WO1998019283A1 - Touch screen based upon topological mapping - Google Patents

Abstract

A new touch screen based upon the mapping of coordinates from an equipotential (32) space defined by a simple set of screen electrodes to some other more useful coordinates, such as Cartesian. The key idea is that unique coordinate mapping can be achieved with each sensing pair of electronic readings.

Description

TITLE

Touch Screen Based Upon Topological Mapping INTRODUCTION

Since their introduction in the early 1970s, touch screens have afforded attractive alternatives to keyboards for certain computer applications. In many situations the keyboard and mouse are eliminated, because the touch screen provides the user with a much easier access to the computer. As a consequence, the market has grown to a substantial size, and a continued rapid growth is anticipated. However, current touch screens are difficult to produce, which creates a price barrier limiting growth into many new areas, such as education.

In this disclosure, a new concept is discussed that virtually eliminates design constraints and provides more freedom for the configuration of touch screens. Examples are given to illustrate this new freedom in design parameters. These design concepts provide a basis for producing touch screens at a much lower cost, without sacrificing quality. Furthermore, the creation of new designs for special sensor size, shape, or electrical characteristics is greatly simplified with the concept described herein and reduces research and development costs .

BACKGROUND OF THE INVENTION

A substantial portion of touch screens produced today are based on the measurement of electrical potentials on substrates that are made of a transparent medium such as glass, coated with an electrically conductive material. Uniform electrical fields must be maintained on the substrate, and these are applied sequentially in the x- and y-directions . In other words, equally spaced equipotential lines are generated orthogonally in a timed sequence. A voltage (or equivalently, a current related to the local potential of the touch point) measured when the field is in the x-direction is directly proportional to the distance along the x coordinate and is independent of the y coordinate . Conversely, a voltage measured when the field is in the y-direction is directly proportional to the distance along the y coordinate and is independent of the value of x.

According to present designs, resistive touch screens are often mounted on LCD or CRT displays, but perhaps most commonly on CRTs used as computer monitors to use as data input devices . As shown in Figure 6 a typical monitor 10 will comprise a back case 11 into which is set the CRT. A glass panel 12 with a uniform resistive coating 15 (shown in Figure

7) such as ITO (indium tin oxide) is placed over the face 14 CRT 13. A polyester coversheet is tightly suspended over the top of the glass panel, preferably separated from it by small transparent insulating dots 16 as described in Hurst, U.S. Patent No. 3,911,215 which is incorporated herein by reference. The coversheet 17 has a conductive coating on the inside and a hard durable coating 18 on the outer side. A more detailed view of the layers of the touch screen is shown in Figure 7. A bezel 19 is then placed over the coversheet.

A simple computer or controller 20 (shown in Figure

8) is used to alternate a voltage across the resistive surface of the glass in the X and Y directions. When a touch on the coversheet causes the inner conductive coating to make electrical contact with the resistive coating on the glass, an electrical circuit connected to the controller digitizes these voltages and transmits them to the associated main computer 21 for processing. Alternatively, the electrical circuit may digitize the amperage. As shown in Figures 8A and 8B, the controller 20 may be mounted internal to the monitor 10 or in a slot within the associated main computer 21.

In practice, the implementation of these concepts, as disclosed in the Patent of Hurst (U.S. Patent 3,798,370, March, 1974) leads to the production of touch screens of excellent quality. Yet production costs are high, because of three factors :

1) The substrate must have very uniform conductivity. Conductive materials are applied to a substrate (usually glass) in elaborate vacuum chambers. When a large substrate is being prepared, the chamber must be still larger, and even then, several sources must be used to cover the substrate uniformly. Some of these coated substrates do not meet specifications and have to be rejected.

2) A resistor divider network must be added to maintain straight equipotentials by eliminating edge effects associated with the field switching matrix. This has independent quality demands that further add to production costs and increase rejection rates.

3) Finally, rigorous post-fabrication testing must be done on each completed screen. These statistical quality- control tests are expensive and are directly associated with the problem of maintaining accurate equipotentials.

