Author Options:

RGB TV Output of Gameboy Color? Answered

 I'm absolutely stumped by this. I have my gf playing my old gameboy color but the screen cover is pretty scratched up and I don't feel like peeling it off until I find something to buff the scratches out.

I've nearly googled it to death and all I come up with is random websites selling gameboy advanced tv out adapters. I know this HAS to be possible as www.hackaday.com/2008/11/29/gameboy-color-on-an-led-matrix/

The guy says in his comments on the youtube page that he does IN fact have a real live gameboy color hooked up to the LED matrix. Yet he does not explain how he hooked it up, obviously he did it somehow.

Does ANYONE have any bloody clue as to how this is done? I'm not afraid to do some soldering as I know that'll probably be required but really how hard can it be.


If you have a Super Nintendo, there is a cartridge adapter which allows you to do so. Sometimes the games like pokemon yellow have specific frames around the screen with pikachus for example.

if the screen cover relly messed up you can buy a cheap replacement cover on ebay for a few bucks as for tv out im not sure

You'll need to find the video out connections where they go to the screen, disconnect the screen, and solder wires onto that. Then solder those to RCA cables. For the sound, much the same except with the speaker instead, and I'm not sure if Gameboys have 1 or 2 channels. 

 I do believe the gameboy has a regular 3.5mm headphone jack and that I can deal with. As far as video out though what am I looking for exactly is the question. Should I be looking for a pinout diagram of some kind or a schematic of the entire gameboy if one even exists??

Oh! Try a 3.5mm to rca cable - they sell them all over. Otherwise, yes, that's exactly what you need. :)

Are you winding the guy up or do you really believe you're going to get video out of (no, it's got to be a wind-up)


Er, no - I've used this exact cord to connect the headphone jack on a laptop to a large flatscreen TV - I can assure you it works. How do you think the "hook any mp3 player up to your tv" cords work? 3.5mm to rca, while perhaps unlikely sounding, does work, at least in some scenarios.

Laptops output video through the sound-card? That's a new on to me, which model was it? - I'll have to find out more.


Some do, I guess. Ipods, too, for a certainty, and other mp3 players with video. You may not find the specs online to say either way, but I swear it does work - HP dv6000 series.

Tjanks. I'll not say no to this, but as yet I can't understand why or how someone would output video through the audio-jack. I'll commence looking harder tomorrow (it's 12:03)


 it would make more sense to just have an svideo/composite jack on the side of the laptop no?

I think it may be a (composite plus L&R audio)-jack rather than the audio-jack. So you'd use a 3-channel cable. Mind, if the device has video-out it would say so in it's specification.


Could be. I haven't got the manual handy; the port is only marked by a headphone symbol (two identical ports, exactly - headphone 1 and 2). So it can't hurt to try, worst case scenario is it doesn't work and it's back to the drawing board...

I'd agree with that. My video camera has a socket v much like a standard audio, but it's as I described above.


 Aye I hear ya, kinda like my video card provides svideo and composite outputs, along with spdif output for the tuner card.

As I say, I have no idea if it's done with the Gameboy, but it's definitely done on at least some devices, including iPods (and, apparently, my laptop).

 I completely agree, that piques my curiousity to hear of such a way of connecting a laptop, that's a first for me. A mp3 player or handheld camcorder maybe, but a laptop?

 I'm trying to learn more about the different kinds of video signals.

You suggest using RCA cables. Those are the 'standard' plugs a lot of devices use right? Like the yellow plug for image and red/white for sound.
Those use 1 plug for image, but you are talking about finding the "video out connections (plural). Do you mean it's really as easy as just connecting the + and - of the video signal to a RCA cable?

Yes, exactly!  

Well, I've never opened a gameboy, but I imagine you would find some sort of ribbon cable going to the screen. I have a hunch it _ought_ to really be that easy, based on similar projects, but I don't know for sure.

In my experience it's actually surprisingly easy to hack that sort of thing (audio/visual cables) but success depends on several things: quality of wires/soldering - doesn't have to be wonderful but an unintentionally shoddy job won't do - and, of course, whether you're soldering the right wires to the right wires. :D A pinout would definitely help with this.


8 years ago

The process is detailed pretty well on his blog...

Here's a breakdown:

-- Sixteen separate power supplies for the matrix (with a relay-driven power-up sequence, so his circuit breaker won't trip from the inrush current of all 16 simultaneously.)

--144x144 matrix. 20,736 discrete LEDs.

-- An matrix decoder/LED controller built from an ALTERA FPGA board. FPGAs are user-programmable logic gates--essentially a roll-your-own microcontroller.

In short--pretty heavy stuff.

If you knew how to program an FPGA, you might be able to create an RGB output--by decoding the matrix output of the Gameboy board. Would take a pretty good understanding of video timing, etc. too.

Frankly, this would be easier than controlling a large 144x144 LED matrix. But it still wouldn't be easy, even for a seasoned engineer...

 Holy christ, well I had gathered a bit of info on the project but I didn't know it had 16 different power supplies! Another thing, I know what FPGA's are and have a bit of an understanding on how they work, but I wouldn't for the life of me be able to program one, let alone set aside $150 or so and buy one.

They would be a nice toy for my collection though, as FPGA's are usually used for some rather unique tasks that aren't normally possible with regular circuitry, they intrigue me quite a bit.

In the short and curlies of it all, yeah you're right, it's not worth it... sigh but I thought there would be some possibility that someone had already done this WITHOUT an FPGA.

On my previous experiment (BaBar) we deployed Xilinx Spartan 3 FPGA for two critical tasks:  the Level 1 trigger system used them with firmware to do three-dimensional charged track reconstruction (a five-parameter fit to a helix) in less than 10 us; and the drift-chamber used them to do digital-waveform integration and feature extraction on-board the front-end electronics.

