Quantcast

Page 4 of 5 FirstFirst 12345 LastLast
Results 46 to 60 of 65

Thread: Possible rainbow banding fix [idea]

  1. #46
    Road Rasher
    Join Date
    Mar 2010
    Posts
    308
    Rep Power
    16

    Default

    Quote Originally Posted by andrebrait View Post
    Except they are not pin compatible, AFAIK.
    judging by datasheet, they're very similar, most of the pinout is the same and 1645 requires less components as Y/C mixing is done internally. Only thing is, it doesn't buffer mono audio.

    Might actually try on my test MD

  2. #47
    WCPO Agent Helder's Avatar
    Join Date
    Jun 2013
    Posts
    752
    Rep Power
    14

    Default

    Quote Originally Posted by andrebrait View Post
    Thank you!

    I'll look into that, as I'm sorry to report my attempt has failed. Converting S-Video to Composite resulted in somewhat weird colors artifacts (not many, but stilk noticeable) and sharp composite video, which is ezactly what I didn't want.

    EDIT: what exactly do you mean by tweaking the resistors in the sync line? Do you think I should replace them with ones with higher or lower resistance?
    I mean in the circuit I linked to, the sync line has a few resistors and you can replace them with a pot to adjust as needed but the extra cap on sync towards the output can and should be removed.

  3. #48
    Road Rasher
    Join Date
    May 2015
    Location
    East Coast AUSTRALIA
    Posts
    446
    Rep Power
    7

    Default

    Quote Originally Posted by andrebrait View Post
    Except they are not pin compatible, AFAIK.
    Near enough, just cut the tracks that are different and connect them where they need to go.

  4. #49
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    Hoping to revive this thread with some additional information.
    I'm trying to get a rainbow banding free composite picture as well. To that end I've now tried replacing the CXA1145 with a modern CXA2075.
    The 2075 has all the high precision parts internal and comes with a nice Y-Trap filter.
    Alas.. No amount of tweaking the filter helps with the horrible rainbow banding.
    It appears there's some flaw in the Genesis video generation design that makes composite encoders produce rainbow banding.
    Quote Originally Posted by andrebrait View Post
    Converting S-Video to Composite resulted in somewhat weird colors artifacts (not many, but stilk noticeable) and sharp composite video, which is ezactly what I didn't want.
    I get the same with the CXA2075 Composite: Sharp pixels and artifacts.

    Looking at the scope, I believe this is caused by some kind of phase error on the subcarrier clock or maybe in the way the RGB + CSync behaves when mixed with that clock.
    I did try shaping the subcarrier by introducing coils, resistors and capacitors. It does change the look of the artifacts but even a nice sine wave still has them.

    For comparison, the same CXA2075 works perfectly fine in a Playstation.

  5. #50
    Nonconformist Hedgehog-in-TrainingWCPO Agent EyeDeeNo76's Avatar
    Join Date
    Dec 2015
    Location
    97km north of New York City
    Age
    43
    Posts
    891
    Rep Power
    25

    Default

    It's nice to read that someone is still poking around this disorder. Sounds like something that I/we will just have to live with.

  6. #51
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    I believe I will find the problem by continued elimination of factors.
    - The encoder itself is now out of the picture.
    - The 32X demonstrates that re-encoding the amplified Genesis RGB also fixes the issue.
    - Common suggestions like recapping help with some issues, but will not affect the banding.

    That really only leaves the system clocks (phase, signal shape), especially the subcarrier, and possibly the RGB / CSync circuits out of the VDP.
    Unfortunately, I only have a basic analog oscilloscope and I don't have experience with using it for diagnosing clock jitter and phase problems.
    But yea, some day a fix will be found

  7. #52
    WCPO Agent CrossBow's Avatar
    Join Date
    Feb 2017
    Location
    Ivory Tower Collections
    Posts
    751
    Rep Power
    10

    Default

    Was going to say that s-video modding a Genny 1 with the CX1145 doesn't remove the banding effect. In fact, it replaces the banding with what we call jailbars. The reality is that the jailbars were there all along and on composite it shows up as a banding effect with mixed signal from composite. With S-video and I would guess RGB as well, you will notice the jailbars as they truly are.

    Now having said that I did mod my 32x with separate s-video and I can still see faint jailbars on darker hues in large areas of the screen. Like the sky in Sonic..etc. However, TVs do make a difference. On most CRTs I've displayed the s-video and even the composite on PVMs, the jailbars and rainbow banding aren't seen. On my cheaper 42" Insignia LED...you could see it everywhere. On my higher end Samsung LED...not so much unless you look up close to the screen.