Currently, design changes would require retooling in all of the factors mentioned above. However, these may be virtually eliminated in the new concept of the present invention. This concept is explained so that it will reduce all three of these cost factors and, at the same time, provide much more flexibility in the design of sensors of the required shape, size, and electrical specifications.

SUMMARY OF THE INVENTION

It is therefore a purpose of the invention to provide an improved touch screen allowing improved screen yield with inherent tolerance for individual and lot variances. It is a further object of the invention to permit simplified manufacture requirements for touch screens including less-demanding conductive-coating application; fewer electrodes -- only four, for example; and no divider resistors required. It is yet another purpose of the invention to provide compatibility with current manufacture with identical analog-to-digital electronics, and similar calibration/testing procedure. It is yet another object of the invention to permit manufacture at low additional cost, more than offset by savings in screen manufacture. It is another purpose of the invention to permit liberated design of touch screens with changes readily implemented to accommodate new screen configurations .

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1(a) shows a contour plot of theoretical equipotentials in the x-direction on a touch screen sensor according to the present invention with four electrodes and non- linear equipotential lines;

Figure 1(b) shows a contour plot of theoretical equipotentials in the y-direction on a touch screen sensor according to the present invention with four electrodes and non- linear equipotential lines; Figure 2 shows a three-dimensional plot of the potential distribution in the sensor with the configuration shown in Figures 1(a) and (b) ;

Figure 3 shows a plot of current flow lines in a sensor having four electrodes at the corners.

Figure 4 illustrates a contour plot of theoretical equipotentials in a sensor with non-uniform conductivity to simulate conditions that might result from vacuum evaporation and deposits of the conductive substance.

Figure 5 illustrates the special condition when the severe bulging of an equipotential causes a given equipotential to intersect a rectangular function box at four points;

Figure 5A illustrates the location of cells by the process of border mapping in the usual case.

Figure 6 illustrates a typical monitor with touch screen input device;

Figure 7 illustrates a detailed view of the layers of a resistive touch screen;

Figure 8A shows a representative touch screen monitor with an internal controller;

Figure 8B shows a representative touch screen monitor with an external controller; and

Figure 9 illustrates a contour plot of theoretical equipotentials in a rectangular sensor with an electrode in the center of each side.

DETAILED DESCRIPTION OF THE INVENTION

For the purpose of describing the invention, it is best to define a space in terms of electric equipotentials. In principle, coordinate mapping can be obtained using any set of electrodes that generates monotonic equipotentials. For a surface, lines drawn at the same potential in the space between the opposing sets of electrodes are called equipotential lines. Full two-dimensional mapping on that surface can be achieved using two sets on monotonic equipotentials in two different directions. The potential lines need not be straight or uniform, but the key idea is that any point on that surface must have a unique value for the pair of potentials at that point.

Consider, as diagramed in Figures la and lb, a two- dimensional surface 30 that is slightly conductive (or, if you prefer, resistive) . A very simple geometry with electrodes 31 attached at each of the four corners can be used to illustrate the basic idea. The exact solution for the potential distributions also provided by LaPlace's equation as described below. The usual Cartesian space with x-y coordinates can be mapped by two sets of equipotential lines. One of these sets is obtained when a source of electrical potential such as a battery is applied so as to produce an electrical field generally in the x-direction; equipotentials 32 will run generally in the y-direction (Figure la) . When the battery is switched to produce a field generally in the y-direction, equipotentials will run generally in the x-direction (Figure lb) . The word "generally" is used to stress that nowhere are we assuming uniform fields or equipotentials that run parallel to the x- or y-axes. There is distortion (i.e., the equipotentials are not evenly spaced nor are they parallel to the x-and y-axes) , since the electrodes are not at all designed to produce uniform fields, and because the electrical conductivity need not be uniform. Obviously, one set of equipotentials need not be orthogonal to a second set when the voltage source is switched from one direction to another.

