MiSTer FPGA Forum

Greetings!
I recently assembled mister setup and stumbled across one problem - I observe non-uniform gradient ramps in 240 test suite (version 1.22) when outputting through VGA port on analog IO board. Has anybody experienced such a behaviour and what can be done to fix it? Can it relate to output produced by the core? Can it have something to do with 18 bits of color precision of vga output (does genesis core produce 16 bit - 565 rgb - output internally)?
I checked RGB output from real Mega Drive 2 console and the gradients it outputs were free from color deviations. I also checked this with two CRTs - JVC TM-H150cg and a 14" Sony Consumer CRT KV-M1400K - both show different gradients.
Maybe the problem is that I am using non-official analog IO board - my board was purchased on Aliexpress and it is actually a combined analog IO + usb Hub board sold by Gamenaissance Store (btw they also sell another version of the board - with saturn output port and that version should be able to output s-video and composite, selectable with dip switches).
Can anybody with an official IO board compare gradients on vga and hdmi outputs? And maybe somebody can comment on the internal format of the picture produced by the core and how it then get converted to analog signal?

This is probably a mismatch between the core's output and the lack of precision on the VGA port, like a rounding error.

VGA port doesn't have full color depth, so that very well could be the culprit. You need direct video for full range IIRC.

With the 6 bit limitation of the IO board you might see uneven brightness but since grey color means 3 equal R,G,B signals there shouldn't be any tint. Unless you are using YPbPr.

Check the grey ramp in the SNES core. It has a more linear color output than the Genesis.

You should also make sure you are not using Gamma correction.

I saved screenshot from 240p test Suite - each bar in gradient have equal r, g, b values. Checked the gamma - there was no gamma applied.
Color tinting also visible in SNES 240p test suite.

Hdmi output is fine except one thing - it is efficiently outputting 240p super resolution 240p if I am not mistaken. The point is - every original pixel becomes 3x, 4x or maybe even 8x horizontal pixels in super resolution. This changes the picture drawn on a CRT - transitions between neighboring pixels get more sharp, edges of scanlines become more rectangular. Not exactly a bad thing, picture gets even more sharp on a good screen, but not as authentic as original hardware.

I experienced this first when compared native 320x240 vs 2560x240 from PC emulation, picture got razor sharp, almost like an LCD (was especially visible with NES emulation with its 256(x10)x240 picture).

But strange thing - I tried to compare VGA and HDMI direct-video on Mister and I cannot see such drastic sharp super resolution picture as I got from a PC (I don't know if direct output does some interpolation on pixel transitions when it is forming super resolution output line - that would explain softer picure compared to PC emulation).

It's interesting you mention those subjective observations about MiSTer video output, sersei. I have one as well.

Using the MiSTer's analog VGA out (YPbPr) to a consumer Sony CRT I can't shake the feeling that the scanlines become indistinct / swimmy when backgrounds or sprites move vertically, which makes the pixels of the affected objects look disagreeably blocky/pixelated in a way that I don't recall them looking on original consoles or RetroArch's CRT switchres feature outputting to same CRT.

But maybe it's always been like that across all hardware and I just misremember. I will have to pull out the consoles sometime to test.

To close this issue - it seems that the problem was in some of the the RN1-RN11 resistors. I tried shorting some pins of these resistors to see if it will have any effect on the analog output signal (and as expected there were some red/green/blue tints to various parts of the gradient) . And as I was shorting the pins grayscale gradients suddenly became normal and now stay uniform, free of any initial tints. I do not know if it was a bad soldering or shorts on these, but now everything is beautiful. Will reflow solder on these resistors later to be absolutely sure they are fine. Probably one should be cautious when purchasing non-official boards.

Is that a big solder blob on RN16?

I think there is a capacitor soldered to power button...