GB2344245A

GB2344245A - Cell-based image sensor

Info

Publication number: GB2344245A
Application number: GB9922088A
Authority: GB
Inventors: Laurence Ray Mccolloch
Original assignee: Hewlett Packard Co
Current assignee: HP Inc
Priority date: 1998-10-05
Filing date: 1999-09-17
Publication date: 2000-05-31
Anticipated expiration: 2019-09-17
Also published as: GB9922088D0; GB2344245B; JP2000151905A; DE19927991A1

Abstract

An image sensor system (10) has an array of plastic moulded lenses (30) which are used to de-magnify an image. Each lens works with one photosensitive semiconductor die (32) as a cell, and together the cells have slightly overlapping fields of view. Each die has extra pixels (82,84,86,88,90) to allow for manufacturing tolerance and recombination of the image. The individual dies 32A-E can be separate carriers or on a single carrier. The sensor system is preferably part of a colour scanner using LED's as the light source. The arrangement avoids the use of SELFOC lenses and does not require the precise positioning of the sensors required in prior art systems.

Description

1 2344245 CELL-BASED IMAGE SENSOR The present invention relates generally

to image sensors and more particularly to scanners for document scanners, copiers, facsimile and multifunction peripherals that include scanning devices.

In the past, image sensors of different types were used in scanners, copiers and facsimile (fax) machines to gather an image, one line at a time.

For example, in a conventional flatbed scanner, a single lens optical system is used. Originally, an image on a document would be focused through a single lens which would view the entire width of the document and provide the image to a single photosensitive semiconductor die. The lens usually consists of several optical surfaces. The difficulty with this approach was that the distances between the document and the lens and between the lens and the semiconductor die had to be fairly large which meant that the scanner tended to be fairly tall or thick. This led to extremely bulky machines.

Single lens optical systems in flat bed scanners have very high 'T' numbers. The "f" refers to the angle that light is collected. The higher the 'f' number, the less light is collected. The high 'J" number is the primary reason that very bright lights are required in single lens scanners.

A significant improvement was achieved by the creation of flatbed scanners where the optical path is folded with mirrors to shorten the height of the machine. Unfortunately, this increased the cost of the machine. First, a number of additional precision rnirrors were required. Second, the mirrors contributed additional light loss. Third, expensive semiconductor die was required. The die had to be long enough to have enough pixels to see the entire width of the image. The width of the die commonly had to be wider than minimum to have an acceptable aspect ratio (length/width). If the die had too high an aspect ratio, it would not be stable and may break during semiconductor processes such as sawing and die attach. Since the die is wider than minimally required for the circuitry to get an acceptable aspect ratio, it is more expensive than if it could be made at minimum circuit width. And, fourth, the machines still tended to be relatively thick.

I 2 To develop less expensive and more compact machines, SELFOC lenses were used in a different type of scanner sensor called a contact image sensor. The SELFOC lenses are a plurality of rods placed on end. The path of light is along the axis of the rods. The material that makes up a SELFOC lens varies in index of refraction with the radial distance from the axis of the rod. The radially gradiant index of refraction allows the rod to act as a lens. Each rod acts as a lens with a small field of view. By stacking the rods side by side in a line, they can transfer a linear image. When the SELFOC lenses are stacked in line, the field of view of each rod lens overlaps. To create a linear image, the rods must be cut to a length and positioned such that the image of each lens is a positive (non-inverted) image at 1:1 magnification.

Contact image sensors are much more compact than single lens systems. They are generally less expensive, have a much smaller depth of field, and produce a lower image quality.

There are two methods of manufacturing lenses. One is to use glass rods. These glass rods are made by dipping them into a dopant and letting the dopant diffuse into the rods. It is the dopant which provides the gradient of defraction. Then, the rods are cut to length, bound by an epoxy, and the ends ground. This tends to be a relatively time-consuming and expensive process.

