Fixing a Super Nintendo That Won’t Output Color

I have a Super Nintendo
Entertainment System, which normally works great. Today I found that when I started it up, I wasn’t getting any color out of it. The screen came up black and white. In the image, you can see that I'm not getting the proper colorful display I expect. However, interestingly, it's not purely black and white. Especially if you look in the lower right, just past the word Nintendo, you'll see that some red seems to be showing up. How strange! Let's talk through how I fixed this.

Super Nintendo (of course)

A game

Gamebit screwdriver (for opening console)

Phillips screwdriver (same)

Some way to measure a signal (ideally oscilloscope, I used a Saleae Digital Logic Analyser)

Before we jump into repairing this, let's talk about how old-style analog video signals work.

Essentially, there are two components. Luma, and chroma. The luma defines luminance (brightness) of the display at each location on the screen, and the chroma defines the color. You could almost think of luma as being a magnitude and chroma being a direction (which direction on the color wheel the luma is pointing). This structure is used because when analog video standards were being defined, all the TV’s were black and white. Once color TV was invented, they wanted to be able to transmit a color signal in such a way that, if you were still using an old TV, you could still receive the signal, even though you wouldn’t be able to see the color portion.

So to be clear here - you're probably familiar with RGB signals (red, green, blue) to define the color of pixels on a screen. We're in the analog world here, so we don't use those individual components. Instead, we have luma, and chroma to deal with.

The Super Nintendo outputs using an RCA Composite cable. That's the little yellow circular plug you're used to seeing on TV's and consoles and things of that nature. It carries luma and chroma all in one. Take a look at the picture above for an example of that. More modern devices that want to get higher quality use what's called Component cables, where you use 3 separate cables to carry the luma and chroma data. Basically by separating out the individual signals, you reduce the amount they impact each other, and you get a better video signal on the other end. But we're in composite land, where we're using one cable for everything!

Basically the chroma signal is more high-tech, and relies on a strict clock signal. If the chroma signal isn't coming at exactly the right time, the system may fail to recognize it. On the other hand, the luma signal is simple and easy, and usually the receiving device can tolerate a little bit of drift in the timing of the luma signal. It doesn't have to have perfect synchronization. If the signal's timing is a bit off, then the luma might come through, while the chroma fails to be recognized by the TV. What that means is you still get your lightness information, but the color is gone. And just like that - a black and white image.

So in this case, I'm suspecting that as the cause of our issues. But there's only one way to be sure - we'll have to open up the Super Nintendo and check on its clock lines.

Taking apart a Super Nintendo isn't too hard. It's just a few screws, nothing fancy like in a lot of today's electronics. The big thing to be aware of is that the first set of screws, in the base of the machine, use a special head known as a GameBit. Nintendo used these for their cartridges and consoles for several years. If you're into retro games and repairing them, they're a must-have. Here are a couple eBay links that should do you well:

$6 for two screwdrivers with handles:

https://www.ebay.com/itm/402453918947

$4 for standard interchangeable bits:

https://www.ebay.com/itm/112253805867

Turn the SNES over, remove the 6 screws on the bottom, flip it over again, and lift the top off. You should now see the circuit board, with its RF shield on top. Note in the image above, the large red arrow. This points to the trimmer capacitor on the board. The trimmer capacitor is responsible for setting the exact timing of the system. In the subsequent steps we'll talk about fine tuning it, but if you don't have the needed equipment to measure it, you can take this step and be done here. Use a small screwdriver and try to turn the trimmer capacitor and see if you can get color back. Good luck!

Otherwise, if you have an oscilloscope or a Saleae logic analyzer, read on.

Remove the cartridge ejection lever. Next, we need to remove the RF shield (the silver plate of metal covering most of the motherboard). See the image above for the locations of the 6 screws that must be removed. Note that one of them is under the main power switch, so you'll probably want to remove the switch first. Once you get the RF shield removed, disassembly is complete. Now we need to put on our thinking caps.

