170 likes | 357 Views
Low Cost Infrared Touch Screen Bezel for POS Systems. Rohan Verma, Jeremy Taylor, Freddie Dunn III. Georgia Institute of Technology School of Electrical and Computer Engineering. December 4, 2009. Touch Screen Design. Phototransistor. LEDs. ADC. Bezel. USB. Microcontroller. PC.
E N D
Low Cost Infrared Touch Screen Bezel for POS Systems Rohan Verma, Jeremy Taylor, Freddie Dunn III Georgia Institute of Technology School of Electrical and Computer Engineering December 4, 2009
Touch Screen Design Phototransistor LEDs ADC Bezel USB Microcontroller PC
Project Overview • Infrared LED bezel that attaches to a computer monitor converting it into a touch screen • Lower LED/Phototransistor count when compared to the current market products to make for a more cost-efficient and affordable device • Development costs and a production run of 5000 units totals $458,000 with a single unit cost of approximately $91.60
Bezel Design • The current orientation of the • LED’s/Phototransistors produced • the best results for touch • recognition • By placing an acute angle LED • (~20 degrees) we are able to • more accurately detect touches • that occur in the middle of the • screen Red Line – Ground rail for device Green Line – Power rail for device Blue Circle – Phototransistors Yellow Circle – Wide Angle IR LED Orange Circle – Acute Angle IR LED
MCU firmware nuances • 10-bit Analog to digital conversion, scaled down to 8-bit, of the phototransistors converts their voltage into a digital reading on a scale from 0 to 255 • By polling the sensors with the LED’s on and then off, and computing the difference of the two, we are able to eliminate ambient IR levels • Take a pseudo running average of current light levels, valuing the most recent reading as only 1/10 of the current reading to minimize random fluctuations of readings (Must fill pipeline) • Take 200 A2D readings at boot to fill the averaging pipe and determine the unobstructed IR levels • Touch is determined by creating a touch threshold of 90% and 110% of the unobstructed levels and comparing to the current readings
Testing Procedures • Implemented a serial debug interface for viewing raw ADC values from the phototransistors and later to detect touches • MCUs UART (9600 baud), Viewed over HyperTerminal • Methodology • By choosing a series of random points (~20 points) we were able to test various spots in our design for “blind spots” and by altering the orientation of the LEDs and phototransistors, we were able to achieve a much better touch detection sensitivity and accuracy • USB interface was written separately and tested by having the mouse move to the center of the screen and continually right-click (via code)
Did it meet requirements? • Proposed: A device that can accurately identify a set of distinct points on the screen and is able to move the mouse to that spot • Result: Achieved through the testing of a series of points that clearly are defined and allows the mouse to be placed at 9 unique locations • Proposed: The ability to achieve single-click, double click, and drag • Result: Achieved through the USB interface which allows, through code, that the mouse will “click down” as long as a touch is detected and will be “released up” once a touch is no longer detected. • This allows for drag to be implemented as long as the “mouse down” action is in effect
Problems/Issues • Encountered many “dark spots” with original configuration • Resolution: Improved upon by altering orientation • Too sensitive when objects are in close proximity (~3 inches outside the bezel) • Getting the optimum number of phototransistors
Recommended Future Work • Continue to increase the efficiency/accuracy of the touch screen bezel by altering/adding the LEDs and Phototransistors. • Find a functional relationship between light levels and finger placement to allow for interpolation and determination of touch location • Find new LEDs/phototransistors that increase sensitivity for more touch points