SNES-Mini RGB Measurements

The data sheet states that the 800k has a 20% tolerance too.

 

Which is a possible range of 640000 to 1440000. Can tune that sucker with a scalpel and an ohm meter.

 

Hahaha agreed. I think using the circuit I posted with a 150 ohm as R3 and a 21 ohm as R2 would be OK, or 1k and a 143 ohm. This would negate the tolerance of the 800k.

 

Hm, so the 150 ohm resistor is the passive current sink (and converts DAC I-to-V). dc-coupling should be OK here-- dc-coupling is preferable anyway.

The voltage divider load should be as large as possible so as to not affect the DAC voltage, but I don't think using the internal 800k is a great idea, just an option. I'd go with a 8.2k series & 75k parallel PD, this would yield almost exactly 100 IRE. The divider input should attach before the series coupling capacitor in case anyone's confused, attaching after will screw up both the S-RGB and THS' biasing.

 

I'm not sure about the 150 ohm resistor. I would hope the 1-chip would take care of the DAC I-to-V conversion internally. The 150 ohm resistor might be required for the S-RGB chip. But, we don't have a datasheet for it.

I was looking at the BA6592F data sheet just to see if there was anything at all to compare. On the last page, they have the recommended circuit. It shows a 560 ohm resistor between the signal and GND placed before the 1uF cap. Maybe the 150 ohm is recommended for the S-RGB?

 

No, the resistor is certainly a current sink in that circuit, it's a very common arrangement for IC DACs. Basically the DACs are composed of hundreds of matched weak transistors acting as current sources. Without the pull-down ac-coupling wouldn't even be possible since the cap could never discharge. The encoder is effectively decoupled from the node by the series cap and its high input impedance (all of the encoder's input biasing is internal).

Decreasing the resistor alone would lower the signal level but could affect the linearity (probably negligibly). Might be something to try for people with hot-air stations since it's easier than inserting a divider in front of the THS. 133 ohms (1%) would presumably bring the signal to the desired 100 IRE, that's what I'd start with.

 

Ok, thanks for the clarification. I'm used to seeing op amps to convert current to voltage on DAC outputs, so I thought that would have been included in the 1-chip. Makes sense now. I'll post a photo later, but the resistors and caps are easy to get to, I can leave the 150 ohm in and connect off the front side of the cap to the voltage divider network.

I'll try replacing the 150 ohm with 133 first, to see if that makes a difference. If that works, then it would be even better. Thanks for your insight!

 

OK, so I got around to modding my SNES-mini to see if the RGB signals could be brought to normal specifications. I removed R6, R7, and R8 (150 ohm resistors) which, as Calpis said, are a passive current sink to convert the DAC current to a voltage level. The current produced by the DAC for a 0.8V (current highest intensity R, G, B signal) signal level is 5.33mA. To get 0.7V output, the resistance needed would be (0.7V / 5.33mA) = 131.33 ohm. For common resistor values, a 133 ohm resistor was used.

Here are the results, measured at the Multi-Out. 240p Test Suite, white screen used. These measurements were using the outputs on the S-RGB chip with a 75 ohm resistor in series. R, G, B waveforms in that order:



Values, as you can see, are ~ 1.44Vpp. With the 150 ohm resistors in (stock), I was getting 1.8Vpp.

Also, I measured the CSYNC on Pin 18 for reference:



I took some before and after photos, using my phone (didn't come out to great). I'll post them once I look at them on the computer and see if they are usable. I do notice a difference in the image. The colors actually do look more vibrant, since they aren't being clipped anymore.

I will post the parts I used from Digikey. The SMD resistors I used are a package size too big, but they worked. A smaller size would be better, I'll figure out the ones to use and post them. To be honest, if SMD soldering is hard for people, you could easily use axial through-hole resistors, and solder them to the SMD pads once the SMD resistors are removed. There is enough space under the board where they won't be a problem, just make sure the leads are isolated so they don't touch anything.

I plan on testing using RetroRGBs board too. Once I do that, I'll put up instructions on how to do it with the current board.

Big thanks to Calpis and Kel. Kel actually pointed out the resistors earlier in the thread, as his 1CHIP has 160 ohms.

 

Great work Ultron. I can't wait to try this out.

I had a hunch that changing the resistors would affect the voltage level but I just didn't understand the circuit enough to know why. It seems so obvious now after seeing your calculations above and Calpis's earlier tip that the 150 ohm resistor is the passive current sink (and converts DAC I-to-V).

I think that I was also a bit confused because thinking about it now the 1chip with 160 ohm resistors is even worse than the mini, not better.

