SNES-Mini RGB Measurements

Discussion in 'Modding and Hacking - Consoles and Electronics' started by Ultron, Aug 6, 2014.

  1. bigshowpl

    bigshowpl Member

    Joined:
    May 22, 2012
    Messages:
    19
    Likes Received:
    4
    Thanks Ultron!

    Going to make things more complicated now, I looked into 0.7 vs 0.714 and as i'm using a Super Famicom, on a natively PAL TV (that was made in Japan haha) , i'm guessing 0.7v (after the 2x75 ohm taken into account) is what to shoot for as it's only NTSC-U that has the blacks from 7.5 IRE

    I found a few threads from Voultar on other forums where he's determined the gain of the S-RGB encoder is not 6dB (2:1) but more like 6.5dB/6.6dB. Based on that, i'd be shooting to lower the input to the S-RGB to more like 0.65v to get 1.4v out, so the closest resistor would be 560 Ohms (or 510, depending which figures you go with.)

    I guess the only accurate way to do it would be to scope every SNES PPU output and S-RGB output and tailor it individually depending on the system....
     
    bond.san likes this.
  2. bigshowpl

    bigshowpl Member

    Joined:
    May 22, 2012
    Messages:
    19
    Likes Received:
    4
    Actually I bet you're supposed to attenuate the output rather than reduce the input.... more thinking required
     
  3. Ultron

    Ultron Spirited Member

    Joined:
    Jul 6, 2014
    Messages:
    110
    Likes Received:
    15
    Can you send me the link where Voultar talks about this? I remember reading something about the linearity of the DAC in the S-CPUN. I don't remember reading about the gain.
     
  4. bigshowpl

    bigshowpl Member

    Joined:
    May 22, 2012
    Messages:
    19
    Likes Received:
    4
  5. Ultron

    Ultron Spirited Member

    Joined:
    Jul 6, 2014
    Messages:
    110
    Likes Received:
    15
    Thanks I think I found another post in the chumps forum. I understand what he is saying, but the output of the S-CPUN and the I-to-V conversion is wrong, and should be corrected. That's what the resistors are doing, correcting the output. With this corrected, you can use any buffer amp you choose and the output will be correct.

    From what I gather, Voultar shows that the resistance needs to be increased, that is incorrect. It needs to be decreased. The current value of 150 ohms (160 in the 1-CHIP) is too high. Ohms Law: V (voltage) = I (current) * R (resistance). The current(I) is output from the DAC internal the S-CPUN and is a current source. This current only changes depending on the R, G, and B values (each line has its own current source). The 150 ohm resistor is used as a simple I-to-V converter. We want an output so that a 255 (max) value is = 0.714V. With a resistance of 150 ohms, the value is ~0.8V.

    Using Ohm's Law, I = V/R = 0.8V / 150 ohms = 5.33mA.
    The value of current at 255 brightness is 5.33mA.
    R = V/I = 0.714V / 5.33mA = 133.8 ohms.

    To make the mod easier, instead of replacing the SMD resistor, you just add one in parallel. In this case, 1.2k in parallel gets us to 133.33 ohms.

    I haven't done measurements on a 1-CHIP, so I don't know what the output voltage of the circuit is. But I do know that they have a resistor of 160 ohms. If using the same S-CPUN which I'm pretty sure they do, I would assume the output current will be the same. So, you can use the same value of 133.8 ohms as the I-to-V converter. Someone in this thread made 1-CHIP measurements, but the photos are no longer hosted.

    1-CHIP would be:
    For 0.714V: (160 || 800) = 133.33 ohms. But 800 ohms are hard to find in 1%. Use something closer. 806 ohm would work.
    If you are looking for 0.7V, 0.7V / 5.33mA = 131.33 ohms. A 750 ohm resistor on a 1-CHIP (160 || 750) = 131.86 ohms.

    Could there be variation between systems. Of course! But I expect it to be small. Changing the resistor value will give much better image quality over the setup used by Nintendo, though.
     
  6. Sp33dFr34k

    Sp33dFr34k Spirited Member

    Joined:
    Jul 13, 2015
    Messages:
    141
    Likes Received:
    32
    I was just about to buy a THS amp board from Voultar for my SNES Mini, but all this tech talk confuses me slightly. Are you guys saying there's going to be (overscan) issues with the board as it is?
     
  7. bigshowpl

    bigshowpl Member

    Joined:
    May 22, 2012
    Messages:
    19
    Likes Received:
    4
    Hey Speedfreak, no this is just about colours being a little washed out and being corrected with some resistors.

    The bypass board is using a 6dB amp so the usual resistor values used to correct brightness are fine.

    Just debating if they should be different for those using the existing SRGB A In the snes itself, if because of it's higher gain should we decrease the input below .7v or not.
     
  8. Sp33dFr34k

    Sp33dFr34k Spirited Member

    Joined:
    Jul 13, 2015
    Messages:
    141
    Likes Received:
    32
    Thanks for the clarification @bigshowpl. I was planning on using Voultars PCB as that seems to be a very nice and clean solution, so there's no risk here then :)
     
  9. voultar

    voultar Rapidly Rising Member

    Joined:
    Sep 4, 2013
    Messages:
    76
    Likes Received:
    63
    The 1CHIP ASIC has a variably wild output swing on the analog outputs. It doesn't output 800mV statefully across all systems. I've measured systems outputting as high as 880mVP-P, other systems have measured as low as 670mV-P-P. Let's first examine the fixed gain output of the S-RGB encoder. We'll take a look at scope plots that I've captured for analysis.

    We'll start by examining the charcteristics of the S-RGB encoder found in both SNES 1CHIP (Phat) models as well as the SNES-Mini.

