Klaus and I spent several hours on Friday getting his new monitor to work with his slightly aged graphics card. His new monitor is a Samsung 2048×1152 LCD/TFT.
The G550 specs claim that it supports this resolution through the VGA output. Unfortunately, somewhere along the line, the DDC data was being ignored. The display specs stated that it supported a pixel clock up to 157 MHz, but Xorg told us otherwise and would limit it to 140 MHz.
So the first thing to do was to get it to ignore the DDC data, as this was definitely wrong. The “NoDDC” option in xorg.conf does this:
Section "Device" ... Option "NoDDC" ... EndSection
After much testing of various modelines and configuration, we discovered the undocumented “ReducedBlanking” option that allows higher resolutions to be sent over VGA connections with a short blanking period between lines. Modern TFT displays support a reduced blanking interval, as they don’t have to allow time for an electron beam to move back to the other side of the screen.
This did make me wonder whether in the days of CRTs anyone had thought of alternately reversing the direction of each horizontal line supplied through a VGA cable. I think this would allow for a reduced blanking period, higher resolution or reduced pixel clock. I guess CRTs never got up to the resolutions where this was an issue.
The trick is to put this in the “Monitor” section of your xorg.conf:
Section "Monitor" ... Option "ReducedBlanking" ... EndSection
This isn’t mentioned in the xorg.conf man page. It should be!
Once we’d found that option and used Erich Boleyn’s CVT Timings Program to generate the required ModeLine with a reduced blanking time, the monitor and graphics card worked fine.
The next challenge was getting a second monitor to work on DVI output of the graphics card. Unfortunately we came across a bug that would have required us to rebuild the mga driver from source. After spending many hours getting the first monitor to work, this was too much for us to bear for one day.
There is a patch in the Xorg bug tracker that will hopefully make it’s way into a release of the mga driver in the future (unfortunately I can’t remember the bug details). Then Klaus’ monitors will both work.
Site by Robert Spanton. © 2008 - 2018