SNES-Mini RGB Measurements

Hey Ultron, I happen to have a SNS-1Chip, SNS-CPU-GPM-01, and SNN-CPU-01 (SNES mini) all cracked open at the moment. I can measure CSYNCIN pin 7 and CSYNCOUT pin 18 for all encoders and report back.

 

Awesome!! I'm interested in the output going to the CSYNC pin coming from the transistor amp that was highlighted earlier in this thread. The 1CHIP and the GPM might use the same amp. So I guess you can check pin 3 on multi-out and the output of CSYNC coming from the PPU (going to pin 7 on the S-RGB). I expect pin 7 to be the same as what I measured earlier in the thread.

 

Welp, this is embarrassing. In the process of beginning to carry out my measurements on the 1CHIP, I may have blown the 1.5 amp pico fuse I was replacing the regulator so that I could use the SNES without a heat sink so that it wouldn't get in the way since it obstructs access to the encoder. So I'll have order some new ones and replace that first.

 

Quote from here: https://www.assemblergames.com/foru...ard-3-Wire-Mod&p=798612&viewfull=1#post798612

Well - I can add this kind of holes in the design once it has been finished. At the moment I use pads with holes directly within the pads in a lot of my desings... So one has the choice between using a connector or a direct wirering.


Quote from here: https://www.assemblergames.com/foru...ard-3-Wire-Mod&p=798391&viewfull=1#post798391

I've done two designs I want to test.
- One with the additional pull-downs to reduce R6-R8-values. I do not do something with jumpers to hold the design quite small. At a PAL-SNES you need other resistor vlaues. So my option is to simply adjust the values according to the needs
- Next to that solution I have a design which uses from the S-CPUN decoupled RGB-signals with a DC-Bias to have a proper working point for the THS7314. Thankfully the caps C6-C8 together with the S-RGB does the job quite well. So I just have to the signal directly from pin 1, 3 and 5 of the S-RGB for the amp

After modding I want to have a RGB cable without any additional components in the RGB-lines; neither caps nor pull-downs! So one have simply to remove them

Once everything is working the final step would be to replace the THS7314 by the THS7374 to add the /CS-signal from the S-CPUN to the amplification circuit. This is just for for the 1Chip-03 and Mini/Jr.

 

Measurements were taken with Agilent 54622D using a 10074C 10:1 probe and using a ground bayonet. I used the 240p test suite ROM with the White and RGB Test Screens test pattern loaded while displaying the black pattern. The specimens under measurement:


1CHIP #1
Serial No: UN310740788
Board Rev: SNS-CPU-1CHIP-01
CPU: S-CPUN A (RF5A122)
Encoder: S-RGB A (BA6596F)


Measurements taken at the encoder on pin 7 - CSYNC IN


Measurements taken at the encoder on pin 18 - CSYNC OUT



1CHIP #2
Serial No: UN308308600
Board Rev: SNS-CPU-1CHIP-01
CPU: S-CPUN A (RF5A122)
Encoder: S-RGB (NO CHIP MODEL)



Measurements taken at the encoder on pin 7 - CSYNC IN


Measurements taken at the encoder on pin 18 - CSYNC OUT


MINI #1
Serial No: UC305346582
Board Rev: SNN-CPU-01
CPU: S-CPUN A (RF5A122)
Encoder: S-RGB A (NO CHIP MODEL)



Measurements taken at the encoder on pin 7 - CSYNC IN


Measurements taken at the encoder on pin 18 - CSYNC OUT


Is this what you wanted to see? My voltage levels don't look quite like yours though. Were you measuring after the amplifier and at the multi-out pin directly? Or maybe you were adjusting your Y cursors manually eyeballing it? I'm measuring directly at the pins with no amplifier in place and sync NOT extended from the encoder to the multi-out at pin 3.

