Originally Posted by Mr Bob
Can't tellya exact box size, but reducing the o'scan on the Elites causes geometric distortion ANY way you do it.
The boxes in the internal pattern are all supposed to be the same size. The bottommost set is always a lot taller than those above it, once you can fully see them, after reducing your o'scan. That's the first thing that has to be done, is make that bottom row of them shorter, same size as the others.
The first row of boxes, farthest to the left, is always wider than those to the right of it when doing the redux via sm only. That has to be taken care of.
When reducing hor size via the H Size in Coarse convergence mode, it only goes so far before it becomes a H Linearity register instead, inexplicably but always. It's better to do H Width via the stand-alone register, the one for HD.
The boxes become non-linear vertically very easily upon reducing the o'scan. You have to go back and forth among them, till they are all the same height, whatever height that turns out to be.
That much I am glad to give away here.
And the first 5 columns of boxes to the farthest left behave in a VERY curious manner, which I am going to keep to myself. Be glad to share off board about that, but it took me years to figure it out, so I'm going to "breast my cards" on that one, sorry...
Reminds me of a quote from the Dark Night, as the Joker says about killing the batman, "If you good at something don't do it for free". Yes your free advise has been valuable, and your experience is the "art". Sometimes I have to read your posts 3 or 4 times to get the concept.
I figured out the geometery would be different based on the new sizing so I focused on getting all the lines straight as possible and all the horizontal and vertical lines as evenly spaced as possible. I used the string method which helped me see the cause/effect but it was a lot of work setting them up.
Just luck more than anything, I avoided the global changes that could be made with H size / V size mainly because there was not vertical phase adjustment (equivilent to the static adjustments in the coarse convergence). Also, because the FULL and FULL HD mode required different adjustments. I'm not concern about fixing the other modes like Cinema, Zoom, etc...
After lots of trial and error, due to some geometery issues that I can't correct around the edges, I'm at about 5% overscan on everyside. I used the Veritical static adjustments to center the image, then I had to actually increase the H Size because I had a lot of red bleeding from the left side on FULL HD, and some blue on Right side. That was also ok because it was needed to make the circles perfected anyway.
Funny thing I learned is that 480p signal uses the FULL mode so I see the red bleeding on the right side with this mode. However I can't go into coarse covergence while feeding a 480p single, it does sync well. I will try to use static registers to shift both to avoid this problems. I think this relates to your comments about the tricks with the left columns. I found it very difficult to get covergence working around the top, the top/right and the left most edge.
So far I have easily spent 15 hrs on what is probably simple geometery corrections but then learning curve is high. That does not included about 6 hours for the shimming mod.
I have corrected my top overscan from 10-12% to 5% which is a big improvement (with the centering of the image). The image is starting to improve. I'm now realizing how my greyscale and color saturation is being negatively affected by the brightness being raised to +12 and contrast at +10. I have ordered a Eyeone Display LT and I plan to measure to see just how far off my greyscale is currently. I will then slowly attempt to adjust greyscale using the trimpots for cuts (marking my position) and SM for drive ( cross my fingers).