Composite video to VGA, Intel 8275

Hi all,
I am about to convert a composite video signal (CRT screen) on an old robotics controller to fit a VGA (TFT screen). I have ordered an ATLONA BNC video to VGA converter and managed to get an image on the TFT. The problem that remains is that some characters are unclear on the TFT screen e.g the curved part of the character "P" and "R" also the diagonal part of "R". These parts are only represented by single dots and is therefor unclear.
The characters are housed in an Eprom. The Intel 8275 addresses the prom and the character is converted from paralell to serial data which forms the composite video signal.
The problem can relate to either : 1) a weak composite signal 2) reading the data sheet of the 8275 says that by making the "Dot Clock" longer it is possible to get wider dots.
Making the characters wider in the Eprom is not possible.
Does anyone have an idea how to improve the image?
Kind regards , Bo - SWEDEN -

Schematics_01Schematics_02Block diagr. 8275

While the standard-definition composite video bandwidth is compatible with VGA resolution, your TFT may not be. If the display panel is not capable of 640x480 pixels in each color (RGB), its controller will interpolate the video signals to fit, and this often produces less-than-desirable results. In general, flat-screen controllers resample the input video in whatever way they see fit, and some do a better job than others, so you might try other video screens and see if the problem lies there.
I don't know if a "weak" composite signal is a likely source of the problem, as long as it is reasonably close to 1 volt peak-to-peak (sync tip to maximum white) at the 75 ohm termination of the converter input. But that level is worth checking. Also, my experience with Atlona video converter products has not always been positive, so you might try another brand and see if the output picture is better.
I think that changing the dot clock may not get the result you want, since reducing its speed to lengthen the dots will also reduce the number of dots you can display per video line. A 40-character line IS going to be more legible than a 64-character line. But that may be a tradeoff you wish to make for your application.
Good luck!

Dear Mark,
regarding composite video to VGA-TFT.
Thanks for the promt and kind reply!
What I´ve disovered so far is that the composite signal level is about 4V positive peak out. That is due to that the signal is generated through a transistor emitter follower with a 100 ohm resitor from the emitter to ground. Maybe the level is to high to give a clear image? I will try to introduce a voltage divider instead of the 100 ohm resistor to lower the signal.
My TFT has different input options: BNC (which I´ve now tried), DVI, S-Video and more. Is there a possibility to get better shape by using another converter converting to e.g. S-video or DVI?
My oppinion is that the problem lies in the composite itself or as you mention the converter or that the screen has to interpolate to give the characters in a good and clear way. All comments are welcome.
Kind regards , Bo

Oh, I looked at the schematic - that is an ugly way to send video, but typical of digital designs of the '70s! 4 volts p-p will almost certainly severely overload the converter's input. A voltage divider at the emitter will work OK, though ideally you should have a 75 ohm resistor in series with the output (video is an early controlled-impedance transmission method). Set it for 1 volt p-p at the input of the converter. I don't know what the series capacitor is but it needs to be about 470 uF, with a 0.1 uF in parallel.
S-Video is slightly better for signal quality than Composite, since the chroma (color) information is sent separately from the monochrome. They're combined in Composite Video, and a decoder has to separate them in the display, with resulting possible errors/losses. Since it appears that your video generator is monochrome (no color), this might not make a difference, though it depends on the display's decoder circuit -- some of them not only expect but even REQUIRE the color signal to work properly. You can feed your video output into the Y input (pin 3 of the connector), with the same 1 V p-p requirements as Composite. See http://en.wikipedia.org/wiki/S-Video

Good luck!

I´ve tested different voltage dividers dimensions and have got the level to 2V p-p but its just slightly better. A 75 ohm serial resistor makes it worse.
The earlier specified parts of the characters (small dots in the diagonal parts) is likely not caused by the composite signal. ((it is not seen on the CRT cause of the beam lag on the phosphore) and the processing on the converter I use and probably also the processing in the TFT screen is the cause of the smaller dots. I do not know what happens inside the converter or the TFT but its clear that the diagonal parts are becomming smaller after processed. I don´t know if changing the converter or changing the TFT would help?
Of what I can see the signal is processed so the specific parts of a character (dots) are diminished. Is there any way to make the specific parts clearer?
I attach an image of what a diagonal character can look like (diagonal part of a "K")
This looks better on the photo than in reality (don´t have a good macro).
Many thanks for the engagement in the problem.