The notation V(X,y) is used to mean an equipotential when the voltage supply is connected in the x-direction (Figure la) , and the lower-case y is shown to indicate that the equipotential also depends on y, due to the distortion. Similarly, V(Y,x) is used for the equipotentials when the voltage supply is connected in the y-direction (Figure lb) . Near the center of the screen, there is little or no distortion and it is possible to approximate V(X,y) with V(X) , i.e., the equipotential in x is essentially independent of y. Of course, a similar statement could be made of the complementary case, V(Y,x). Near the edges of the sensor there is appreciable distortion, which is permitted here, but would be fatal to the performance of conventional touch screens that require uniform potential distributions.

Because of this interdependence of potential upon both x and y, a potential measurement cannot uniquely specify either x or y. Consequently, it is no longer possible to measure x independent of y, and y independent of x. This must be given up in the present considerations; however, the pair of potentials [V (X, y) , (Y, x) ] uniquely transforms to a point P(x,y) in the Cartesian space, under certain conditions. It is this uniqueness that is important to the invention. So along as the uniqueness holds, operations can be found that will map the pair [V(X,y) , V(Y,x) } to a unique point P(x,y) in Cartesian space.

Certain conditions are required for this uniqueness . One condition is that the field (i.e., potential gradient or change in potential) be continuous over the entire area in each of the directions of application. A related condition is that the field has no singularities over an appreciable area of the substrate. These field conditions imply that the equipotentials must increase continuously in the direction of the applied potential. These conditions, in turn, impose some moderate conditions on the substrate prepared by vacuum coating. The coating need not be uniform, but it must be continuous without isolated areas of no conductivity. Further, the coating must not be so heavy in the other areas so as to substantially "short" them. Both of these conditions are much more easily satisfied than is required for present touch screens. Finally, there is another condition related to how much variation in the field we can allow in practice. A weak-field region presents a problem for precisely determining the sensed point.

Many designs with alternative electrode placement can be conceived. However, the simplest design is that shown in Figures la and lb: attachment of one electrode 31 to each of the four corners of a rectangle. It will be understood that this panel could be planar or could be curved to fit onto a computer monitor. If the electrodes are located in the middle of the panel edges, the fields near the corners are very weak as shown in Figure 9, an undesirable condition. Additional electrodes properly placed can make the field well behaved but add complexity in implementation.

MAPPING OPTIONS

Some, but not all, applications of touch screens based on the potential pairs will require mapping to a Cartesian coordinate system. There are several possibilities to fit any number of applications. Direct address. In principle it is possible to communicate with an attached computer or cash register without ever leaving the potential-pair space. However, this would not appeal to users who have been brought up in a Cartesian world, and as "Cartesianites" would feel uncomfortable working with, for example, curved menu boxes. This would be the least demanding, since little technology would be required for implementation. This possibility is given mainly to set the stage for more serious mapping mechanisms.

Complete Mapping. This term for cases where it is elected to store each point in a built-in table (i.e., memory for each pixel required) . This option would use an auxiliary mapping device with enough memory to define the required number of pixels in the x-y plane. The device would store a large array of pre-defined x,y points corresponding to the field of points in the [V(X,y) , (Y,x) ] space, so that a "lookup" table (LUT) could be used in the mapping. In this case, the LUT would be a device or process whereby a potential pair, [V(X, y) , V(Y,x) ] , in suitable digital form would be used to locate in a two-dimensional table the corresponding real-space coordinate pair, (x,y) , also in digital form. Resolutions of 128x128 to 1024x1024 would require 32Kbytes to 4Mbytes of LUT memory, respectively. This option is becoming increasingly attractive as computer chips drop in price.

Partial mapping. In this case, an active area, such as a menu box, can be defined without complete, one-to-one mapping. For instance, straight-sided boxes (or other shapes) could be defined by their boundaries, stored as potential pairs . A simple logic could be employed to locate the potential pair, [V(X, y) , (Y,x) ] , within or outside the boxes. Typically a limited number of boxes are used in menu selection, so that the memory required would be greatly reduced from complete mapping. Perhaps a small LUT could be used that define areas where this boundary analysis is to be made .

A SPECIAL EXAMPLE OF END USE: MENU BOXES A common end-use of the touch screen is the so- called menu application. Choices are made by the user simply by touching menu items typically enclosed by rectangular boundaries .

