SECURITY NOTICE: Please read the security notice in the site help forum immediately.

SNES-Mini RGB Measurements

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

  1. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8
    EDIT: On the SNES-Mini, to achieve the correct RGB signal levels, resistors must be added to the RGB outputs of the PPU. These 3 resistors (one for each color) will go from the PPU output to GND, to change the output of the PPU's current-to-voltage divider. Please see Post # 45 of this thread for info on where to install the resistors. Installing the resistors will correct the RGB voltage levels before they go into the S-RGB IC (or if you prefer, your THS7314 Amp), allowing them to be buffered at the correct levels. With this, you will not need to change the 75 ohm resistance on the Amp's output to correct brightness saturation, thus preventing impedance mismatch with your TV/upscaler unit.

    -------------------------------------------------------------------------------------------------------------------
    Please see this thread - SNES-MINI-RGB-Amplifier-Chip-Mod-or-Standard-3-Wire-Mod

    It has come up in conversation in the above thread that the THS7314 RGB amp circuit on the RetroRGB page and used in numerous other consoles (i.e. N64 RGB mod) is not the most "ideal" circuit. The data sheet for the THS7314 has a few examples of proper circuit design using the IC, which none of these examples resemble the circuit that has been used.

    Many people have noticed that the when using the RGB amp circuit, the video is a lot brighter than normal. To fix this, they have replaced the 75 ohm resistors with 100 ohm, or others. This has lowered the brightness and made the image "acceptable". The problem here is that the 75 ohm resistor is used for impedance matching with the TV set. The standard these days is 75 ohms (some older monitors might have a switch to change impedance between 50 and 75 ohms). By changing the impedance at the video output of the SNES to 100 ohms, this could cause problems with the video, such as "ghosting", video artifacts and color problems. Some TVs might be able to handle this mismatch, but it would be preferred if we can design the circuit properly to minimize problems from setup to setup.

    The setup:

    SNES-Mini (SNN-CPU-01) and Rigol 1052E Oscilloscope

    SNES-mini.jpg Rigol.jpg

    RGB Measurements taken from vias right off the 1-chip pins

    CPU.jpg

    I also took measurements at the Multi-Out. Currently, I have the RGB lines connected to the RGB out pins of the S-RGB IC with 75 ohm resistors on each line going to the Multi-Out pins. I also connected the CSync pin of the S-RGB to the Multi-Out.

    The measurements were taken with a 10:1 probe (in 10x mode on the oscilloscope and probe) and in DC Coupled mode. I used the 240p Test Suite's White and RGB screen mode.


    The measurements at the RGB pins (R, G, B):

    RGB-CPU.jpg

    R is slightly lower in amplitude(?). Amplitude ~0.8mVpp.

    Measurements for RGB at the Multi-Out (R, G, B):

    RGB-Multi.jpg

    Amplitude is ~1.6Vpp. Strangely, R's amplitude is the same as G and B, where in the signals off the CPU, it was lower.

    Black level at R pin near CPU, and R Multi-Out (Cursor B shows offset of signal). Last is CSync at Pin 3 of Multi-Out (Run from S-RGB IC):

    Blk-Sync.jpg
     
    Last edited: Aug 24, 2015
  2. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8
    Just so people understand, the output is supposed to be double the input. With the 75 ohm impedance on the output, and the 75 ohm impedance on the TV input, the input going into the TVs (or scaler's) video circuit will be half of what is measured on the Multi-Out. With a 100 ohm impedance on the output, this would cause a further reduction in the signal, and thus wouldn't be the true output from the 1-chip.
     
  3. xmog123x

    xmog123x Peppy Member

    Joined:
    May 7, 2010
    Messages:
    359
    Likes Received:
    7
    you could've made 2ch measurements
     
  4. keropi

    keropi Familiar Face

    Joined:
    Feb 2, 2011
    Messages:
    1,012
    Likes Received:
    43
    Finally someone measured things, now hopefully knowledgeable people can use the data and design a better amp circuit.
    Kudos on making the measurements Ultron!
     
  5. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8
    What would you have liked to see on the 2nd channel? Looks better with each signal separated.
     
  6. APE

    APE Site Supporter 2015

    Joined:
    Dec 5, 2005
    Messages:
    6,360
    Likes Received:
    93
    I need to get a scope for myself to do similar measurements and to help calibrate lasers better. Kudos!
     
  7. kel

    kel Rapidly Rising Member

    Joined:
    Aug 13, 2012
    Messages:
    89
    Likes Received:
    4
    Thanks for taking these measurements Ultron. I wanted to do this myself on this console and others but unfortunately for me the oscilloscope that I ordered from ebay seems to be non existant so now I am left awaiting the results of a paypal case instead :(

    So am I correct in thinking that from these measurements that the input to the S-RGB chip is too high and should be 0.7Vpp? Could this be due to the resistors from the RGB lines to GND before the S-RGB chip being 150 ohms as instead of 160 ohms like on the phat 1CHIP? I've never heard any complaints about the brightness on the phat 1CHIP. Unfortunately without a scope to check my phat 1CHIP I'm left guessing.
     
    Last edited: Aug 6, 2014
  8. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8

    Hey kel,

    I don't think the resistors and caps have anything to do with signal level. They are there for coupling to the RGB chip. The 150 ohm resistors go between the signal and GND. The caps are 100nF and are in series with the signal to the pin on the S-RGB IC. I blew the one on the G line when testing, have to replace it now (might steal one from my N64 for now).

