My first industrial job after graduating with a PhD in theoretical physics was that of an R&D engineer at RCA, which at the time was a major conglomerate in the US. They manufactured TV sets among many other products which used cathode ray tubes (CRTs) as display devices. The charter of my group was developing magnetic deflection coils for CRTs which could deflect electron beams inside the tube to their desired positions. These coils had been traditionally designed by veteran engineers using their experience and intuitive skill following empirical trial and error methods a time consuming and expensive process. Our goal was to develop a CAD method using sophisticated computer programs so that design could be done faster and more accurately, even by less experienced engineers.

The staff members at the Research Laboratory, consisting mostly of PhDs in physics and mathematics, had already developed a program which they hoped would serve the purpose. Our job in the R&D group was to confirm its accuracy and give the research guys feedback so that they could continue to improve the software. The CRTs used in colour TVs used not one but three electron beams, one for each of the three “primary” colours, red, green and blue; colours were generated by phosphor stripes corresponding to each color, deposited on the inside of the TV screen. The desired performance was convergence of all three beams at every point on the TV screen.

Advertisement

Our first task was to model an existing coil system to see how close its computer-predicted convergence characteristics came to its experimentally measured performance. While the computer model correctly predicted general trends such as geometric distortions, change in performance by incremental changes in design parameters, ideal location of the coils on the CRT, it never gave us a satisfactory number for “misconvergence error” – the separation between the red and blue beams, the most significant performance parameter. The largest error allowed in specifications was at the corners and typically of the order of 1-1.5 mm. Much to our........

© The Statesman