Some options for the mapping of the measurement space [V(X,y) , V(Y,x) ] are :

1) Do the complete mapping by using a programmable chip with a "look-up" table, as described earlier. This would remove the distortion and provide a unique mapping onto the Cartesian space.

2) Do not map into the Cartesian space; accept menu boxes that are defined with curved lines as required by the distortion of the measurement-space compared to the Cartesian-space. While this is functional, it is not aesthetically appealing.

3) Do not map into the Cartesian space; but, give the appearance of a rectangular space. This can be done by creating an invisible band around the perimeter of the box, so that the area defined by the raster lines on the computer screen can appear rectangular. This could be accomplished with an electronic logic that prevents errors in the selected box, by removing any point in the pair- potential space that has a questionable box assignment. There would be no sacrifice of aesthetics, but, in the case of severe distortion, it would produce an "inert" band that would be unsatisfactory for some uses. 4) Map only around the perimeter (boundary) of the box. Define the points around the perimeter of the rectangular box, in terms of the pairs [V(X,y) , V(Y_f x) ] and use an electronic test to determine if the point in the measurement space is within the rectangular box.

Expanding on the discussion of option 4, Figures 5A and 5B illustrate the definition of a rectangular box 40 by means of the measured coordinates based on equipotential pairs. This illustration makes it clear that any pair of potentials measured within the box so defined can be assigned uniquely to the box defined by Cartesian coordinates. So, the main design choice is the definition of the perimeter of the box in practical terms.

It is beyond the scope of this patent application to discuss the many options on software methods for analyses such as this . But perhaps some general comments are needed to attach plausibility to the electronic test that determines if the measurement space is within the selection area. In the box illustration of Figure 5A, note that its entire perimeter could be defined with as few as eight measurements of the potential pairs at the indicated Cartesian points. From the three measurements on each edge, a polynomial of, for example, three terms (e.g., a+bz+cz²) could be fitted to both of the measured members for each of the potential pairs, giving two sets of coefficients in the polynomial. Thus, for the four sides only eight sets of these coefficients (a total of 24 coefficients) are needed to completely specify (with good accuracy) the perimeter of the box. Assuming that each coefficient is an eight-bit byte, only 192 bits, or 24 bytes, of storage space is required.

Having defined the box, where the measurement space is now made to conform to the rectangular Cartesian space, a test can be described to see if the measured point is within the Cartesian space. Imagine, as in Figure 5A, that either of the two potentials measured at a particular point, P, is on an equipotential 32 that intersects the box at two places. Neither of the two equipotentials 32 alone will select a particular Cartesian box, as shown. However, the pair of equipotentials 32 will intersect only at one point and therefore in a particular box. Thus, a boundary analysis, a procedure that searches all boxes and finds the four points on the perimeter of a box, will select that particular box. A box is selected by finding just two potentials on its perimeter, provided that the two potentials are complements; i.e., one value belongs to V(X,y) and one to V(Y,x) . With modern data processing, this search procedure is routine. For instance, the two potentials measured could first be stored in a register until the boundary analysis described above is completed.

In the case of sever bulging of an equipotential 32 shown in Figure 5B, it is possible that a given equipotential will intersect a box at four places rather than two. This could produce some ambiguity in assigning a pair of potentials to a box. However, this can be avoided by applying a special test. In the above paragraph, a regular test is described that will be by far more common than the special test . In the special test a concept of complementary potentials is employed. At every point in the working area, there is an intersection of two equipotentials. The discussion will focus on just one of these, and the other will be considered its complement. At each of the places where one equipotential 32 crosses the boundary of a box 40 there will be a value for the corresponding complement. The special test looks at the complement in potential actually measured and tests to see that its magnitude lies intermediate to the complements created at the boundary crossings. In the case where there are four crossings this test could be applied and in three regions of the sensor, as illustrated in Figure 5B .

The special test just described removes any possible ambiguity due to four crossings of one equipotential on the box boundary. However, a more attractive alternative could be simply to electronically abort any touch that results in four crossings. This would create very small inactive areas, which would go unnoticed in most applications . This test procedure is not restricted to rectangles. For instance, the control "box" could be a circle or an arbitrary enclosure. However, for complex figures, boundaries or perimeters, definition becomes correspondingly more complex.

