Designing my dream AV Switch

Hey guys, First post here but been reading through threads here for awhile.

So I've recently caught that lets get all my old consoles working with the best possible video signal on my HDTV bug. A lot of it was due to the release of the HDRetrovision component cables. I've always knew composite was crap but I didn't want to start hacking up my consoles to add svideo (even though I have the skill set for it) or spend a load of money on framemeister. Seeing what the guys at HDRetrovision have done without any (or minimal on the n64 for example) mods got me excited about getting all my consoles running though cables like that. So I started some testing and I found that I can put any consoles RGB signal through my Genesis HDRetrovision cables.

Now we get to my initial switch plans. A box that has multiple mini-din9 (genesis style pinout) in and 1 mini-din9 output. I could make multiple cables that adapted RGB consoles output to mini-din9 and hook them into this switch and output out my HDRetrovision cables. Then using something like a Bluetooth arduino micro-controller control the whole thing with a cell phone app. I have already mocked up on a breadboard using 8 bit bus switches (SN74CB3Q3245PWR). This allows for switching all pins except the 5V pin (that kills the trace in the chip I found out on accident lol) which I could use as a detect pin maybe later for auto switching when a system is turned on. My breadboard prototype is able to switch my x'eye and my n64 thought the HDRetrovision cables nicely. Well at least nicely for a thing that is a bundle of wires with no decoupling. There is some interference coming though in the picture but I accepted that with this setup. There is also the issue that the audio seems to be able to leak though at a really low volume from a unselected input. I'm gonna have to play around with that see what I can do to prevent that. My guess is the open switch is acting as a capacitor and the AC signal of the audio is able pass through at certain frequencies. If any computer engineers know a solution let me know.

Now this brings me around to why I'm making this post. I had an idea while trying to design this switch to be module and flexible. This idea I wanted to share with a community that is also interested in these kind of things who might have some input and insight that I might have missed. Also I am a Software Engineer not a Computer Engineer so I am by no means an expert to circuit building and analog signals, but I do work with some rather brilliant CE's and have learned a great deal along the way so I'm not completely in the dark. So any CE advice from guys on the forum will definitely help going forward.

So lets get to my big idea and what I would consider my dream AV switch. A module switch that had a module for each input type (RGB, Component, HDMI, etc) that could stand one its own or be joined to other modules to suit your needs. The switch I described above would be the RGB module 8 RGB inputs and 1 RGB output. Then there could be a component module that has for example 3 component inputs and 1 component output, but could also have a circuit similar to what's in the HDRetrovision cables to convert a RGB module's output to component as well. So you could get a RGB module and a component module hook them together and then output all your signals through component. Then the last module I had in mind would be an HDMI module that used the OSSC circuitry to take in RGB or Component and output it out HDMI as well as possible switch 2 or 3 HDMI ports. This would in theory allow you to have 1 switch that you can hook every console into, but also you would only have to buy the modules you want.

The way I can see this working so far is buy having a controller module that contains the software and the microcontroller and maybe a LCD display and a few buttons. And then using something like a 20x2 header that has preset pinout for each possible signal ( R, B, G, S, CVBS, R audio, L audio, Y, Pb, Pr, TMDS 0, TMDS 1, TMDS 2, TMDS CLK, i2c clk, i2c data, etc). That way each board can receive a signal from any other board and the controller module can use software to determine all the switching and settings.

This post had already gotten to long but I have so much more on what I've already figured out and designed. I'd like to hear what you all think? Any one see a fundamental flaw in my thinking that would prevent all this? I think next I'll make diagrams to adds so visuals to this wall of text. Thanks for reading.

 

Welcome to the forums! I am not an electrical engineer, so please take the following with a large grain of salt. I'm just trying to get the conversation going, as I find this to be an interesting topic as well.

I think you want to use a low voltage multiplexer switch rather than a digital bus switch. The gscartsw uses the MAX4638EUE. I can't explain to you exactly why this is important, but I bet there's a reason (adding too much resistance and/or impedance on the lines?). You may not be able to get rid of your interference/leakage issues using the chip you mentioned.

As for the modular idea...I've thought about the same thing. I'd love to have one "super box" that just took care of all of my retro gaming need. However, I think you run into some serious problems really quickly.

Let's start with the nitty gritty of what you proposed. Component and RGB signals can be treated essentially the same for switching purposes, so the circuitry would be identical for such switching modules. I don't believe there are any 3:1 multiplex switches, so you'd be building a 4:1 switch or an 8:1 switch. It seems like you're proposing to chain a RGB switch to a RGBs to YPbPr transcoder to a component switch to a digitizing line doubler (the OSSC). This creates an unnecessary transcode for RGB consoles. The OSSC doesn't care if it's being fed RGB or YPbPr. Supposedly, it can handle either on input AV 1, but I don't have one yet to test.

