Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!rice!uw-beaver!uw-entropy!dataio!aez From: aez@Data-IO.COM (Adam Zilinskas) Newsgroups: sci.electronics Subject: Re: input device Message-ID: <2179@dataio.Data-IO.COM> Date: 23 Oct 89 21:17:08 GMT References: <2300@tutor.tut.fi> Reply-To: aez@dataio.Data-IO.COM () Organization: Data I/O Corporation; Redmond, WA Lines: 58 In article <2300@tutor.tut.fi> jr67960@tut.fi (jr67960@tut.funet uunet!mcvax!tut!jr67960) writes: > > ... ... "better than paper and pencil" ... > >-j rantala There are several "touch screen systems" I know of, most are not too useful except for selecting items off menus. 1. Brute force method: I saw a piece of test instrument from Fluke that had a mechanical contact touch screen. It consisted of two closely spaced mylar sheets with a very thin gold plating forming a cross-point switch mechanism. The gold was thin enough to be relatively transparent but would cause current flow if the mylar sheets were squezed with a finger. The problems with it was that the switch contact areas were fixed locations, the gold could tarnish/damage causing "bounces", and since physical contact was required, the screen would get mucked up easily. 2. photo-beam grid. I think HP developed this but I could be wrong. Here, a grid of infra-red beams criss-cross the screeen face. Placing anything near the screen would disrupt two or more beams and you could tell approximately where to "touch" was. Best place I seen it used was in EPCOT center in Florida where the "electronic maps" had touch-menus, you needed to only go within an 1/2" to the screen to activate but people still always tried to break thier fingers getting the maps to work faster (the Bernoulli affect, harder the key is pressed, the faster the data is squirted out of the machine :-) The problems here is again the physical limits of the beam sizes make the active areas fairly large and fixed. 3. doing it with mirrors. I never saw a real machine with this just specs. Here instead of a grid of beams, a small rotating beam and sensor sits in one corner of the screen and the opposite edges are mirrors. The beam sweeps across the screen ala RADAR station and the sensor catches the return from the mirrors. Placing a finger in the screen will disrupt the beam twice, once during the direct beam at the finger, and once during the carom shot off the mirrors at the edge of the screen. The finger location is then triangluated. The problem here is first, a mechanical element is constantly moving and the computations to triangulate are typically more complex than the application. I have seen talk about systems where a headset calculates the head position to move a cursor, fancy optics "watch" eye movement and follow it on the screen, all sound cumbersome, expensive and useable only in situations like fighter jets where the user is so strapped in, little adjustments are needed. There was a thread of discussion about non-keyboard interfaces and I think the results were not conclusive, some could type 90 WPM and others scribble at that rate. Personally, I think keyboards (QWERTY, Chord or otherwise) is best for textual input (non ambigouous) while mouse/digitizer pads are best for graphical input. A better approach is to design a keyboard so that reaching for the trackball/mouse is easy so that the transistion between between keyboard (figners at home position) and mousing (move-pointer with fine eye-hand coordination) is quick. A good location may be below the space bar in the QWERTY keyboard for thunb operated trackballs. Adam Zilinskas If it does have anything to do with PLDs it isn't Data I/O's concern