And, of course, the special test can be avoided altogether by using the concept of "cells" that are small enough to eliminate multiple crossings. A menu box would then be configured from any number of cells . Reasonable sensor design to avoid excessive curvature of equipotential lines would greatly reduce the number of cells needed, perhaps to one per menu box. Interpolative mapping. In practice, intermediate points between two tabulated points can be determined by interpolation. This option would store calibration points and fill in all intermediate points by interpolation. In a sense, interpolative mapping can be regarded as a processor-based method of achieving complete mapping that uses less memory than a complete LUT. In this connection, a mathematical solution of the boundary-value problem would be very powerful, especially if the solution is able to adjust to substrate irregularities. For instance, a math chip or programmed processor could be used to solve a partial differential equation known as LaPlace's equation, and the interpolation between points could be based on this solution. It is anticipated that the solution to the partial differential equation would automatically take into account non-uniformity in the substrate, and thus serve as an accurate interpolation independent of substrate characteristics. An explanation of the application of LaPlace's equation follows:

The electrical potential distribution of a conducting sheet is determined by the configuration of electrons, the potentials applied to them, and the conductivity, δ, of the sheet. In general δ = δ (x,y) is a function of position. We further assume that the conductivity is isotropic (but not necessarily uniform) for conventual conductive coatings applied to screens . Then if V(x,y) is the electric potential at (x,y) , the resulting current j (x,y) is given by:

j (x,y) = -δ (x,y) W (x,y) . (Al)

We assume that charge cannot accumulate at any point, hence: vj (x,y) = 0 , (A2 )

and from subst itution of Eq . (Al ) into Eq . (A2 ) :

v [δ (x,y) W(x,y) ] = 0 (A3 )

This is the equation that is used to solve for V (x,y) , for a given distribution of applied electrical potential on the electrodes. The electrodes may be of any shape, including circular spots 31 as shown in the illustrations of Figure la and Figure lb. In that case, the sheet is 20 x 28 cm with 1-cτn radius circular electrodes centered on the four corners of sheet with uniform conductivity .

Figures 2 and 3 show alternative ways of displaying the configuration of Figure la: three-dimensional potential and current distributions, respectively. These are useful to further understand the distorted space of non-uniform potentials that occurs with the simplified electrode configuration .

Experiments approximating this model were done by making electrodes in the form of discs of 7-mm diameter (using highly conductive nickel paint) applied to electrically conductive paper. These intuitively designed experiments gave excellent results. Even for a line drawn less than 1 cm from the two electrodes (spaced about 20 cm apart) on the left of the figure, there was less than a factor of two variation in the electrical potentials as read with a digital voltmeter of high input impedance . At 2 cm from the electrodes this factor was reduced to 1.5 and at a distance of 4 cm, this factor was about 1.25. At the center of the 28-cm conductive paper this factor was about 1.0. Thus, while the equipotentials are far from vertical (along the y-axis) there is no region of the sensor that deviates from norm by more than a factor of two. There were no "insensitive" regions where a change in position would give essentially the same potentials.

Another simple design was tested utilizing a single electrode at the center of the four edges of a rectangle. However, as shown in Figure 9, actual potential measurements on this design revealed regions near each of the corners that were "insensitive" and therefore would not be a good touch screen design. This effect could be anticipated simply by looking at the equipotentials in the vicinity of the corners. In these regions the equipotentials spread out indicating that the electrical fields are weak in these regions in comparison with the central region.

The examples illustrate an important point : it is quite easy to arrive at sensor designs where every point on the planar surface can be characterized by a unique pair of potentials [V(X,y) , V(Y,x) ] and where the fields vary by less than a factor of two, so that all regions of the sensor can be regarded as responsive. Of course, these examples are far from exhaustive; the designs could be combined to have eight electrodes. The electrodes could be rectangles instead of circles, etc. There is great flexibility because we have liberated our design from the requirement of uniform electrical fields.