Both of those are tasks which had previously been done with C++ code running on a Motorola 68040-based farm.

When I headed the drift chamber upgrade project, I was blown away by what these modern FPGAs can do!  We move all the feature extraction code down onto the FPGAs, and had enough room left on them to also run an emulation of the previous readout hardware, with software-selectable switching between the two.  And we ran two copies of each set of code for fail-over redundancy in case of single-event upsets.  We didn't two three-way majority voting the way space-qualified FPGAs do, but we got pretty close.

How many 68040's did that single FPGA replace?

Well, it's the other way around.  The front-end electronics on the BaBar drift chamber consisted of 48 modular boxes; the output of these boxes were multiplexed onto four optical fibers (using HP's GLink), each feeding to one 68040-based single-board computer. 

Each processor ran the feature-extraction code sequentially, processing all the channels from 12 boxes (a total of up to 1,776 waveforms) and putting the output onto a local gigabit network.

We put one FPGA into each front-end assembly, so it did the feature extraction for up to 196 channels per event, then shipped the output data for multiplexing and fiber transfer.

Not only did this result in a great gain in processing speed per event, it also reduced the data volume by a factor of 5:  The raw waveforms are 32 bytes per channel, while the feature extracted data average six bytes per channel.

The upgrade allowed the chamber to record over 10k events per second with zero deadtime, compared to 2k events with ~5% deadtime before the upgrade.

Ah, I interpreted:
Both of those are tasks which had previously been done with C++ code running on a Motorola 68040-based farm.

...as meaning the 68K CPUs had been replaced. So instead part of their functionality was offloaded to the (much faster) FPGAs.

The FPGAs replaced the FEA assemblies pictured on that page? Those pic show boards populated with Xilinx (and some Cypress) ICs. But not Xilinx FPGAs, eh? Perhaps those were older CPLDs or some other programmable gate array...

Nothing like replacing a software solution with dedicated custom hardware. FPGAs are amazing. Too bad I don't have a good excuse to play with one...

In fact, exactly the same boxes were used.  The three "ROBs" on the photos page are the boards that were replaced.  Identical footprint boards (ten layers) were fabbed for installation into the same aluminum boxes (the custom ASICs used for digitization of the signals are still there). 

The Xilinx's you see on the old photos were ~1996-era FPGAs, along with supportting EPROMs and separate data buffer chips.  They were so small that we need three per board just to do the waveform data packaging and multiplexing to go out on copper cable to the fiber interface boards.

The new boards used a single Xilinx Spartan 3 each to replace all of those tasks, including the data buffers, and to do the feature extraction work as well.

We also designed the new boards with a JTAG interface connected to the fiber-optic command and control channel.  That allowed us to upload new firmware to the boards in situ. The detector is on the beamline, behind a concrete shielding wall, and with a multi-ton concrete and steel "plug" covering the end of the drift chamber, so doing reprogramming with a JTAG cable was not an option.

Nice. One new FPGA replaces six of the older arrays (and then some.) With some FFT (?) analysis thrown in...

I don't know enough about FPGA's to know if you program the gate array directly, or the eeprom memory (or both.)

They sure are a powerful tool (especially considering the low cost._

Our feature extraction doesn't need FFT.  What we needed to do was take a digitized single pulse -- baseline, middling-fast rise, slow fall back to baseline -- subtract pedestal, get the timing of the leading edge, and integrate.  The hit time and charge integral are what get used for the physics reconstruction later.

For the programming, you store the binary image in memory (EEPROM or in our case flash RAM), and the FPGA knows how to load that image into itself at power-up time.  Our trick (which others have done, of course) was to set up a reverse path:  we could ship a new binary image down to the FPGA, which would write it out to the flash RAM via JTAG, and we could then tell it to reload itself from the new image.

Here's a poster I did for the 2005 IEEE NSS/MIC meeting about the project.

Non-FPGA solution? Possible, but it would be a nightmare with discrete logic chips--like building a working CPU with simple TTL chips (which actually has been done, but you'd have to be a masochist ...)

He's also using two 16 channel RGB LED controllers in addition to the FPGA board. And when you think about it--20,736 color LEDs and each one can have 32,768 different color variations. That's a lotta data in the pipeline...

All that wasn't cheap, I'm sure.

Here's a patent application in which they describe exactly what you were speculating about. Like I said, RGB video output would actually be a lot simpler than an LED matrix. But even that patent uses an ASIC chip (very much like a FPGA, but it's internal gate programming is permanent, with big upfront costs for developers.) The LED matrix on the gameboy would need to be decoded, including all that color information...

The LCD screen is a matrix. I'd guess he's mirrored the matrix in LED and hooked-up the display output to that. Probably involves pre-assembled panels and microcontrollers.
RCA output it won't have.


 Sigh... I figured as much from what I COULD find on google that it doesn't have the circuitry to do it. And yes, I can just imagine him rigging up his LED display to each pixel on the LCD itself, lots of wires.. makes me cringe.

But heres another thing, you can get an adapter for the gameboy advanced to give RCA outputs, and I doubt it's just a matter of the adapter tapping into a socket on the gameboy because the unit from the pictures I have seen, makes the gameboy twice the size.

Gameboy Advanced is something else, but I'd heard of that.
W/ref what you have, most of the job is in the LED matrix and control. After that you just need the right connections (I wouldn't know)


 Eck... Too much work.. I do have an "emulator" on here, and including the video filters I get with it, I actually get better video quality than I would right on the actual unit. But I thought it would be cool as heck being able to play on my big 32" TV with my little gameboy color.