# [mythtv-users] RE: [ Matrox G400 ] VGA - SCART cable WAS: Hotwo enable flicker filters

Bernd Müller caret at gmx.de
Sun Mar 20 01:48:42 UTC 2005

Hello;

I didn't build the cable until yet. But will do next days. Your last posting was
a little bit scarring me, as you wrote that your informations are "only"
hypothetical. Up to this point i've had believed you had build this cable and
are using it successfully with the G400. As i've had spent so much  time in the
G400 and had also build three different VGA-SCART cables, which all do not
worked for me, my motivation was not so high to test an new hypothetical cable.
But as Peter wrote he is using this cable, although with a simple CAT5 wire, i
will now hot my soldering iron.
I've to solve an other problem before. I'm not eble to find any valid PAL
modeline for X. I tried the following ones

Modeline "768x576" 14.767 768 769 839 945 576 577 581 625 interlace -csync
Modeline "768x576i" 14.75 768 784 864 944 576 582 588 625 interlace -csync
Modeline "720x576" 13.875 720 744 808 888 576 582 588 625 interlace -csync

but get

(==) MGA(0): Min pixel clock is 12 MHz
(==) MGA(0): Max pixel clock is 360 MHz
(II) MGA(0): PAL TV: Using hsync range of 15.00-80.00 kHz
(II) MGA(0): PAL TV: Using vrefresh value of 50.00 Hz
(II) MGA(0): Clock range:  12.00 to 360.00 MHz
(II) MGA(0): Not using mode "768x576" (bad mode clock/interlace/doublescan)
(II) MGA(0): Not using mode "768x576i" (bad mode clock/interlace/doublescan)
(II) MGA(0): Not using mode "720x576" (bad mode clock/interlace/doublescan)

This is a little bit confusing me, because minimal dotclock is 12 MHz. The modes
should be all valid. I've i general problems to use any interlaced mode for X.
If i try some non-interlaced 720x576 at 50Hz modes with dotclocks around 27 MHz X
will accept them. But they are not interlaced. Interpreting the above error
message i would say, the G400 is not able to generate PAL dotclocks around 14
MHz. But from web i know the card is capable.

The importend lines from my XF86Config:

Section "Device"
Identifier  "Device[1]"
Driver      "mga"
EndSection

Section "Monitor"
Identifier  "PAL TV"
HorizSync  30.0-80.0
VertRefresh 15.0
Modeline "768x576" 14.767 768 769 839 945 576 577 581 625 interlace -csync
Modeline "768x576i" 14.75 768 784 864 944 576 582 588 625 interlace -csync
Modeline "720x576" 13.875 720 744 808 888 576 582 588 625 interlace -csync
EndSection

The matroxfb support is compiled in the kernel, but X don't use the framebuffer
with this configuration. So i assume it dosn't matter.

At least i would like to use the G400 with this general VGA-SCART cable. The
solution with the framebuffer, i'm using it until yet, is frustrating me. First
i've spend much time to configure kernel, frambuffermode and X, soldering the
special matrox VGA-SCART cable an now i've to notice the whole setup does not
work reliable. Changing framebuffermode, setting the registers for contrast, hue
ist working in general bu sometimes not. Sometimes i simply get a greyscale or
distorted image. Assigning frambuffermode or contrast a seconed time leads to a
good image. I don't know if the problem is in hardware or software. But i know a
solution using the general VGA-SCART cable has to be more reliable and more
flexible 'cause some newer grafic cards, especially the ATI Radeon 7500 and
higher, are supposed to be capable to generate valid PAL/NTSC signals

(c; bernd

Cory Papenfuss schrieb:
> On Fri, 18 Mar 2005, Peter van Gils wrote:
>
>> Hi Bernd,
>>
>> Just a tip: you can save some money next time you need such
>> a cable. I use Cat5 UTP for a 13m VGA cable and it works
>> great! (Running 1024x600 @ 60Hz on a 32" LCD.)
>> You have to keep signal and ground together in one
>> twisted pair and add 22 ohm resistors in series with
>> the signal wires at the TV end. (The resistors are
>> needed to prevent ghost images.) The shell is left unconnected.
>>
>> My \$0.02.
>>
>     What you are describing can be done, has been done, and there exist
> commercial products for doing so:
> http://www.svideo.com/coaxbalun.html
> ... and many others.
>
>     There are some issues there, though.  The one you propose fixing is
> the impedance one... video signals typically are run on 75 Ohm cable,
> the and nominal impedance of Cat5 is 100 Ohms.  It's also \pm 15 Ohms, I
> believe, so that's not so good.
>
>     There's also the balanced/unbalanced issue.  Coax is unbalanced but
> shielded.  Cat5 is balanced (differential), but unshielded.  If you put
> the unbalanced signal on it directly, you don't get the benefit of
> differential signalling and it radiates a lot of power.  This guy says
> it better:
> http://hardware.mcse.ms/message159780-2.html
>
>     There's also the attenuation.  The high frequencies of baseband
> video (6MHz) will be attenuated 5dB or so over a 100 ft runk, but the
> low frequencies not as much.  It will soften the image, and possibly
> even cause other high frequency (read: chroma) artifacts.
> http://www.chem.ox.ac.uk/it/staff/Network%20Documentation/standards.html
>
>     That said, what you propose will probably work OK for short runs,
> but coax is definately better.
>
> -Cory
> *************************************************************************
> * Cory Papenfuss                                                        *
> * Electrical Engineering candidate Ph.D. graduate student               *
> * Virginia Polytechnic Institute and State University                   *
> *************************************************************************
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users