NON-UNIFORM CONDUCTIVITY

If the conducting material has a non-uniform thickness, the potential will be affected by a non-constant δ (x,y) in Eq. (A3). For example, if δ (x,y) = {l+0. 01 [ (x- 10) ²+ (y-14³) 2] } ^"*, then the conductivity at the center of the sheet is twice that on a circle of radius 10 cm and centered at the center of the sheet. Such a conductivity function might represent that generated in vacuum deposition of conducting material from a single source located over the center of the substrate . Figure 4 shows a contour plot of the equipotentials for the same configuration as in Figure 1A, but with the above spatially varying conductivity.

Even simpler would be a modest resolution LUT (say, 256 x 256 points) and a linear- interpola tion algorithm. For example, a pair of 12-bit (4096) potential readings could be trimmed to 8 bits (256) before interpretation by the LUT. The 4-bit remainders would then be used for linear interpolation between adjacent points in the LUT. The resultant answer would yield complete mapping at 12-bit (4096) resolution. The code required for this is very small, so that even a modest- sized LUT with a simple programmed processor would be used.

Each of these general mappings may be chosen, depending upon existing production capabilities and specific application. Complete mapping is often preferred due to its conceptual simplicity. For special applications, such as menu selection, partial mapping would be quite satisfactory. Interpolative mapping might be the most practical way to achieve mapping at the highest-possible resolution.

MANUFACTURE Coatings

Returning now to the problem of achieving uniform electrical coatings over large areas, the present invention provides some interesting design considerations. Take the typical case where a vacuum evaporation chamber is of limited size with interior dimensions that are not much larger than the substrates themselves. In this case, the corner regions of the substrates tend to receive a thinner coating than the central portions. This certainly would be the case if there were only a single source of the coating material located at some distance away from the center of the substrate.

The design of Figure la and Figure lb would give partial compensation of this problem of irregular conductivity. That is, the equipotentials could become straighter in the corners due to the higher resistivity in these regions! (Conversely, designs with electrodes in the middle of the sides would only exacerbate the problem.) With the new design concept, it is entirely possible that coatings from small evaporators (that are currently unusable) would even be preferable to those of more uniformity. Additionally, less stringent requirements upon coating uniformity could allow economical manufacture with simpler in-house equipment rather than using specialized out-sourcing. Hardware for Complete Mapping

A convenient scheme for complete mapping is to use a decoding integrated circuit to convert sense readings. Chips are already manufactured very economically that provide this function for 256 x 256 and higher screen resolutions. Such a chip can be combined with the existing electronic sensing circuit to convert a pair of readings, one corresponding to V(X,y) and the other to V(Y,x), to their corresponding Cartesian space coordinates . Specific examples for various screen resolutions follow. For a screen with a resolution of 256 x 256, the raw data in potential space will consist of two 8 -bit measurements. To convert these, a LUT memory component is needed which will accept two 8-bit addresses that point to two 8 -bit values that have been previously loaded during calibration. Chips are available as programmable read-only memory (PROM) or erasable-PROM (EPROM) .

The Am27C1024 is a 1-megabit (65,536 x 16-bit) CMOS EPROM that meets the requirements for 256 x 256 resolution. This component is readily available from its manufacturer, AMD, or from a distributor such as Hamilton Hallmark. Typical power consumption is only 125 milliwatts in active mode and only 100 microwatts in standby mode. Only 8 seconds are needed to program the component while look-ups can be performed in 55 nanoseconds.

For a screen with a resolution of 512 x 512, the raw data in potential space will consist of two 9-bit measurements one corresponding to V(X,y) and the other to V(Y,x) . To convert these into comparable Cartesian space, a LUT component is needed which will accept two 9-bit addresses that point to two 9-bit values that have been experimentally determined during calibration.