A cheaper approach is to use plastic which is extruded with a number of extrusion processes, each with a different index of refraction. The multiple extrusions are later fused to create a more continuous gradient in the index of refraction. The plastic SELFOCs generally still have a stepped index that does not smoothly change, so they have poor image quality. A major drawback with the conventional contact image sensor is that the photosensitive semiconductor dice cannot be niade as a single piece. Therefore, a large number of the dice must be cut to exceedingly precise tolerances and then exceedingly carefully assembled side to side to create the long line which is required to match the width of the document. This creation of a long line of dice is the most difficult assembly process in making a contact image sensor. The sensor dice is also the most expensive part of these sensors.

The semiconductor dice must be assembled to maintain continuous pixel spacing, even between the pixels between adjacent die. There cannot be an accumulation of tolerance variation, across all twenty or so of the semiconductor dice it takes to make a contact image sensor. The pixels of a conventional contact image sensor have a poor "fill factor". The pixels at the ends of the die must be reduced in size by what ever amount is required for the 3 die sawing process, the gap between the die and the minimum spacing for active area of the die to the saw edge. To make all pixels in the array respond equally to light, all the pixels must be reduced in size, the same as the end pixels. Fill factor is the active area of the pixel divided by the area available for a full pixel (Uthe pixel per inch specification). A 600 pixel per inch contact image sensor typically has a 30% fill factor. Low fill factor means that the pixel collects less light and either the light has to be made brighter or the scanner slower. As the technology advances toward 1200 pixels per inch, it will not be possible to make contact image sensors.

In a conventional contact image sensor, die sawing and die attach processes are difficult in manufacturing because the tolerance the die have to be sawed to is very precise (10 microns is typical), and achieving assembly tolerances to maintain the spacing is very difficult. It is too difficult to do with standard automatic die-attach machines.

The sawing and assembly processes often cause the semiconductor dice to chip and crack, causing poor yield and high cost. The width of the die is often about 0.027 inch wide by 0.5 to 0.8 inch long. The non-square aspect ratio makes it very difficult to saw the dice. Naturally, the fewer dice required, the fewer dice have to be butted together. The aspect ratio that is acceptable seems to be between 20 and 30 to I which means manufacturers often make the dice wider so that they can use fewer dice for the width, but this results in wasted die area, which results in added expense.

Alignment is so complex that it generally tends to be done by hand. But this often leads to severe losses almost to the extent of 30% yield loss even in mature operations. Alignment errors often result iii the scrapping of an entire array of sensor die. Since the die are the largest cost in a contact image sensor this yield loss has a large impact on cost.

Scanners previously have had white light illumination and did color separation by prisms or filters. The white lights were typically a tube lamp, such as a florescent. The lamps were not color stable, and required regulation to control intensity and color. They also had limited life.

With LED's, it is now possible to do color separation by illumination. Using three colors, or more, the detector is read three times with the color of the illumination changed for each exposure. More than three LED's can be used to provide more realistic color (i.e., improve color rendition).

A further difficulty relates to mounting dice on PC boards. The pattern of PC boards tends to float to the edges of the board, which means that from an optical point of view, it is I 4 difficult to precisely position, the semiconductor dice in such a way that it aligns to the optics of the sensor.

A less expensive, compact system that has quality similar or better than the single lens system has long been sought by those skilled in the are, but has long eluded experts in the field.

A series of plastic molded lenses are used to image a line onto a series of photosensitive die. Each lens works with one d ' ie to create a cell. The cells have slightly overlapping fields of view to allow for manufacturing tolerance. The overlap may be more than minimum to help with reconstruction of the image. The lenses can have magnification (demagnification) such that the die can be smaller than the image.

The lens in each cell can be a single optical element, but most often will have several optical surfaces. The use of more optical surfaces can be used to correct for optical aberrations, such as chromatic aberration. The lenses could also be reflective, or could be a mix of refractive and reflective lenses. The optical surface shapes may be spherical, aspherical, diffractive, or a mixture thereof.

The advantages of the present invention will become apparent to those skilled in the art from a reading of the following detailed description when taken in conjunction With the accompanying drawings.