I was unable to measure the SNS-CPU-GPM-01 because I stole its pico-fuse to repair the 1CHIP. My replacement fuses should arrive tomorrow and I can measure it over the weekend and update this post. One thing though, the SNS-CPU-GPM-01 has the BA6594AF encoder and the datasheet shows only a SYNC IN pin. No SYNC OUT. Does that sound right to you?

As an aside - my waveforms looked a bit funny while displaying any of the color patterns, especially the white pattern. Voltage levels were the same as with the black pattern though.

EDIT: I have a SNS-CPU-1CHIP-02 a SNS-CPU-1CHIP-03 available to measure as well if it makes any difference.

 

GPM #1
Serial No: UN20406813
Board Rev: SNS-CPU-GPM-01
PPU: S-PPU2 B (5C78-03)
Encoder: S-ENC B (BA6594AF)



Measurements taken at the encoder on pin 8 - CSYNC IN

 

Measurements look good. Thanks! One thing though, I don't think the 1CHiP uses the CSYNC OUT of the S-RGB, it has it's own buffer circuit. Could you measure at pin 3 of the multi-out also?

i believe I measured CSYNC at pin 3 and 18 on my mini.

 

This would be fine as well! Holes behind the pads or on the pads... As long as there is a way of neatly soldering on right angle headers It would be a very thoughtful addition. Thanks borti.

 

As per post #28 of this thread:


"The current produced by the DAC for a 0.8V signal level is 5.33mA. To get 0.7V output, the resistance needed would be (0.7V / 5.33mA) = 131.33 ohm."


It was then suggested to install resistors in parallel with the 150 ohm input resistors to make installation easier. Equivalence resistance formulae for resistors in parale is:

Rtotal = (R1 * R2) / (R1 + R2)

Given R1 = 150ohm and Rtotal = 131.33ohm, best R2 will be 1055.14 ohm. So one can actually use 1.1kohm or the much easier to fine 1kohm resistor.

 

Yes, this was stated in post # 35:

 

Yeah. But if one doesn't have 1.1k I'm saying 1k will also work according to the math. Right?

 

The math says to use exactly 1.25k should 150 ohms yield 0.8 Vpp.

1 / ((100 * 1.0 Vpp / 140 IRE) / (0.8 Vpp / 150 ohms)) = 1 / 150 + 1 / x

In this case it's better to round down, so I'd use 1.2k 1%. (Of course it's even better to replace the 150 ohm loads since they're probably 5%.)

 

Ahh good point. I should have used 0.714V in the calculation instead of 0.7V.

I think the 150 ohm resistors are actually 1% tolerance. The ones I measured were all within 1%.

 

Surprisingly, I measured the same RGB-amplitudes at the S-CPUN of a 'normal' 1Chip-SNES (PAL-Version) as you did, Ultron. (I have to look on my computer for the curves: I have stored them somewhere...)
This surprises me, because we have 160Ohm pull-down resistors instead of 150Ohm. However, by keeping in mind that Nintendo has been changed the S-CPUN for the mini, this could be the case. This leads me to the conclusion that putting 750Ohm in parallel to R6 - R8 could be incorrect.

I decided to use a high-ohm voltage divider prior to the THS7314 instead of simple pull-downs. I calculated with 0.77Vpp I measured at red (not with 0.8Vpp of green and blue) and used 750Ohm in serial and 7.5kOhm to ground afterwards. However, with the internal 800kOhm +/- 20% pull-down inside the THS7314 I have something close below 0.7Vpp at R and something close above 0.7Vpp at G/B. One could also use another divider for G/B.
By using the voltage divider, I also do not change the input for the S-RGB (in case somebody wants to use composite video sometimes... or due to other reasons I would never understand ).
The second effect is that I can use the same values for every system

Here is my beta-board installed in a 1Chip-02 PAL board.


The /CSYNC-pad is just for testing and goes to pin 3 of the multi-av. A proper RGB-cable has no components at the RGB-line: neither caps nor resistors. For standard PAL-SNES-cable one could solder 75Ohm resistors in parallel to the other 75Ohm resistors at the THS7314 output.