The Am27C4096 is a 4-megabit (262,144 x 16-bit) CMOS EPROM that meets the requirements for 512 x 512 resolution. This component is readily available from its manufacturer, AMD, or from a distributor such as Hamilton Hallmark. Typical power consumption is only 125 milliwatts in active mode and only 125 microwatts in standby mode. Only 32 seconds are needed to program the component while look-ups can be performed in 90 nanoseconds. Both of the components cited above can be purchased in either a package with a ceramic window permitting erasure via ultraviolet light (and thus reprogramming) or in a sealed package for one-time programming. The one-time programmable part has the advantage of being slightly cheaper but the reprogrammable part has the advantage of allowing recalibration after some period of customer use.

For a screen with a resolution of 1024 x 1024 or higher, multiple LUT memory components can be used or a single-package component can be custom designed specifically for this purpose. There is an initial non-recurring engineering expense involved for a custom component but its piece price would likely be less than two EPROMs . Therefore, this approach might prove more cost-effective if 100,000 or more parts are needed. Development of a custom VLSI chip may be justified whenever the quantity of parts needed is sufficient to amortize the one-time engineering effort.

The contents of the custom component may consist of either full LUT memory just like the EPROMs or a reduced number of memory locations and some associated calculation logic. The exact balance of these resources is dictated by the resolution desired and the area required for logic versus that required for memory. Calibration

The screen-response calibration can be determined either empirically, theoretically, or by a combination of both. A purely theoretical approach presupposes a model geometry and a particular screen-conductivity distribution such as detailed in the particular example used to describe LaPlace's equation and would ignore variances that occur in manufacture. A purely empirical approach would involve pressing the screen in a pattern of points to generate all the values that transform potentials into useful coordinates. This latter approach would automatically account for variances but may be too slow or labor-intensive to be cost-effective. The combination approach would determine the transformation data for a number of points and interpolate the rest based upon theory.

Calibration values could be determined on an individual basis for each screen or each screen-lot manufactured. Thus, the component would be personalized to correspond to the coating of a particular screen and many non- uniformities, distortions and manuf cturing defects would be compensated, producing much higher screen yields at significantly reduced cost.

An example of screen calibration compatible with economical production involves manually or robotically touching a grid of points on each screen and interpolating using a computer. The computer applies the data generated by touching the grid points to a theoretical analysis. The Cartesian set of transformation values is generated by the computer and "burned" into the LUT stored in the PROM or EPROM. The number of points is determined by the resolution desired and the amount and kind of manufacturing defects. The program may also indicate defects in a screen and possibly highlight the positions of a few additional points that could immediately be touched. A refined calibration set can then be generated. As a result, quality control is automatic while rejects are reduced. SUMMARY

The present invention liberates the design of sensors for touch screen applications. Furthermore, this versatility comes with great simplicity and with no sacrifice of quality. Several versions of the concept have been explored in which a space is defined by measurement of a potential pair on a surface with electric fields applied sequentially in two general directions . Acceptance of some distortion of this space, with respect to a perfect Cartesian space, is the key to simplicity and freedom of sensor design. This distortion poses no fundamental limitations, since the potential-pair space can be uniquely mapped onto a Cartesian space . Complete mapping would use an auxiliary computer of adequate storage for the number of desired pixels.

In some "menu" applications, complete mapping with auxiliary equipment is not required. Several sub-options are available when the auxiliary computer is not used. (1) The measured space can be matched directly to the control space by accepting some distortion in the edges of the menu box. (2) If distortion of menu box edges is not acceptable, electronic blank-out can be used to give the appearance of straight edges. (3) Direct matching of the distorted equipotential space to a rectangular box can be made in our technique of boundary analysis. When boundary mapping is made along the perimeter of a function box, an auxiliary computer is not necessary, since little storage space is needed for the definition of boundaries.

Thus, there are many options for the broad principle -- topological mapping of potential pairs to real space. Regardless of the option used, it is believed that the overall cost of touch screen production is considerably reduced without the loss of any quality, compared to existing terminology. At the same time, new designs can be implemented without excessive engineering efforts. The combination of design freedom and the much-reduced production costs should impact the industry in a positive way; especially since there are markets, such as education and home entertainment, that cannot be penetrated with the price structure of existing technology.