FIG. I is a schematic side view of a system incorporating the cell-based image sensor of the present invention; FIG. 2 is a schematic of the light-sensing portion of the present invention; FIG. 3 (PRIOR ART) is a partial portion of the assembly of the contact image sensor semiconductor die array; and

FIG. 4 is a partial portion of the semiconductor die arrangement of the present invention.

t Referring now to FIG. 1, therein is shown an image sensing machine 10 which would be like a scanner, facsimile (fax) machine, copier, etc. The image sensing machine 10 is set up to sense an image 14 on tile top surface of a document 12 which is intended to move longitudinally in the directions shown by the arrow 16 so that "one" scan line across the width of the document 12 may be sensed at a time.

The illurnination to the top of the document 12 is providcd by a scrics of differclit colored I light-cmitting diodes (LEDs). The LEDs' 18, 20, 22, and 24 could be red, blue, green, and yellow, not necessarily in that order, to provide the ability to sense different color images 14. The LEDs in figure I are shown in the plane of the page. The LED's may also be in a single line perpendicular to the plane of figure 1. The illuminator could also be a end feed illuminator.

Light travels along the path 26 from-one-of the light-emitting diodes 18 during one scan. The light passes along the line 28 through a molded plastic lens array 30 to the photosensitive semiconductor die 32, which is embedded in a molded lead frame 34.

For a full color image, three scans are used, commonly referred to as red, green, and blue scans. The spectral distribution of each scan can effect the scanner's ability to reproduce a true color. More than one color LED may be turned on per scan to create a better spectral distribution.

The photosensitive semiconductor dice 32 are connected to circuitry 36 for taking the signals therefrom and converting them into signals which can be used by the particular scanner, fax machine, or copier, etc, The cable 38 provides power to the image sensing machine 10.

Referring now to FIG. 2, therein is shown the width view of an image sensor 40 of the image sensing machine 10 (which is shown lengthwise in FIG. 1). The image sensor 40 consists of the molded lens array 30, which consists of individual lenses 30A, 3013, 30C, 30D, and 30E as examples. The lenses are shown schematically as single element lenses, but most commonly would be multi element. As shown in FIG. 2, each of the lenses covers a cellular field of view designated respectively for lens& 30A, 3013, 30C, 30130 and 30E as cellular fields of view 42A, 4213, 42C, 42D, and 42E. As evident, the fields of view overlap as overlap 44A, 4413, 44C, and 44D. The lens array 30 respectively focus tile cellular fields of view 42A, 4213, 42C, 42D, and 42E onto respective photosensitive semiconductor dice 32A, 3213, 32C, 32D, and 32E. Each lens and related di(form a cell of the image sensor 40.

6 The molded lead frame 34 has registration features (not shown) that allow accurate positioning of the optics 30 to the molded lead frame. The registration features allow the molded lead frame to be accurately positioned in assernbly equipment and the registration features allow the final scanner assembly to be precisely place by the final manufacture in the overall scanner assembly. The precision registration allows the semiconductor die 32 to be placed precisely relative to the molded lead frarne and therefore to the optics 30.

Referring now to FIG. 3 (PRIOR ART), therein are shown prior art photosensitive semiconductor dice 50A and SOB. Manufactured into the surface of the semiconductor die 50A arc a series of pixels 52, 54, 56, 58, and 60. Similarly, the semiconductor die SOB has pixels 62, 64, 66, 68, and 70. The pitches 53 and 57 between pixels must be maintained the same and also must be maintained between the pixels at the ends of the semiconductor dice, such as at the pitch 61. There cannot be an accumulation of tolerance across all ten or more semiconductor dice 50. This means that the semiconductor dice 50A and 50B must be sawed to very tight tolerances such that the gap designated as gap 72 must be sawn within a tolerance of typically 0.0002 inch.

The pixels of the prior art have poor "fill factor". This means that they have to be reduced from the maximum possible size to keep the spacing between all the pixels tile same, including the pixels at the ends of the dice 60 and 62. Assembling the dice to maintain constant pixel pitch across the gap between dice is difficult. It is too difficult to do with standard automatic die-attach machines. The precision die attach, can be done with very expensive custom machines, but is usually done manually.

The sawing process for the semiconductor dice normally produces chipped and cracked edges. The dice for conventional contact image sensors can not tolerate the normal amount of chipping and cracking, and the sawing process must be maintained at a fine level.