I was thinking, would it not be easier just to connect some 1.1k ohm resistors in parallel with the 150 ohm resistors to achieve 132 ohms? That way there would be no need to remove those tiny SMD resistors, you could just connect some axial resistors from the via holes to GND just after the DAC.

 

The 1.1k going from the via holes to a GND point would work too. A much easier solution than replacing the SMD components.

 

Here's a cruddy comparison I did taking photos with my iPhone. I don't have a way of doing any video capture, but I think you can tell the slight difference. This is probably not a good test, as there isn't much on the screen. Some of the other photos came out horrible and kind of unusable for a comparison. The difference in person is much easier to see. The picture is slightly darker, and all the colors are more vibrant and less washed out. Using Link to the Past as an example, Link's magic boomerang looks completely different (easy to see, as it is mostly red).

Here's photos of Link at the Save selection screen. The one on the left was the basic RGB mod, with wires coming from the S-RGB chip with a 75 ohm resistor in series going to the Multi-Out. The one on the right is using the RGB mod described above, where I changed R6, R7, and R8 to 133 ohms to lower the voltage range of the RGB output signal closer to 0.7Vpp standard.

Left is the old mod, right is the new mod:



You can see there is less of a "washed out" effect with the colors.

If anyone wants to request photos of anything in particular, let me know.

 

Huh? Surely they can't both be the one on the left.

 

Good catch haha, thanks. I edited the post above

 

So to clarify, the new mod is to simply change the SNES onboard resistors R6, R7, and R8 to 133 ohm. Do you still need to amplify the signals using the THS7314 and the 75 Ohm resistors on the lines or is this not necessary any longer?

 

Yes the new mod is replacing the 150 with 133 ohm resistors, wiring up pins 20, 22, and 24 of the S-RGB chip to 1, 2 and 4 on the Multi-Out with 75 ohm resistors in series, and Pin 18 of S-RGB to pin 3 on Multi-Out.

As far as the 75 ohm resistors, you will need a 75 ohm resistor and 220uF capacitor in series between the output of the S-RGB and the TV or upscaler. If you have a cable that has both resistor and cap, the 75 ohm is not needed inside the console. If you have a cable that only has a cap (like mine), then the 75 ohm is needed in the console. If your cable has neither, then you will need to add both a 75 ohm resistor, and a 220uF capacitor in series between the S-RGB and the Multi-Out pins. The THS7314 is and was never needed.

Some people believe that bypassing the S-RGB chip and using a THS7314 helps eliminate the "bright white band" seen on dark screens. So some have used it in the 1-chip and Minis, wiring up the inputs for the THS7314 to the output lines of the raw RGB signal coming from the CPU/PPU (I've never seen this problem on my Mini BTW).

To make it easier, instead of changing the resistors, you can wire a 1.1k ohm in parallel with the 150 ohm resistors. One side of each of the 150 ohm is connected to the vias that are used in the THS7314 mod, the other side to GND. Add a 1.1k to each of the vias, tie the other ends of the resistors together, and connect them to a GND point.

Here's the vias seen from the top of the board:



If someone wanted to use the THS7314 amp, the easiest way would be to leave the 150 ohm resistors, wire up the 1.1k resistors like above. Connect 3 wires on the via side of the 1.1k resistors, and connect them to the THS7314 inputs. This would use the THS7314 in DC Coupled input mode, and should work OK. I am going to try this out later this week to make sure it works properly, then I'll post install pics of both mods.


Can someone check the 1-Chip's R6, R7, and R8 values for me? If they are 160 ohms, like kel said, then instead of using 1.1k resistors in parallel, you can use 750 ohm resistors to get the same results.

 

It just so happened I had a 1chip I was just putting back together in my hands right this second.



If you cant make it out, code is 161, so 160 ohm.

My next question is, whats in the cable you are using? PAL cables (on a PAL 1chip) may have differing results. Also IIRC, the PAL phat snes has slightly differing output circuit. The cables have the 75ohm resistor as a shunt.

 

Thanks! It looks like the 750 ohm resistors can be added to the vias right next to the resistors. Can you check and see what side of the resistors is connected to GND?

 

Updated the picture. Also check my edit at the bottom.

 

I'm using a NTSC Mini. The cable only has 220uF electrolytic caps on the RGB lines and the cable is shielded, bought from retro_console_accessories on eBay. Good point, should have mentioned that.

The ends of the resistors with the black line are the GND side, I take it. Do the other side of the resistors connected to those vias?

 

Black line = ground.
Other side of resistors = They connect to the vias below the black line