Re: How does a Graphics Tablet work?

Hi Terry!,

I admire your desire to build your own graphics tablet. Like you, I try to build my own equipment whenever possible, but I'm afraid this is one project I wouldn't attempt without some very sophisticated and very expensive equipment.

Let's look first at how a mouse operates. Please bear with me if this sounds a little basic, but I think it's important to understand the mouse to really appreciate a tablet. The mouse operates on a 'relative position' basis, based on x/y coordinates. It needs to know 'from whence thou cometh' and 'to whither thou goest' to determine which pixels to turn on.Every movement of the mouse (or trackball) turns an interrupter - a thin disk with about 20 evenly-spaced aperture slots cut into it. This disk interrupts the light beam from a LED to a photo-transistor, creating a pulse every time the light beam is broken. There is one LED/disk/photo- transistor combination for the x (vertical) axis, and another for the y (horizontal) axis, mounted at a 90 degree angle to each other. Each pulse will move the cursor one 'UNIT' in the appropriate direction. I use the term 'unit' because it could represent one pixel, or one point (1/72nd inch), one inch, or any other unit of measurement; as defined by the software.

For our purposes, let's use the pixel. Assuming your CRT is set up for 600x800 resolution, you can see that it would take quite a few turns of the disk, and a lot of movement of the mouse, to draw a straight, diagonal line from the bottom left to the top right corner of the screen. We could make the software interpret one pulse as 10 or 100 pixels, resulting in less movement of the mouse to cover the same distance on the screen. But what happens to our line? Sure. It becomes jagged because we are moving 10 (or 100) up, then 10 (or 100) right, and so on, until we reach the top right corner. Great! - if all you're doing is drawing boxes or staircases; not so great for highly accurate drawings, or artwork. What you have done is sacrificed resolution for speed.

Enter the graphics tablet! Unlike the mouse, it uses 'absolute positioning'. To draw the same line, you would use the 'line' drawing tool provided by the software program, position the stylus at x,y=(0,0), click the stylus, move it to x,y=(600,800), click again, and Presto! You have your diagonal line. By adding a radii factor between any 2 points on the line, you can add curves, etc. How does the line look? Pretty good, when you consider that newer tablets have a resolution of around 2500 lines/inch. They accomplish this by using - yes, you guessed it, WIRES! If you look at one of the newer motherboards, the etched 'wires' are extremely close together; around 40-50 lines/inch. When I make a board, the best I can do is about 15. Obviously, some other technique must be used to get 2500 lines/inch.

Two methods are in current use. One, the ion diffusion process, 'sputters' copper ions, forming lines on the substrate under high vacuum and high heat. The other is to etch the lines on a copper-clad substrate by burning away the copper to form lines, with a carbon dioxide laser. Either way, you end up with an x/y grid. The position of the stylus is determined by using electromagnetic inductance.

In one method, the stylus contains a battery powered rf oscillator. When the grid is scanned, it detects the presence of the rf signal and translates it to x,y coordinates thru a timimg function, i.e. if a signal is detected at time=n, the stylus must be at (x=n,y=n). The second method, while using the same timing function to determine position, has an added advantage - it doesn't need batteries! With this method, the stylus contains a simple (well, maybe not that simple) R/L/C circuit. The grid first becomes a transmitting antenna, and charges the R/L/C circuit in the stylus, which starts to oscillate. The grid then switches to a receiving antenna, and detects the rf oscillations. Pretty slick, huh? Future tablets may use lines of single copper (or gold, or...) molecules laid end to end, and separated by single molecule lines of a non-conductive material. Can you imagine the resolution? How wide is a copper molecule? Another method may be to grow photo-transistor arrays right on the substrate, as they are now doing with CPU's and other LSI circuits.

If I were to try building a tablet, my first choice would be one which used opto-electronic components such as the Texas Instruments TSL218 or the TSL401. These are both linear light sensor arrays; the 218 being 512 pixels covering 2 1/2 inches, the 401 is 128 pixels covering 1/2 inch. That translates to 205 and 256 pixels per inch, which is far more than a crt, which has a resolution of 72 pixels per inch. I have experimented with both devices with excellent results. However, you need a very accurately focused light source (diode laser) to take advantage of this high resolution. Even then, it doesn't come close to the wire grid array. If you would like to check into other devices, I would suggest a CCD (charge-coupled device). You can find information on them at http://www.ti.com Click on "Products", then "Analog and Mixed Signal", then go to "Video and Imaging". Whichever method you choose to use, I believe the stability and accuracy of the timing circuit will determine your success. Make sure you choose quality components such as crystal oscillator clocks, polypropylene capacitors (1% tol.), 1% resistors, proper rf signal shielding, signal pin guarding, power supply filtering, etc.

If I can be of any further help, you can contact me directly at Karl- Kolbus@email.msn.com I live in Wauwatosa, Wisconsin - a suburb of Milwaukee.

Good Luck!
Regards,
Your not-so-mad scientist, Karl