The assembly process can also chip and crack die. The die can be chipped and cracked by banging them into each other. This requirement makes manual or automatic die attach operation more difficult than normal, and the failures are the reason for high yield loss.

Referring now to FIG. 4, therein are shown semiconductor dice 32A and 32B. It should be noted that.the new semiconductor dice are approximately 1/3, ,or less, than the size of the semiconductor dice 50A and 50B of the prior art. Because the pixels have a 100% fill factor, they are adjacent to one another, and there is no problem with pixel gap placement. For example, the pixels of the semiconductor die 32A, which are the pixels 82, 84, 86, 88, and 90 can be positioned with conventional semiconductor manufacturing processes. Since

7 the exact position in which the image 14 falls on the semiconductor die 32A is not critical and will be corrected by the image processing. The overlap 44A and image processing means that the distance between the dice 104 is not critical. The distance from the sawn edge of the die and the pixels 102 is also not critical.

In operation, the document 12 with the image 14 passes by the image sensing machine 10. A narrow band of the image 14 is illuminated sequentially by the light-emitting diodes 18, 20, 22, and 24, and the light follows the path indicated by the arrow 28 to the lens array 30. If the lens array is designed to be achromatic, the lens array can focus each of the illumination colors from LED's 18, 20, 22 and 24, onto the plane of the semiconductor die 32.

As shown in FIG. 2, the length of the image 14 is reflected into the lenses 30A, 30B, 30C, 30D, and 30E with each of the portions of the image 42A, 42B, 42C, 42D, and 42E being focused typically with between 3:1 to 6:1 de-m agni fi cation through the lens array 30 with the overlap indicated by overlapping areas 44A, 44B, 44C, and 44D.

Each of the portions of the image 14 is focused onto a photosensitive pixel on a semiconductor die 32. In the region of overlap 44, the object image 14 may be imaged on both semiconductor die 32. It should be noted that the precise alignment is not critical since image processing will match and remove the overlap to recreate the original image 14.

It has been determined that the entire image-sensing machine 10 can be made much more inexpensively while providing higher quality than in the prior art.

For example, the price of the molded lens array 30 is a factor of more than 6 times lower than a glass SELFOC array and a factor of more than 2 times less expensive than a plastic SELFOC array.

Als o, molded lead frames are a factor of 2 times, or more, less expensive than printed circuit boards, which were required with the prior art. Molded lead frames also simplify assembly to other components. In addition, due to the de-magnifi cation of between 3:1 and 6: 1, the semiconductor die can be reduced in size to 1/3 to 1/6 that of the prior art contact image sensors, resulting in a proportional reduction in cost. This arrangement would also eliminate most of the expense of the 30% assembly yield loss in prior art contact image sensors.

Further, the semiconductor die aligranent is no longer critical and would speed assembly. At the same time, the quality of the image would be better since the lens array can work at a higher resolution than the SELFOC array in contact image sensors. Better I 8 resolution means that the image sensor 40 could operate at 1,200 dots per inch or more, which is much more than the prior art SELFOC system, which is limited to approximately 600 dots per inch.

SELFOC based image sensors are limited to approximately I mm depth of field when used in 600 pixel per inch scariners. The new device 40 can be designed to have a much higher depth of field using conventional optical design techniques. This means that the distance between the document 12 and the lens array 30 is not so sensitive and could have a greater range of variation.

While the high f number requires more light than a similar SELFOC, the improvement of fill factor more efficiently collects light. Because contact image sensors lose fill factor quickly as the resolution go over 600 pixels per inch, the advantage of full fill factor may give the new sensor superior light collection at higher pixel per inch resolutions. This is a dramatic effect as resolution increases.

The semiconductor dice 32 may utilize CCD, or photo diode technologies. The dice could be made in CMOS, bipolar or amorphous silicon.

While the invention has been described in conjunction with a specific best mode, it is to be understood that many alternatives, modifications, and variations will be apparent to those skilled in the art in light of the aforegoing description. Accordingly, it is intended to embrace all such alternatives, modifications, and variations which fall within the spirit and scope of the included claims. All matters set forth herein or shown in the accompanying drawings are to be interpreted in an illustrative and non-limiting sense.

