Making a component RGB video input module

cadet
#1

So.
I want to make a component video input into my Cadet system.
The goal is to have full RGB video input instead of the single black and white luma I have now.
I think I will make 1 big pcb out of this. Or maybe 3 joined / sandwiched smaller ones… not sure yet.

Lars and Pbalj have given me some tips (see below), but I’m a bit stuck.

I have copied the layout for the video input pcb in Eagle.
The idea (I think) is to copy the sync info that is coming into the first input to the other 2 inputs.

Now I’m looking at the sync points that need pinouts. Somewhere at the LM1881?

Should the video input be RGsB or YPbPr?

Or is it already possible with 3 “vanilla” RGB inputs, by patching ?

the tips:

pbalj:

you could also get component in to cadets if you use green in to the sync, and then in to video input, as well as r and b in their own video inputs as well.

Lars:

You would need to modify the secondary units to share the sync signals generated by the first, but you could definitely use 3x Cadet IIIs to get component video (YPbPr and RGsB) into your Cadet rig!

#2

As far as I can tell it’s almost possible just with three straight modules, except for the back porch sync to the cd4053 on the R and B inputs. In component video, the green has the sync, so as Phil said you’d need to run the green channel into the C1 to get the sync (which it then shares with the C2), then from there into a mostly unmodified C3, and that’ll grab the back porch from the sync on that channel. The one modification the green C3 needs is to send the back porch from the lm1881 to the EN pin on the red and blue C3’s. And the red and blue C3’s can omit their own lm1881’s.

I think that’s the concept in theory. Anyone please correct me if I’m wrong!

#3

Actually looking at cd4053 specs, you might not need three of those. I think you can use the same one for r,g, and b. So you’d need to feed three input circuits into x0, y0, and z0, but the lm1881 is only hooked up to the green one? Maybe?

#4

it starts making sense… I was wondering about the use of the cd4053.

#5

I don’t actually know what the 4053 does. I suspect it might be blanking the input signal during the back porch part of the signal, which I think just strips the color burst timing signal? So maybe you don’t even need it for component input? I don’t entirely know what a component signal is like but I don’t think it has the same color burst timing thing since the color is on separate channels, eliminating the need for a color sub carrier.

#6

I’ve been thinking about this a bit more (boring day at work) and reading a bit about component video… Sounds like YPbPr is more common than RGsB? I don’t have experience with either so I don’t know, but if that’s so it might be worth making this work with YPbPr instead? It’s a little more complicated but not terribly so, since you just have to do some math with your signals. Strip the sync from Y and then do some adding and subtracting and dividing. Or, perhaps even easier with YPbPr, use a LMH1251. That chip even has a switch that switches between two inputs, YPbPr and RGBHV, so you might be able to use the RGBHV input to handle a RGsB input without much work too, just by stripping the sync out of G before putting it in there, maybe? (I’m imagining using a LM1881, and then you get H and V signals to use too, if it actually needs those to work.) It’d be quite different than the triple C3 method, but doesn’t seem much more complicated.

I’m curious to see where you go with this, what direction you take! Now I’m going to look into composite to RGB, since I don’t have any component stuff but I would like to use some of my color composite things…

#7

maybe you want to help me design this?
We can do it the same way as with the sync busboard. like an open project .

I’m no expert in electronics. more like a beginner with some experience :slight_smile:

#8

The CD4053 in Cadet III is for DC restoration, to restore black level of external video to 0V level. For an RGB input I would omit the LM1881 from the Red and Blue C3s and just wire the connections from the original build (Green.) If you’re doing your own board you can get away with just a single CD4053.

If you want to do YPbPr to RGB (like Visual Cortex) it is more complex and you need a colorspace matrix converter (opamps and precision resistor values) after the DC restoration.

Here are the steps to follow…

  1. Terminate with 75 ohms and buffer your input.
  2. DC restore black level to 0V (burst gate extraction with LM1881, 4053 switch to 0V, and then opamp buffer). Look at C3 circuit.
  3. Gain from external video levels to LZX levels. This is a gain of 1.4X (0.714mV to 1.0V) scale.
  4. Buffer with opamp to 499R resistor output. Steps 3 and 4 can be combined. Cadet 3 also includes pots for gain/offset adjustment in all this.

Some other optionalsteps you could consider:

  1. Chroma notch filtering, to remove color subcarrier from a composite video input.
  2. Sync clipping. After DC restoration, run thru half-wave rectifier to remove sync from the signal entirely. Note that if you do this with Pb/Pr you’d be clipping off the negative half of the color component (which is bipolar instead of unipolar like RGB.)
  3. Colorspace Matrix. Note that all signals must be DC restored first for this to work correctly. Opamp math for the following:
    R = Y + 1.140Pr
    G = Y - 0.395Pb - 0.581Pr
    B = Y + 2.032Pb
#9

Hah , sounds like a good challenging project!
I would not mind help from other DIYers :woman_genie:

For starters, I’ll try to make a RGB input.
I’ll get started and post my progress.

I think most component video is YPbPr, so that direction might be preferable.

#10

I don’t think i’m up for helping in design, but i’d be happy to help test… including building up boards for testing

#11

I can peek on here every now and then and offer some thoughts, but I’m also no expert, and I have some learning/projects of my own I should be devoting my time to… :slightly_smiling_face:

#12

Awesome help there, Lars. Thanks! I guess what I had figured out was kind of close, but not all the way.