Numerous alterations of the structure herein described will suggest themselves to those skilled in the art . It will be understood that the details and arrangements of the parts that have been described and illustrated in order to explain the nature of the invention are not to be construed as any limitation of the invention. All such alterations which do not depart from the spirit of the invention are intended to be included within the scope of the appended claims.

Claims

CLAIMSWe claim:

1. A method of determining the location of a selected point on a sensor apparatus having an electrically conductive area and an associated set of electrodes through which electrical current may be applied to establish electric potential distribution on the conductive area, an electrical circuit that measures the potential of points on the conductive area when brought in proximity thereto, and a controller, comprising the steps of:

(a) introducing an electrical current to the electrically conductive area through a first group of at least one electrode selected from said associated set of electrodes, and thereby establishing a first electric potential distribution on the conductive area;

(b) bringing the electrical circuit in proximity with the selected point on the conductive area and thereby measuring a first potential reading of said point ;

(c) storing said first potential reading;

(d) introducing an electrical current to the electrically conductive area through a second group of at least one electrode selected from said associated set of electrodes and wherein not all of said first and second groups of electrodes are identical, and thereby establishing a second electric potential distribution on the conductive area; (e) measuring the second potential reading of the selected point with the electrical circuit;

(f) processing said first and second potential readings to determine the location of the selected point.

2. The method of claim 1 wherein the first and second electric potential distributions are isotonic .

3. The method of claim 1 wherein the first electric potential distribution defines a first set of potential lines on the conductive area and the second electric potential distribution defines a second set of potential lines on the conductive area, and said first and second sets of potential lines substantially intersect .

4. The method of claim 1 wherein said first and second readings are uniquely transformed into spatial coordinates for the selected point .

5. The method of claim 1 wherein said first and second readings are mapped into the Cartesian coordinates for the selected point utilizing a lookup table.

6. The method of claim 1 wherein the processing of the first and second readings is determined by the intersection of equipotential lines.

7. The method of claim 1 wherein the electric potential distribution is established by applying a predetermined voltage sequentially to said first and second groups of electrodes.

8. The method of claim 1 wherein the location of the selected point is determined by the intersection of equipotential lines.

9. The method of claim 1 wherein it is determined whether the intersection of equipotential lines identified by the first and second readings are within a defined box on the electrically conductive area.

10. The method of claim 9 wherein a box is defined on the electrically conductive area such that an equipotential line has no more than two boundary crossings with said box.

11. The method of claim 1 wherein an auxiliary lookup table is implemented in an integrated circuit to achieve higher resolution by interpolation during the processing of the first and second readings.

12. The method of claim 1 wherein an algorithm is implemented in software run on a controller to interpolate the first and second readings.

13. The method of claim 12 wherein the algorithm is determined by boundary conditions applicable to the electrically conductive area.

14. The method of claim 11 wherein the lookup table is filled with values determined by a calibration step in which readings are taken for a set of preselected points .

15. The method of claim 12 wherein the algorithm is determined by a calibration step in which readings are taken for a set of preselected points.

16. A position sensor comprising: (a) an electrically conductive area; (b) a set of electrodes attached to the electrically conductive area and connected to an electrical power source;

(c) an electric circuit that measures the potential of the selected point on the conductive area when brought in proximity thereto;

(d) a controller that sequentially switches electrical power from the power source to a first group of said electrodes thereby establishing a first non-linear electrical potential distribution on the conductive area, and then to a second group of said electrodes thereby establishing a second electrical potential distribution on the conductive area; and

(e) a controller that processes sets of potential measurements of said first and second electrical potential distributions from the electric circuit to determine the location of the selected point.

17. The sensor of claim 16 wherein the conductivity of the electrically conductive area is not uniform.

18. The sensor of claim 16 wherein the electrically conductive area is generally rectangular and an electrode is located near each corner of said area.

19. The sensor of claim 16 wherein at least one electrode is circular in shape.

20. The sensor of claim 16 wherein at least one electrode is L-shaped.

21. The sensor of claim 16 wherein at least one electrode comprises two or more linear segments and electric power may be applied to each segment .

22. The sensor of claim 16 wherein at least one electrode comprises two dots and electric power may be applied to each dot .