Colour and interference issues with with new 3.2 revision (compared to 3.1)

NewHome Forums RetroTink 2 Products Colour and interference issues with with new 3.2 revision (compared to 3.1)

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
    Posts
  • #27118
    xythen
    Participant

    So I recently got the latest 3.2 revision of the Retrotink2x from the VGP store as an upgrade to the 3.1 original I’ve had since near release.

    After hooking it up I immediately noticed the picture was much darker and the colours seemed off, so I set up the two retrotink2x’s and made some comparisons.

    With my AV Famicom and Super Famicom (both unmodded, and using composite video) the picture quality is far worse on the new revision; the picture is darker, hue shifted, and oversaturated. I’ve taken some pictures here:

    Retrotink2x colour issues on new revision

    I don’t have a capture device, so excuse the moire from the monitor. It’s also not the clearest on camera, as compared to in real life where the difference is stark. In the Super Famicom pictures you can just make out that the saturation is overblown (easiest to see in the yellow writing and green grass). Also the white point is purplish (seen in the “SNES” writing, and appears stronger in person). The Famicom pictures show it a lot more clearly and closer to life. It’s obvious that the 3.1 picture has the correct colours for Mario and the sky, whereas the 3.2 is wildly off, like you’re using a garish palette in an emulator or something. You can also see that everything in the 3.2 image has a dark glow around it and looks smudged and less defined (compare Mario and the mushroom between the two pictures). Note that in both cases the setup is exactly the same, I’m just swapping out the two Retrotink boards. I also tried running the signal through an external comb filter which converts composite to s-video, and the results are the same, so it’s not the composite input itself causing issues. I also cycled through all available modes, and tried the limited and full range colour output options with the same results.

    Lastly I tested the only other system to hand, which is a NTSC PS2 Slim, and bizarrely the colours look fine in both composite and s-video. I did however notice strong low-frequency interference on the 3.2 version that isn’t there on the 3.1. It’s hard to see in a static picture, so I made a quick video:

    You can hopefully see the rolling interference pattern on the 3.2 revision. Again, the system and cables are the same, I’m just switching out the two Retrotink versions.

    Any ideas what’s going on here? Do I have a defective unit, or are there still issues with the new firmware? (I’ve seen the posts with people having issues with various systems like the C64).

    #27119
    paulb_nl
    Participant

    Are you by any chance using a European SNES composite cable with your Famicom? That would explain the darker image on 3.2 where auto luma gain has been disabled.

    #27156
    xythen
    Participant

    Hey, I think you were absolutely right! I dug out my old PAL SNES (which is hardmodded for 60Hz output), and the colours looked fine with that model. I’m impressed the PAL60 setting handled it fine as well, since the signal is probably quite far out of spec. I had no idea that the composite cables were different for PAL and NTSC NES/SNES and Famicom/SFC. When it looked fine with the first retrotink version I didn’t think to question it. Do you know what the actual specs are? The only thing I could find was an Ebay listing for a modded cable that added a resistor to the NTSC cables, which I guess must be inside the NTSC models.

    I still have the issue of interference and image noise, but after doing some more comparisons between the two last night I noticed that the interference is still there on the older version just not as prominent, and the noise is more smoothed out. So I guess it’s just the tweaks to the processing in the new firmware, erring on the side of less filtering. Thankfully, when I hook the retrotink to a HDMI -> component box, then into an OSSC, the filtering of the OSSC takes care of 90% of it, so the two are indistinguishable.

    Glad I don’t have a defective unit, and it was just my fault for not checking I had the right spec cable!

    #27157
    James-F
    Participant

    RT2x v3.2 has auto-gain disabled and fixed to accept 750mV without clipping, so obviously you have to use the correct cable like you would with a CRT.
    V3.1 had auto-gain enabled which boosted or attenuated according to the sync voltage which is referenced to 300mV,, this caused clipping with some consoles therefore was disabled for v3.2.
    BTW, with fixed gain of v3.2 it actually processes luma and chroma gain as a CRT would,, so your first post completely wrong and based on personal opinion and a wrong cable. 😀

    #27158
    paulb_nl
    Participant

    You said the hue is shifted and oversaturated. This indicates that the Retrotink is still using auto gain on Chroma and because the signal is so weak with the PAL composite cable the difference between Luma and Chroma level is quite large causing higher saturation.

    PAL cables contain a 75 Ohm resistor to ground on the video line. To convert it to an NTSC cable you simply remove the resistor.

    #27159
    James-F
    Participant

    Yep, v3.2 has auto-gain for chroma and fixed gain for luma.
    That’s how my Sony Trinitron processes color, and most CRT’s back in the day did.
    Commodore 64 comes to mind that relies on this behavior and looks correct on v3.2 like it did on a CRT with Composite signal back in the day.

    BTW, the wavy noise you see is not from the RT2x, but from the console, or your power supply, or poorly shielded cable.

    #27166
    James-F
    Participant

    Matt,, bot warning.
    Posting some nonsense then his link.
    All his posts are the same.
    DON’T CLICK IT!

Viewing 7 posts - 1 through 7 (of 7 total)
  • You must be logged in to reply to this topic.