Y.sub.1 potential has, as a result of minority carrier injection, gone from -10V to -12V, the amplifier output goes positive by 2 volts, thereby causing charge (electrons) to be drawn from the Y.sub.1 gate and resulting in its potential returning to -10V. Since the capacitor 56 works to restore the column voltage--and which voltage is not affected as a result of over-illumination of other column pixels--it develops an information representative voltage-change corresponding to its charge-change, the value of the capacitor 56 being a constant.

The invention has been described in detail with particular respect to implementations thereof, but it will be appreciated that variations and modifications can be effected within the spirit and scope of the invention. For example, while the CID of FIG. 6 is operated according to the readout scheme of FIGS. 2a through 2d, other CID readout schemes which are based on "column" (or "row") capacitance change are practicable with the invention. See, for instance, the readout schemes appearing on page 35 of ELECTRO-OPTICAL SYSTEMS DESIGN, Oct., 1975.