Now, the Super Nintendo's motherboard is fully exposed. See the image for what that looks like. It's relatively sparse, as far as circuit boards go. Look in the lower left and you'll see TC1, the trimmer capacitor that we'll need to adjust. But before we can adjust it, we need to know what to adjust it to. To do that, we'll need schematics. Luckily, some great folks have figured out the schematic for the Super Nintendo, and have uploaded them here:

https://wiki.superfamicom.org/uploads/snes_schemat...

If you look in the top center, you'll find the clock generating circuit. I've circled it here. You can see TC1 in the upper right of that section. If you look at the output of the clock circuit (off to the left), it goes down to U3's pin 31. It also goes off to U2's pin 100. Pin 100 is very convenient for us, because it's right on a corner, so we can easily monitor that pin and find what clock frequency the system is running at. Great!

From here, we have a challenge. You'll need a piece of equipment that can monitor a 21 megahertz signal. For most folks, that probably means an oscilloscope. I'm a college student living in a small apartment, so I don't really have the room or the money for fancy stuff like that.

Luckily, I've found a great alternative that works for me. This is a Saleae Logic 8 Digital Logic Analyzer. It lets you monitor up to 8 signal lines, although we only need one for this project. It fits in the palm of your hand and plugs into your computer, which makes it easy to analyze whatever signal it is you're reading. They're $400 which is way cheaper than a fancy oscilloscope. There's also a 50% discount for students, which makes $200 a great price for such a useful tool. For anyone with limited space or budget, I'd highly recommend this option. It's served all my oscilloscope needs. For this project specifically, its 100 MS/s digital sampling rate will do the job of monitoring our SNES clock signal.

I took my logic analyzer, and found pin 100 of U2. You can see in the images above what it looked like with the probes clipped in. I attached the ground probe to a separate part of the PCB. Be careful with the probe that you are touching ONLY pin 100 and no others! Luckily the probes are pretty tiny so this isn't too difficult to pull off, but if you're not paying attention it would definitely be possible to short something.

Once I was clipped in, it was time to monitor the signal. I plugged the game in again, reattached the console to power, turned it on, and verified that it was still outputting an image to the screen. Then, in the Saleae Logic software, I queued up a 5-second capture. By capturing the signal for 5 seconds, we can count the number of rising edges, then divide by 5 to get the frequency. We can't just directly measure the frequency, because the signal is 21 MHz and our capture is 100 MHz - we're only getting 4-5 samples per signal (you can see in the waveform that each cycle is about 50 nanoseconds), so we don't know precisely how long the signal is high. However, if we monitor for long enough, we can just count the cycles without knowing exactly how long it is between one cycle and the next.

I used the built-in Measure tool to count the number of rising edges, which the software tells us was 107410496. When when divide by 5, we find that our exact frequency is 21.48210 MHz. But we saw in the schematic that the master frequency is supposed to be 21.47727 Mhz. We're running at 0.00483 MHz too high. So we've confirmed that our clock rate is not what it's supposed to be!

To fix it, I went through a few cycles of adjusting the trimmer and then checking the result. I also unplugged the system each time, to prevent any accidents while adjusting.

After a few tries of fiddling with it, I got it to run at 21.47694 MHz, which is only 0.00033 MHz low. After doing that, the color worked perfectly! Apparently, that value is close enough to the proper value that the TV picked up the signal properly.

With the clock frequency adjusted, the color came back! Now everything looks exactly as it should.

If you're having this issue, this is a good thing to check. You could also have a broken trimmer capacitor, in which case you'll want to find a replacement. The original capacitor is adjustable between 5 and 20 picofarads.

Again, I highly recommend the Saleae logic analyzer if you've never tried one. You can buy them straight from the manufacturer at https://www.saleae.com/ - scroll to the bottom to view their full lineup of models. The basic Logic 8 has been great for me, but if you're doing fancier things, you might want to go another step up.

If anyone follows these instructions to fix their Super Nintendo, let me know! I'd love to hear if it was helpful to somebody :)