    I'm going to test the signal at the inputs of the S-RGB, but it shouldn't be any different. A circuit to reduce the voltage (a voltage divider) needs to be designed to be placed on the input, along with a capacitor for coupling to the THS7314 IC. After I make more measurements, I will post a circuit that can be used. I also will hook up the current THS7314 circuit and get measurements from that too.
     
    Last edited: Aug 21, 2014
  9. Bad_Ad84

    Bad_Ad84 Keyboard Error: Press F1 to Continue

    Joined:
    May 26, 2011
    Messages:
    7,508
    Likes Received:
    546
    Please measure csync from pin 18 of the srgb chip, not pin 7.

    7 is csync in, 18 is csync out.
     
  10. niall

    niall Active Member

    Joined:
    May 6, 2012
    Messages:
    26
    Likes Received:
    0
  11. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8
    I'll measure pin 18 tomorrow. I don't expect to see a difference, the levels on pin 7 are TTL and work fine for Sync signal.


    Nice design, but I would never use potentiometers in this kind of setup. There is too much precision involved, pots can get noisy over time, and everyone would need an oscilloscope to adjust them properly. The signals coming out of the 1-chip CPU/PPU are stable and won't change. Pots would be good for testing, but high precision resistors should be used for final design.
     
    Last edited: Aug 6, 2014
  12. kel

    kel Rapidly Rising Member

    Joined:
    Aug 13, 2012
    Messages:
    89
    Likes Received:
    4
    Another reason that made me think that the value of the resistors to GND mattered is that the early N64 also has resistors to GND on the RGB lines between the DAC and encoder chip. In the link below viletim explains that the The amplitude of the RGB signals can be changed by varying the load resistors to GND.

    http://gamesx.com/wiki/doku.php?id=av:n64rgb-ntsc

    Maybe I'm just putting two and two together and making five. Without a way to measure I'm just relying on guess work and bits of info from similar circuits like the N64 and phat 1CHIP.
     
    Last edited: Aug 6, 2014
  13. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8

    In that application, he tried to connect the TV directly to the output of the VDC-NUS after the 110 ohm resistor to GND. This had the 110 ohm in parallel with the 75 ohm (TV impedance). I think it looks like he tried this first before using the amp circuit.

    The 150 ohm resistor in this circuit is on the input side to the amp, so it's a different situation.
     
  14. borti4938

    borti4938 Spirited Member

    Joined:
    May 8, 2014
    Messages:
    166
    Likes Received:
    38
    Nice to see you starting with measurements, Ultron :) KUDOS! Unfortunately, I don't have the equipment doin' similar measurements :(
     
  15. Bad_Ad84

    Bad_Ad84 Keyboard Error: Press F1 to Continue

    Joined:
    May 26, 2011
    Messages:
    7,508
    Likes Received:
    546
    It shouldnt be used directly for the TV though. It should be buffered.

    If we are going to the hassle of making the R,G,B values correct, we should use correct csync too.
     
    Last edited: Aug 7, 2014
  16. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8
    I think I figured out a circuit design that will work. I ordered parts and will test it when I get them in.
     
  17. niall

    niall Active Member

    Joined:
    May 6, 2012
    Messages:
    26
    Likes Received:
    0
    Sorry you've misunderstood. The first link has the design and gerber files to etch your own PCBs. The second link has boards with fixed value resistors for sale (100ohm) and you can find these boards on a few forums and made by a few different guys now.

    However my point (sorry for not making it clearer) is: why not build upon this yourself, rather than starting from scratch? :)
     
  18. Bad_Ad84

    Bad_Ad84 Keyboard Error: Press F1 to Continue

    Joined:
    May 26, 2011
    Messages:
    7,508
    Likes Received:
    546
    Because the circuits in those links are wrong.

    We already know about them, have been using them etc. We are trying to make the circuit correct. Its also pretty simple, there is no need to build upon old work
     
  19. Calpis

    Calpis Champion of the Forum

    Joined:
    Mar 13, 2004
    Messages:
    5,906
    Likes Received:
    19
    blah.PNG

    R2 pull-up is only necessary if the offset needs to be shifted down, in which case R1 will need to be lowered.

    Without it:
    (0.8V / (R1 + 800k)) * 800k = 0.7V

    R1 = 110k

    For better accuracy use an external pull-down, 75k ohms or less so it will be minimally affected by the low-tolerance 800k in parallel.


    Something that also might come into consideration is the DAC's construction. It would be very helpful if someone had an accurate schematic of the full signal path. Typically these old DAC outputs can only sink or source current, so a current sink/source load is necessary to get linear operation (or measurements). Maybe this is internal to the chip, maybe not. If not, the measurements will need to be scrutinized.
     
    Last edited: Aug 9, 2014
  20. Ultron

    Ultron Rapidly Rising Member

    Joined:
    Jul 6, 2014
    Messages:
    97
    Likes Received:
    8
    I'm waffling between using the DC Coupled input or the AC Bias input method. AC Bias would be the "proper" way, but since the RGB outs from the CPU/PPU are already 0V offset (assuming the measurements above are correct), DC Coupled could be used, since the THS7314 applies a bias shift internally. DC Coupled would also cut down on components needed.

    This is the current setup on the SNES-mini:

    SNES-Mini_RGB.png

    Measurements in the first post were taken at the CPU/PPU pins.

    This is the DC Coupled circuit I'm looking at (note, the diodes and 800k are internal to the THS7314):

    New_Circuit.png

    R2 needs to be .143 the value of R3 to get 0.7Vpp from the raw signal. Calpis, you think using the 800k with R2 is a better way for the voltage divider? I didn't even think of using that.
     
    Last edited: Aug 9, 2014

Share This Page