My next steps for the final board:
- remove the /CS pad
- put solder jumper to short the caps on the board -> use 'standard' NTSC-SNES-cables
- design a board with a THS7374 to also buffer the /CS-signal coming from the S-CPUN

 

It's funny you should say that. I was just thinking how this mod hasn't seemed to work as good for my 1CHIP SFC as it has for my SFC Jr. On my SFC Jr I can see the difference clearly but on my 1CHIP not so much. I'll check on my SD2SNES menu later to see if the PPU versions are any different between the two consoles. It could be that the S-CPUN has been updated between consoles as you suggested.

 

I wanted to put a THS7314 amp in my SNES mini with correct voltages in but in the vein of treading lightly, didn't want to tie the S-RGB A pins to ground with resistors that would impact the composite output.

I designed and printed a board similar to the RetroRGB SNES board but with a voltage divider with a very high resistance pulldown. I also added S-Video to the multi-out connector since, why not. I made a voltage divider with a 10.7k ohm resistor and a 75k pulling down to ground. By my calculations that should have been correct, but I forgot to factor in the 800M ohm pull downs in the THS7314.

Here's what it looks like installed:


Here's what it looks like after I cut the traces and swapped the chroma and luma s-video lines since I ran them to the wrong pins of the multi-out connector (DOH!)

In this configuration RGB, s-video, and composite all work properly.

Here's the green input to the THS7314, 688mV peak to peak (just shy of what I was aiming for.)


Here's the green output of the amp:


A little bit higher than 1.4V peak to peak.

The SNES output looks noticeably better than the simplest SNES mini RGB hack, wiring the S-RGB A chip directly to the multi-out with some 75 ohm resistors.

I took pictures of the SNES video test app's color gradient bars, before/after, but my digital camera doesn't capture CRT dynamic range the way our eyes do. With the simple SRGBA mod I couldn't tell the difference between the 4 rightmost bars of each individual color, but with the THS amp and voltage drop I can.

If anyone wants to print the board it's on upverter: https://upverter.com/akramer/9c424d25d57b40d3/SNES-mini-RGB-out-board-ws-video/
I've fixed the board so Y and C are connected to the correct multi-out pints now.

Recalculating with the 800k pull down in parallel with my circuit's pulldown, the correct resistance value is ~82.5k. Next time I order some parts I will get some 0805 82.5k resistors and see if they work nicely.

 

I put another one of these boards together tonight. If someone is interested in it, I'll mail it to you if you'll install it and let me know what you think of it.

 

Here my last design for the SNES Mini, SFC Jr. and SNS-CPU-1Chip-03. It can be also used for other 1Chip-designs if Pin 3 at the MultiAV is freed by the /CSYNC (NTSC) or +12V (PAL), which I've also been done.



The design uses a THS7374 to buffer R, G, B and also /CSYNC. I used the signals directly from the S-CPUN. I used a simple voltage divider using resistors for all components. I will upload the design quite soon to my GitHub.

 

I'm about to attempt the RetroRGB SNES mini mod with 7314.
I understand it is not perfect, it seems all that is happening with the evolutions is to recondition the input signal so that at the output we have correct values (and there's 2 RGB inputs possibilities as I see it, namely before and after the onboard encoder).

I did not check yet, but if the 7314 is not clipping, wouldn't is be possible to have a voltage divider at the output?
Namely change the 75ohm to whatever it is needed so series with the TV 75 ohm one obtains what is needed, as it is all 75ohm drivers are actually relying on output voltage dividers (75 ohm at the TV and 75 ohm right at the out of the ICs, so already the signal is scaled by a factor 2)

Again, and I have no scope to measure, my suggestion is assuming that the 7314 is not clipping and it is not entering non-linear zones (distortions) at which point obviously only an input reconditioning would work correctly.

 

I really don't know anything about clipping or non-linear zones but wouldn't what you are proposing cause an impedance mismatch anyway?