I think you've fallen in love with the HDRetrovision cables for the wrong reason. The "magic" of the HDRetrovision cables is that a total noob can buy them, use them on an unmodified console, and get appreciably better video quality from their console. Each cable HDRetrovision makes is "tuned" to the quirks of the console it's designed for. However, in its most basic form...it's just a RGBs to YPbPr transcoder. It's not a scaler or a line doubler. The end result is completely dependent on how your TV handles 240p. As component video inputs start to be depreciated and removed from TVs, the HDRetrovision cables become as useless as the stock RGB SCART cables you could have bought with the consoles back in the day. In my opinion, the future is HDMI or bust. The OSSC is a huge step in the right direction, but you still have to feed it the best quality RGB and sync as possible to get the best results.

The industry standard connector is still SCART. Yes, I hate SCART plugs too, but it's the way consumer RGB (outside of Japan) has always been done. If you want to cut all custom cables for yourself, that's cool. But most are not going to want to go that route for cost reasons...

...which brings me to my final point. Cost. This would make an awesome passion project, but if you're just looking to save money...this is probably not the route. If you're looking to make a salable product, you're going to have to restructure things to use more standard connectors or offer a complete line of cables. Even then, I can't see such a product costing less than a gscartsw and a component switch.

Superg has pretty much designed the ultimate SCART switch (with a similar component switch in the works). Micomsoft has pretty much designed the ultimate 240p gaming scaler. I know you don't want to hear this, but if your end goal is RGB to HDMI scaled properly, just go and "spend a load of money on a framemeister"...and a gscartsw.

 

Thanks for the welcome.
I am looking into a different chip to handle the switching and thank you for model number of what superg's switch I was trying to find what he was using but none of the images show the chip clear enough to read it off. I might use that or another one. Issue with that one it requires one chip for each channel which in my case is 8 and I'm trying to keep the boards small to keep cost down. One of the reasons the SCART switch is so expensive is its size (needed because SCART connectors are so large) and I believe it is 4 layers. Its definitely higher quality than what I am building for sure.

Right now I would just like to be able to pass my X'Eye, N64, SNES, RGB Mod NES, TG16 and PS2 (RGB) through my HD Retrovision cables. I get that they are just a RGB to component transcoder there just a lot better made then no name made in china converters that I have tried. It's also much easier and cheaper to build cables for Din9 than SCART. And as of right now even the top end 4K TVs still have component in. And my receiver can handle component to HDMI conversion if needed (not sure the lag tho).

This is definitely a passion project but I figured if I'm going to build it I might as well share what I've come up with along the way. Also I will have more than one of these when all is said is done (its not possible to order just 1 PCB). Also if I'm gonna build a RGB switch I might as well design it in a way I can expand it in the future. This is what set me on my common bus idea root. The common bus would allow me to add more functionality down the road. Also you talked about passing the signals through multiple transcoders which is exactly what I am trying to prevent. Common BUS isn't input of one layer is outputted to the next but rather each layer can put what they want on the bus. The software will control what is on each bus and will have to communicate it to the layer it is trying to output. Being a software guy I am design it in such away the hardware does the bare minimum of switching and the software will handle the rest.

The cost of making something from scratch will always be more than buying what is already available I get that. This way I get more of what I'm personally looking for rather than a long daisy chain of hardware that needs to be reconfigured for each system I want to play.

 

The gscartsw only has six MAX4638EUE chips, so I'm guessing that it only switches R, G, B, Sync/composite, R-audio, and L-audio. What are the other two channels you are trying to switch? I assume all grounds are tied together and do not require switching.

I opened up a Shinybow SCART matrix switch and all of the chip labels were neatly sanded off. They were fairly large chips too. So no particular help there...

I'm starting to more clearly see where you're coming from with the software switched common bus idea. Conceptually, I think it's awesome. However, I'm a little concerned that I can't find any examples of analog video signals being treated as such. My gut feeling says that this has to do with interference and lag. It seems like you want to treat analog video with the same approach as you treat digital logic...and I'm not sure how well that'll work. Maybe you can find some clever ways to get around the interference issues, but it looks like you're in uncharted territory.

As for the SCART vs Din9 issue, it seems like you're spot on. The gscartsw does indeed cost what it does because it uses expensive connectors across a large 4 layer board. A Din9 switch could easily be a third of the size and do the same job. However, this means you need to cut all custom cables. SCART plugs also give you the space to add the caps and resistors required to some consoles RGB lines. If you were making a DIN9 cable, you'd have to find a creative way to get those in the circuit without introducing too much interference. My first thought was some kind of small (shielded?) plastic blob on the end of the cable.