    This is a common measurement found on the analog video outputs of the 1CHIP:

    [​IMG]

    Now let's take a look at the outputs of the S-RGB encoder.

    [​IMG]

    1.8vPP under peak luma conditions? That's not a 6dB driver. That's a 6.5dB driver. For everyone who's wanting to attenuate while using the stock, on-board encoder. Treating the S-RGB encoder like a THS-73XX driver won't get you where you're wanting to go.

    [​IMG]

    So having said that, let's get into using typical 6db drivers for video.

    As I mentioned earlier, the 1CHIP ASIC has a fairly wild output swing. And while Ultron is certainly correct in his network formula, it's not correct to assume the output swing is minimally different. It isn't. Like I mentioned at the top, that bastard is really all over the place.

    Let's look at some figures from various SNES systems:

    SNES Mini with 1K2 load for the THS73XX:

    [​IMG]

    Another SNES Mini with 1K8Ω, look at the difference.. These are obviously two different systems, mind you. But let this be an example of the wide swing that we're dealing with.

    [​IMG]

    Ultron is absolutely right. But the problem is that we're assuming that we have a co-efficient that never changes. In other words, we are assuming that variances in the RGB output from the 1CHIP ASIC are small and negligible when in reality we have a fairly wide swing to contend with. So what did I do? Well, I simply chose the middle ground. I felt that it was important to choose a safe, healthy safe place that would accommodate all of these variances. You don't want to go too dark, and you don't want to over-saturate at 255,255,255 (peak luma). My figures were extrapolated from sampling several, several systems in an effort to average them all out. It took a long time!

    To sum it up, I would completely agree with Ultron if we were dealing with a stateful consistent analog output from the ASIC. Unfortunately, things get pretty wild in there!

    As a bonus, I also analyzed the encoded format output of the S-RGB/BAXXX encoder to see how the amplitude/voltage was affected from attenuation.

    [​IMG]
     
    Last edited: Aug 28, 2017
  10. bigshowpl

    bigshowpl Member

    Joined:
    May 22, 2012
    Messages:
    19
    Likes Received:
    4
    Thanks Voultar - looking at the schematic for the SNES Expansion board you released for those of us using the onboard encoder, you've specced up 750 Ohm/1K2 so is that what you chose as a good middle ground?

    I've no scope to measure my SFC 1CHIP-01 console, but on the 240p test suite colour bars test, the last 3 bars of Green and Blue are just one colour (no matter what I do on the TV), wheras for red it's just the last 2 bars. I think the best I can hope for without getting a scope on my system is to make it so those bars are distinguishable!

    I was wondering if the output of the SRGB-A is going to be up as high as 1.58v with 0.7v going in, should we be reducing the input to <0.7v or would that somehow compress the range?
     
  11. voultar

    voultar Rapidly Rising Member

    Joined:
    Sep 4, 2013
    Messages:
    76
    Likes Received:
    63
    Yes, that's meant to be used more of a range. I'd honestly recommend anything between 700-890Ω for the majority of cases. 1K2Ω is more than likely going to saturate the video at 255,255,255.

    It's better to go darker than it is to go lighter. Darkness can be handled. But when you over-saturate, it's on like Donkey Kong!
     
  12. rama

    rama Gutsy Member

    Joined:
    Dec 17, 2015
    Messages:
    477
    Likes Received:
    112
    @voultar
    You probably know about the Vcc > resistor > AVcc pin solution. It reduces the analog voltage for the 3 RGB outputs and fixes all the ghosting, in addition to reducing the brightness to normal levels.
    Have you ever attempted that and measured the effect?
    If the S-CPUN really has such a wide variance, then this is probably the reason why some people see ghosting and some don't.
     
  13. bigshowpl

    bigshowpl Member

    Joined:
    May 22, 2012
    Messages:
    19
    Likes Received:
    4
    Do you recommend a different range for Phat 1CHIP vs Mini as the Phat has a 160ohm resistor vs 150ohm in the mini? Or is that range to cover both?
     
  14. retronerd

    retronerd Spirited Member

    Joined:
    May 6, 2012
    Messages:
    156
    Likes Received:
    5
    Is there any schematics on the 1chip S-RGB chip need to find where the CompositeSync goes, not csync.
    Thanks
     
  15. Bad_Ad84

    Bad_Ad84 The Tick

    Joined:
    May 26, 2011
    Messages:
    8,605
    Likes Received:
    1,376
    Composite sync is csync.
     
  16. Paar

    Paar Rising Member

    Joined:
    Jul 24, 2013
    Messages:
    57
    Likes Received:
    11
    Doesn't CSYNC stand for "clear sync" or something like that? I was pretty sure that composite sync and csync are two different things with csync being superior to the other one.
     
  17. Bearking

    Bearking Konsolkongen

    Joined:
    Aug 2, 2010
    Messages:
    856
    Likes Received:
    100
    Pretty sure it's CSYNC = Composite sync, also sometimes referred to as "clean sync", and CVIDEO = Composite video, which as the name implies uses the composite video signal as sync.
     
  18. Bad_Ad84

    Bad_Ad84 The Tick

    Joined:
    May 26, 2011
    Messages:
    8,605
    Likes Received:
    1,376
    No, it's composite sync. Composite video is the one people try to avoid as sync.
     
    Taijigamer2 likes this.
  19. Taijigamer2

    Taijigamer2 Gutsy Member

    Joined:
    Jun 29, 2015
    Messages:
    498
    Likes Received:
    200
  20. Paar

    Paar Rising Member

    Joined:
    Jul 24, 2013
    Messages:
    57
    Likes Received:
    11
    Thanks. It makes sense now.
     

Share This Page