  8. #53
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    Jailbars are a common issue on retro consoles and they're a relatively simple fix. You just need to employ some denoise methods like shielding and additional capacitors.
    The rainbow banding on Composite is a tough nut though

    Edit:
    Try adding 1000uF low ESR capacitors for each 7805 in model 1 Mega Drives (anywhere near their 5V outputs).
    Also replace the various 100uF and 47uF electrolytics that are near the VDP, CPU and RAM / VRAM with 220uF low ESR.
    This will reduce jailbars to maybe 30% of their original level.
    Last edited by rama2; 02-15-2017 at 03:48 AM.

  9. #54
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    This is a reasonable source for the problem. Posted on the nesdev.com forums.
    Code:
    In 320px mode, the Genesis VDP's dot clock is ~6.712 MHz, and an alternating dark and light pattern would oscillate at half that (~3.356 MHz). 
    But this is very close to the NTSC chroma subcarrier frequency (~3.580 MHz) and thus well within the conventional 3.0-4.2 MHz chroma band. 
    Thus the TV interprets the signal as rainbows repeating every 30 horizontal pixels.
    If that's true however, it will make fixing it quite challenging

  10. #55
    Wildside Expert
    Join Date
    Aug 2013
    Location
    Portugal
    Posts
    199
    Rep Power
    8

    Default

    I believe that the only way to eliminate banding/jailbars is to ground the original chroma carrier to disable it and put in your own crystal. I modded my megadrive with both a PAL 4.43Mhz crystal and an NTSC 3.58MHz and I see no banding on neither cystals and RGB is 100% clean, looks almost like and emulator :P

  11. #56
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    I've tried using a 4.43Mhz crystal before, creating a PAL60 output. The image had color but lots of dot crawl and general ugliness.
    It could've been a poor quality signal though, as I didn't have an oscilloscope back then and made it all blindly with some schematic.
    PAL encoding could bypass the oscillation problem. So I should try again.

  12. #57
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    Update:
    The CXA2075 needs careful denoising and ringing prevention. Then its Y-Trap filter will work very well and fix most of the rainbowing.
    Imgur Album with the results: https://imgur.com/a/t361h
    It's not perfect because it's just so hard to denoise model 1 Genesis consoles. With more tuning on SC and SYNCIN, it can probably be perfect.

    In conclusion: The original CXA1145 is lacking a much needed Y-Trap filter. Replace it with CXA1645 (from a PSX, for example) or a CXA2075 (Ebay), denoise the Vcc lines of the Genesis carefully, enable the Y-Trap filter, done.

  13. #58
    Road Rasher
    Join Date
    May 2015
    Location
    East Coast AUSTRALIA
    Posts
    446
    Rep Power
    7

    Default

    Quote Originally Posted by rama2 View Post
    Update:
    The CXA2075 needs careful denoising and ringing prevention. Then its Y-Trap filter will work very well and fix most of the rainbowing.
    Imgur Album with the results: https://imgur.com/a/t361h
    It's not perfect because it's just so hard to denoise model 1 Genesis consoles. With more tuning on SC and SYNCIN, it can probably be perfect.

    In conclusion: The original CXA1145 is lacking a much needed Y-Trap filter. Replace it with CXA1645 (from a PSX, for example) or a CXA2075 (Ebay), denoise the Vcc lines of the Genesis carefully, enable the Y-Trap filter, done.
    Id be interested to test these on one of my commercial grade CRT projectors, ive only ever used RGB and never had a jail bar issue on these projectors, but ive got to wonder how much the projectors themselves are purifying the image too, id be hard pressed to say it could really improve much.

  14. #59
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    The newer encoders have faster slopes in RGB, theoretically resulting in a sharper image. You won't see a difference in practice though, unless you have a perfect upscaler and use an LCD.
    Everything else, including projectors, is bandwith limited even before the performance of a CXA1145.

  15. #60
    Road Rasher
    Join Date
    May 2015
    Location
    East Coast AUSTRALIA
    Posts
    446
    Rep Power
    7

    Default

    Quote Originally Posted by rama2 View Post
    The newer encoders have faster slopes in RGB, theoretically resulting in a sharper image. You won't see a difference in practice though, unless you have a perfect upscaler and use an LCD.
    Everything else, including projectors, is bandwith limited even before the performance of a CXA1145.
    Well thats not strictly true. These are not consumer grade displays.

    The Barco Cine 9 / Reality 909 CRT projector has FAR great bandwidth than you need for this, it has a max input resolution of 3200x2560 at 60Hz, which is around 180kHz scanrate, so yeah, far more bandwidth than youll ever need for a 15.7kHz signal. It was and probably will always be the absolute sharpest and most capable CRT projector ever made.

    15-20MHz bandwidth is more than you need to show the best out of these consoles, we're only displaying 240p, and ive got over 200MHz, so it is not even close to any limits at all.

    Not sure what bandwidth the CXA1145 has without looking up the specs in the data sheet, but id be amazed if it was even 15MHz.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •