[Coco] New 64-color RGB test chart program (re: RGB2VGA adapter)

Sun Nov 30 00:05:43 EST 2014

if it cannot distinguish the voltage difference between the 2 colors, then
the LUT is useless.

The color spectrum is being divided by 8, and you have 8 levels of red,
green and only 4 for blue. This makes a 256 colors. I have a 3 bit linear
R2R DAC that is used to test each color component: R, G and B.

But some coco3 colors could be borderline in one or more components, what
make it very difficult to capture all the colors.

Start finding the noise problem. As I told you remove this lm1881 it's
totally useless for the coco3 and could potentially generate noise.

For better colors, we would need a better DAC that could reproduce all
coco3 voltages. What can be done: for each color, make a full screen and
try to measure on the scope what voltage each R, G and B have. Once we know
all the voltages for all colors, we can make a DAC that could reproduce all
these voltage levels to test. But this is a daunting task...

Luis Felipe Antoniosi

On Sat, Nov 29, 2014 at 11:29 PM, Richard Goedeken <
Richard at fascinationsoftware.com> wrote:

> Hello,
>
> Today I took the 64-color demo program and rewrote it to have a color
> layout
> which is much nicer for RGB monitors.  I put both this new program and the
> original one on a disk image for everyone to use.  It's a useful test
> program
> for calibrating your monitor color settings.  You can download it here:
>
> http://www.fascinationsoftware.com/media/rgb2vga/CC3COLRS.DSK
>
> Once I ran this on my real Coco, I found that my CM-8 was adjusted totally
> wrong.  It's quite difficult to adjust the monitor so that you get decent
> contrast in both the darkest 2 shades of each channel, and the lightest 2
> shades.  My digital camera didn't capture the image exactly as I saw it
> (the
> contrast was a little bit better in the dark colors than what's shown
> here),
> but this is what a CM-8 looks like when adjusted as well as I could get it:
>
> http://www.fascinationsoftware.com/media/rgb2vga/RGBChart_CM8.jpg
>
> There is still room for improvement with the colors on the RGB2VGA adapter.
> The contrast problem is worse here, and pretty much no matter how you
> adjust
> the 10k pot, either the darkest 2 shades look the same, or the lightest 2
> shades look the same:
>
> http://www.fascinationsoftware.com/media/rgb2vga/RGBChart_RGB2VGA_Dark.jpg
> http://www.fascinationsoftware.com/media/rgb2vga/RGBChart_RGB2VGA_Light.jpg
>
> Could you put an 8-entry LUT on each color channel (one for R, one for G,
> and
> one for B) in between the input and output sections on this video
> adapter?  If
> so, we might be able to tweak it to get even better color reproduction than
> what is possible with the CM-8.  The output circuitry on the coco must
> have a
> very non-linear voltage response.
>
> I will try your suggestions to see if they clean up the noise.  I can
> desolder
> the unnecessary components.
>
> Richard
>
>
> On 11/29/2014 09:04 AM, Luis Antoniosi (CoCoDemus) wrote:
> > well 24 inches is a pretty long cable. I'm using 12 inches maximum.
> >
> > Did you check all the groundings on the board ? check the coco3 pins 1&2
> > and also altera's pins GIPO0 12 & 30.
> >
> > in my prototype the lm1881 is detachable, so it's never plugged when I
> use
> > the coco3.
> >
> > Try assembling another board without the lm1881, 680K, and the 2 caps, or
> > remove them for your current board.
> >
> >
> > Luis Felipe Antoniosi
> >
> >
> >
> > On Sat, Nov 29, 2014 at 11:30 AM, Richard Goedeken <
> > Richard at fascinationsoftware.com> wrote:
> >
> >> I probably need to change the front porch delay, because the left-most
> >> column
> >> of characters is cut off in 80-column mode.
> >>
> >> I think the noise is from a ground loop, but I don't know where.  The
> Coco
> >> is
> >> plugged in with a 2-prong plug.  I was powering the DE0-Nano board from
> a
> >> USB
> >> port on a laptop which was plugged in with a 3-prong.  But I changed
> that
> >> to
> >> use a 2-prong USB power adapter instead.   The noise didn't go away.
> The
> >> monitor has a power brick which uses a cable with a 3-prong plug, so I
> >> guess
> >> my next test will be to block the ground connection on the monitor and
> see
> >> if
> >> that helps.
> >>
> >> I also accidentally discovered that the sync resistors weren't right.
> >> When I
> >> was building up the board, at first I couldn't find the 82-ohm resistors
> >> in my
> >> pile of parts from digikey.  I thought that I had forgotten to order
> them,
> >> so
> >> I looked at the schematic to see what they were being used for.  I saw
> that
> >> they were for the H/VSYNC going to the VGA, so I searched online to find
> >> the
> >> input impedance of these on a VGA monitor.  I found something which said
> >> that
> >> most VGA cables have a characteristic impedance of 60 ohms in the lines
> >> used
> >> to carry the sync, so drivers are supposed to use 60 ohm resistors.  I
> >> looked
> >> around and found some 68 ohm resistors and used those.  Anyway, the VGA
> >> output
> >> is perfect, because the test pattern looks super sharp and clean.  The
> >> noise
> >> is definitely coming in on the input ADC side.
> >>
> >> I actually bought all the parts before you put the digi-key parts list
> on
> >> your
> >> blog.  I was looking at a BOM that you had posted here on the mailing
> list,
> >> and selected all of the parts on the digi-key website.  For some reason
> I
> >> ended up with two 4-position DIP switches instead of one 8-pos, and the
> >> particular 4-position DIP switches that I got are too wide to fit next
> to
> >> each
> >> other.  That's why I don't have any switches mounted yet.
> >>
> >> I do have other Cocos that I could test with.  The RGB cable that I'm
> >> using is
> >> a plain flat ribbon cable, so not shielded.  It's not very long, maybe
> 20
> >> or
> >> 24 inches.
> >>
> >> Richard
> >>
> >> On 11/28/2014 08:37 PM, Luis Antoniosi (CoCoDemus) wrote:
> >>> The customization is about the front porch, the code now is pretty
> stable
> >>> to be generic. But yours seem to work. Steven Hirsch had to change on
> >> him.
> >>> He didn't report about noise. So I can't tell if is a particularly from
> >>> yours or if this board has indeed noise. Until my mules decide to cross
> >> the
> >>> border I can't help you on this.
> >>>
> >>> BTW, Steven has found out theat the correct values for the sync
> resistors
> >>> are 47 ohms instead of 82. I doubt this could be the source of problem.
> >>>
> >>> You don't need the LM1881, the 680K and the 2 caps if you're not
> planning
> >>> to connect any other computers on it.
> >>>
> >>> The scalines help improve the colors. You may turn them on, use a dip
> >>> switch. I have changed the dip switches, the scanline is now the label
> >>> shrink.
> >>>
> >>> You don't have another coco to test right ? Also, how long is your
> cable
> >>> from the coco3 ? Is it shielded ? Do you have anything else that could
> be
> >>> potentially noisy ? Tried connecting the altera's usb cable in a
> >> different
> >>> power source, like iphone or samsung usb chargers ?
> >>>
> >>> Luis Felipe Antoniosi
> >>>
> >>>
> >>
> >> --
> >> Coco mailing list
> >> Coco at maltedmedia.com
> >> https://pairlist5.pair.net/mailman/listinfo/coco
> >>
> >
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> https://pairlist5.pair.net/mailman/listinfo/coco
>