@ibmapc
Sorry I am not prepare to trade this card atm no matter the item offered, I spent quite some time looking for it and currently use it in my EGA setup

I spent some time cleaning my board from flux and I think it helped. Maybe there was some parasitic interference or something and I am pretty sure that once I install the 1% resistor packs things will improve further. I took some pictures:

keropi wrote:

@ibmapc
Sorry I am not prepare to trade this card atm no matter the item offered, I spent quite some time looking for it and currently use it in my EGA setup

That's completely understandable. I have one that works OK in CGA and EGA modes but goes bonkers in Hercules mode. So, I've been attempting to acquire another one. So, yes I know how hard they are to acquire! I saw your picture and had to ask. If you ever find you have one that you don't need, let me know.

keropi wrote:

I spent some time cleaning my board from flux and I think it helped. Maybe there was some parasitic interference or something and I am pretty sure that once I install the 1% resistor packs things will improve further. I took some pictures:

Wow! Now it's perfect. If I was you I wouldn't bother replacing them. It won't make any change. I've been using 1% and 5% and I can't notice any difference.

Right now the board uses only 2 bits (3 for the amber emulation). From 1000 to 500 ohms we are talking about 50%. Using 1% to 5% won't make any difference at eyeball...

You problem was some short circuit

and WHY U NO LIKE SCANLINES ????

retrocanada76 wrote:

and WHY U NO LIKE SCANLINES ????

oh I like them but I wanted to show the colors in the shots, scanlines would make that more difficult
I love both scanlines and aspect ratio 🤣

Idk, the colors still look off to me. Dark gray is still too light and there needs to be more contrast between the light/dark red. The purple on keen's shirt is too "neon" as well. Can you take a screenshot of the gradient colors?

^ I will later when I am back home
about the gray in the gobliiins pic, look at the gray at the upper part of the tree - that's the tone it has but the phone does some color correction or something depending on it's focus point , I did try to make the pics look as close as possible to what I am seeing.

check the brightness. It would be wise to set all to default levels. Also ther is the picture mode: cool, warm, etc.

keropi wrote:

^ I will later when I am back home
about the gray in the gobliiins pic, look at the gray at the upper part of the tree - that's the tone it has but the phone does some color correction or something depending on it's focus point , I did try to make the pics look as close as possible to what I am seeing.

It might help if you used test patterns instead of game shots -- large areas of solid colors, for example.

I use these specific game shots because I know how they look on a real ega monitor so I can compare them easier. Will upload some patterns later since they are a baseline for others to compare

allright this is what I am getting with MobyGamer's suite - at least the tests that work OK with EGA:

and this is what I am getting with CGACAL

I adjusted the gamma a little on the CGACAL pic to represent what I am seeing on the monitor. Light Grey is the same as the one on the sample pic on the site but the light red+magenta seem to be too hot.

My monitor is always at default settings, via HDMI no calibration is needed as the colors/gamma are OK for me.

From that picture, the light colors look too saturated, especially blue, red and magenta. But it could be the camera/monitor? Can you take a "control" photo of CGACAL with the same monitor but hooked directly to a VGA card?

Here is a pic on the same monitor, vga cable and default settings - source is my ss7 build with a Diamond S3 vga:

I think this shows the monitor is not the issue for me.

Yeah dang my board is not generating the correct bright colors.

Look, this is pretty simple. These deep colors, that's because when you get bright blue and am only sending Blue color as full bright, but to achieve this degree of blue you need to add to red and green as well. I am afraid it get to washed out. At least magenta wasn't vivid enough.

I will fix that in a new firmware....

But by the other hand what I need is a real 5152 monitor output for comparison. I'm not mimicking VGA colors but CGA ones.

retrocanada76 wrote:

Yeah dang my board is not generating the correct bright colors.

Look, this is pretty simple. These deep colors, that's because when you get bright blue and am only sending Blue color as full bright, but to achieve this degree of blue you need to add to red and green as well. I am afraid it get to washed out. At least magenta wasn't vivid enough.

I will fix that in a new firmware....

You do realize that keropi is using Dreamblasters board?

But, on another note, I find with your v2.0 board, the colors look pretty good in 640x200 mode, but in 640x350 the red is a little strong. I'll take some pictures later on. I don't have a 5153 to compare, but I do have a Multisync CRT that is pretty decent. I'll try to show some comparisons with that.

retrocanada76 wrote:

But by the other hand what I need is a real 5152 monitor output for comparison. I'm not mimicking VGA colors but CGA ones.

Well the people that made CGACAL have a shot on their page about what CGA colors should look like: http://www.oldskool.org/pc/cgacal , maybe this can serve as a fixed point?
I have 2 multiscan, 1 ega-only and several 1084s that have TTL inputs but you can't really trust them without a fixed point like the shot above, you can easily mess the colors on a crt monitor just by moving the contrast pot a couple of millimeters... the Quadram multisync I have on my desk is especially vulnerable to this.
This is coming from my own experience though, I am not an expert in these stuff 😊

I've been calculating stuff. Does these calculations explain why some colors look off?

VGA input is terminated into 75 ohms, signal voltage 700mVpp.
EGA has two bits per color component, and so the EGA palette in 6-bit VGA DAC world is 0, 21, 42, 63, and modern 8-bit world 0, 85, 170, 255.
So generated analog VGA voltages are exactly 0/3, 1/3, 2/3 and 3/3, or 0mV, 233mV, 467mV, and 700mV.

FPGA output has 3-bit DAC, resistor values 2k, 1k and 500, IO voltage 3.3 volts, right?
I assume that all 3 ladder IO pins are configured as outputs, so the ladder output resistance is about 286 ohms then.
Given the output resistance and VGA termination, the maximum voltage is 686mV.
Not that much less than 700mV for peak brightness, should not matter much.

However, the eight possible output voltages are in multiples of 98mV:
0,98,196,294,392,490,588,686.
When ideal VGA DAC voltages are scaled from 700mV to 686mV, the four EGA color voltages would be multiples of 229mV:
0,229,457,686
So they are difficult to match.

I mean, 3/3 would match 7/7 exactly for 686mV, but 2/3 closest match is 470 (7% error in voltage) and 1/3 closest match is 196 (14% error in voltage).
Note that while DAC voltages are linear, due to gamma/human perception being logarithmic, the brightness difference is actually larger in the darker end.
So there is a greater brightness difference perceived between 200mV and 250mV than from 650mV and 700mV even if both have 50mV difference.

One solution might be to tri-state the FPGA resistor ladder LSB, so the DAC is actually 2-bit. Then the ladder output resistance is 333 ohms and peak voltage 606mV, and the four voltages would have 202mV steps.
Not 100% sure but it might look better, monitor brightness control might just work to bring brightness up while preserving the intended brightness differences.

Or if you are going to use 1% resistors anyway, choose them wisely. The ladder output resistance should be 278 ohms. So for a 2-bit DAC, it would mean ideally resistances of 418 and 2*418, if I calculated them correctly.
Also depending on how much current the FPGA output pins can drive, the output impedance at the VGA connector should ideally be matched to 75 ohms.

keropi wrote:

Well the people that made CGACAL have a shot on their page about what CGA colors should look like: http://www.oldskool.org/pc/cgacal , maybe this can serve as a fixed point?

The color values listed there are the theoretical "perfect" colors based on the RGBI output math and the 5153 monitor schematic (which is where the correct #6 brown value comes from -- it's the monitor that turns #6 from yellow to brown). If you match those colors as best you can, that's ideal.