I 9

Claims

Claims:

1. A sensor system for images comprising:

first photo-sensitive element responsive to light to provide a first signal; second photo-sensitive element responsive to light to provide a second signal; first lens focusing light from a first portion of the image on to said first photo-sensitive element; second lens focusing light from a second portion of the image on to said second photo-sensitive element, the second portion of the image overlapping a portion of the first portion of the image; combining means for combining said first and second signals from said first and second photo-sensitive elements and outputting a signal representative of the image.

2. The sensor system as claimed in claim I wherein: said first and second photo-sensitive elements are spaced apart.

3 The sensor system as claimed in claim 1 or 2 wherein; said first and second photo-sensitive elements contain a plurality of pixel elements; and the light from the first and second portions of the image focus on less than the plurality of pixel elements.

4 The sensor system as claimed in any preceding claim, wherein: said first and second lens elements are a single unit.

5. The sensor system as claimed in any preceding claim, wherein: said first and second photo-sensitive elements are on separate carriers.

6. The sensor system as claimed in any of claims I to 4 wherein-, said first and second photo-sensitive elements are on a single carrier.

7. The sensor system as claimed in any preceding claim, wherein: said first and second lens elements invert said first and second portions of the image; and said combining means includes means for re-Inverting the image.

8. The sensor system as claimed in any preceding claim, including: a source of light proximate said first and second lens elements for illuminating the image,

9. A sensor system for images comprising:

first photosensitive semiconductor die responsive to light to provide a first signal-, second photosensitive semiconductor die responsive to light to provide a second signal; first lens focusing light from a first portion of the image on to said first photosensitive semiconductor die; a second lens focusing light from a second portion of the image on to said second photosensitive semiconductor die, the second portion of the image overlapping a portion of the first portion of the image-,

10. The sensor system as claimed ul claim 9 wherein: said first and second photosensitive semiconductor dice are spaced apart,

11. The sensor system as claimed in claim 9 or 10 wherein; said first and second photosensitive semiconductor dice contain a plurality of pixel elements; and the light from the first and second portions of the image focus on less than said plurality of pixel elements.

M 11

12. The sensor system as claimed in any of claims 9 to I I wherein: said first and second lenses are a single unit moulded of plastic.

13. The sensor system as claimed in any of claims 9 to 12 wherein: said first and second photosensitive semiconductor dice are on separate die carriers.

14. The sensor system as claimed in any of claims 9 to 12 wherein; said first and second photosensitive semiconductor dice are on a moulded single plastic carrier capable of precisely holding said first and second photosensitive semiconductor die with respect to said first and second lenses.

15. The sensor system as claimed in any of claims 9 to 14 wherein: said first and second lens invert said first and second portions of the image.

16. The sensor system as claimed in any of claims 9 to 15 including:

a plurality of light-emitting diodes proximate said first and second lenses for illuminating - the image.

17. A method for sensing images comprising the steps of focusing light ftom a first portion of the image on to a first photo-sensitive element; focusing light from a second portion of the image on to a second photosensitive element, the second portion of the image overlapping a portion of the first portion of the image; combining said first and second signals from said first and second photo- sensitive elements.

18. ne method as claimed in claim 17 wherein the steps of, 12 focusing light from a first portion of the image on to said first photosensitive element focuses light on less than the complete first photosensitive element; focusing light from a second portion of the image on to said second photo- sensitive element focuses light on less than the complete second photosensitive element.

19. The method as claimed in claim 17 or 18 wherein the steps of focusing lights through said first and second lens elements invert said first and second portions of the image; combining said first and second signals includes inverting said first and second signals to re-invert the image and recombine overlapping regions

20. The method as claimed in any of claims 17 to 19 including the step of, illuminating the image.

21. A sensor system for images substantially as herein before described with reference to and as 'illustrated in Figures 1,2 and 4 of the accompanying drawings.

22. A method for sensing images substantially as herein before described With reference to and illustrated in Figures 1,2 and 4 of the accompanying drawings.