Originally Posted by damnsam77
Jon, sorry if this sounds dumb. But what exactly am I looking for when I am testing for convergence in all of the patterns you listed? I remember one thing with te red/green/blue crossline patterns that the blue was the thickest line. While the green and red were thinner.
Unlike a direct view fixed pixel device (eg. a plasma screen), the RS40, and the last several generations of JVC projectors are 3 chip LCOS machines, and produce colour images by overlapping varying quantities of red, green and blue. However, in order to effectively do so, the individual R, G & B chips/panels need to overlap very well, across the entire screen.
When you look at a single pixel thick white grid, you are looking at the results of each individual panel generating its respective red, green, and blue grid which overlap on your screen to generate a white grid. If convergence is 'off' for one or more of the panels, you'll see this as 'fringing' on the single pixel white line. Practically speaking, its almost unheard of to have perfect alignment of the 3 panels across the entire screen, and you will almost certainly notice the horizontal and vertical white lines in the grid having an edge that is redder, or greener, or bluer than the line itself. The more 'out of whack' this 'fringing' is, the greater your convergence error.
Because of the projector's ability to shift each panel's output horizontally and vertically in single pixel increments, you shouldn't really be off by more than 0.5 pixel - at least in the centre of the projected image. The real problems arise when this divergence of up to 0.5 pixel at the centre, increases to more than that at the periphery.
Blue is tougher to see than red & green , and I wouldn't worry about blue appearing thicker (less focused?) than the other two colours. The real issue is how well all three overlap, and to what extent that changes as you move away from screen centre. If you want to see what really bad convergence looks like, feel free to shift the output of any one of the panels (red is easiest to see) within the service menu converge adjustment function, and you'll get a sense for what really bad (more than a pixel) misconvergence looks like.