Quantcast

Results 1 to 5 of 5

Thread: Random ruminations regarding the CXA1645M

  1. #1
    Outrunner Eep386's Avatar
    Join Date
    Sep 2014
    Location
    Not far from Sacramento
    Posts
    535
    Rep Power
    19

    Sweat Random ruminations regarding the CXA1645M

    Hello all,

    Not sure if this is the place to talk about this, but here are my findings about the CXA1645M encoder.

    - None of the released technical manuals for the Genesis / Mega Drive consoles, at least the ones I've seen thus far, make any mention of the CXA1645M, only its older CXA1145 ancestor.
    - It has a lower part count than the 1145M, requiring fewer SMD capacitors and resistors and only one inductor coil. It has a higher external part count than the KA2195D however.
    - On the VA4 using a CXA1645M, the inductor used on L2 is 12 uH, going by the color band code. The pad for L1 is unpopulated. Also, the value of R21 is 47k (473), where on other systems using other encoders the equivalent resistor (R54 on the VA1, say) is a 24k (243). On the VA4, the values of the capacitors at C34 and C28 are as of yet unidentified. (I'd have to desolder them and test them with my multimeter, and I'm kind of afraid to mess up this VA4.) The pad at C31 has a 20k resistor installed instead of a capacitor.
    - Composite picture quality is generally quite good with this encoder, though it can pixellate really badly if the TV's sharpness is set too high. (Now that I've turned down the sharpness of my TV, I can really see this for myself. A 50% setting has only subtle visible pixellation on my 32" RCA TV.) Also, VA3's using this encoder have heavy jailbar artifacts too; I am especially eager to find out what causes this. Unlike the KA2195D though, it doesn't always respond so readily to a boosted capacitor value at CE23.

    I am eager to get another VA2/2.3 or VA3 with CXA1645M encoder so I can study its circuit in depth. Ironically I used to have a VA2 with a 1645, until I changed it to a CXA1145M just to see if I could. (Yeah, I know, derrherrherr...)


    ----

    Oh, and for the record, I recant what I said about the 1645 being 'notorious' for pixellation. In fact, it only pixellates if the sharpness is set way too high on the TV, or on VA3 Model II's for some reason I've yet to discern. If anything, I want to find out how to adapt KA2195D systems to the CXA1645M, in theory it should be easier than adapting them to the CXA1145M.
    Last edited by Eep386; 05-06-2017 at 02:26 AM.

  2. #2
    Hedgehog-in-Training Hedgehog-in-TrainingSports Talker
    Join Date
    Sep 2015
    Posts
    40
    Rep Power
    0

    Default

    You probably know this but Sony's PSX uses this encoder and there's a service manual available for it (PU-18 board).
    On that machine, the encoder works very well. You don't find any jailbars or rainbowing, even if the software uses a 320px wide resolution.
    If you want to know how to use this encoder to its best capability, you should study it in a PSX.

  3. #3
    Outrunner Eep386's Avatar
    Join Date
    Sep 2014
    Location
    Not far from Sacramento
    Posts
    535
    Rep Power
    19

    Default

    I wouldn't mind learning how to change any given Model II to a CXA1645M, as even at its worst it generally looks better than the awful KA2195D. (The VA3 Model II has a pretty ugly picture with a CXA1645M, but even then it's sharper, not quite as pixellated and doesn't suffer from off-tint color like the KA2195D. Nothing says visual failure quite like a pinkish Knuckles taunting slightly cyan-green Sonic, on a KA2195D.)

  4. #4
    Road Rasher l_oliveira's Avatar
    Join Date
    Jul 2010
    Posts
    258
    Rep Power
    13

    Default

    PS1 feeds the CXA1645 a triangle waveform (no, it doesn't use attenuation with a RC circuit like the CXA1645 datasheet suggests, it uses a real triangle waveform output by the graphics chip) to it's CXA1645 and that's why you can't see "edges" on the color transitions. The edge on the waveform is what causes the artifacts. Optimal waveform for such video encoders are a real sine wave but because CXA1645 doesn't have a cristal oscillator driver (remember, the reason why CXA1145 has artifacts on the blue signal is due to inductive interference from the crystal oscillator driver which is the actual pin right after the blue input) you have to feed it an external waveform.

    If you use a free running crystal oscillator with a CXA1145 you get a different problem, called "dot crawl". It's caused by a difference between the timing of the video chip horizontal synchronization and the actual waveform generated by the chroma carrier crystal. I believe this is related to the discrepancy of actual timing of the waveform generated by the oscillating crystal versus the alloted time slot it will be squeezed into (following the signaling coming from the video chip in the form of the colorburst signal).

    Modulate the color information with the clock from the color oscillator then fit it into the slot for color burst into the sync signal. So if the clock for color comes from the videochip it will be synchronous with the timing of the color burst slots and no dotcrawl will happen (but you will have edge artifacts). Should the color carrier signal drift in frequency (quartz oscillators do drift) you will have a "wavy misty" pattern crawling through the screen.

    Do you think this information is of interest for you?

    Besides the complexity levels, these SONY encoders are pretty much similar. Of course the CXA1145 is much better than the newer ones if properly configured but achieving a proper configuration is hard. Also cost wise the newer encoders beat it heavy so they were preferable on newer designs.

    PS1 is the showcase of how to get the best possible image out of a SONY RGB video encoder.
    I'd like to share some wisdom:
    Quote Originally Posted by TmEE
    I'm too lazy to be unstoppable.
    you can have shame, as long as you don't feel it

  5. #5
    Outrunner Eep386's Avatar
    Join Date
    Sep 2014
    Location
    Not far from Sacramento
    Posts
    535
    Rep Power
    19

    Default

    I see, that would explain why putting a 1000pF cap at C31 (the capacitor component of the colorburst clock RC filter circuit) visibly reduced the edge pixellation, but also made the display a bit blurrier and the color a little bit off-tint. (I tried this on both a CXA1645M and a KA2195D, and the KA2195D benefited the most from it - sure it became much blurrier, but the edge pixellation was considerably reduced.) I'mma try 560pF next once my next shipment of parts arrives, and see if that strikes a better balance.

    The non-synchronous clock coming from a direct crystal osc would also explain the HORRIBLE dot crawl I got when I tried feeding a KA2195D a colorburst clock directly from a crystal osc. Interestingly once I disconnected the colorburst clock, leaving the KA2195D to display a luma-only picture, and it actually didn't look half bad at all, I couldn't see any pixellation. I guess the KA2195D is *extremely* sensitive to colorburst clock noise and waveform.

    Also, before anyone thinks they can fix the KA2195D by changing caps, bear in mind that even after putting an excessive cap at C31 it will still be blurry as heck, the tint will be off and the pixellation doesn't completely go away. The only real fix for a system with a KA2195D is to replace the encoder - the KA2195D simply isn't a good fit for a Genesis due to the noisy colorburst clock, Sega should never have used it for this application.
    Last edited by Eep386; 08-17-2017 at 03:02 PM.

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
  •