nesdev.com

So I was wondering if something like this is possible. I want to split RGB video over scart and figured I should be able to make something myself. I only have a pretty basic grasp of electronics but I figured I could split the signal with opamps. I found this example schematic straight from a opamp datasheet -



Which was more or less what I had in mind, split/buffer the signal for each required scart pin with an opamp.
Still, I'm finding very little info trying to search for others doing something like this. The opamp used in this image (LT6206) appears to be specifically for video signal amplification - is this required or could i perhaps use more commonplace opamps? I'll need to look up what kind of bandwidth rgb signals have, but I get the feeling that something like the generic TL07x with 3mhz might be too low.

Is this at all feasible to do? If not with generic opamps then at least with this specific one? Too bad i'd have to order it internationally and it only comes in a very small IC package (MSOP-8) haha.

Yes, it's plenty doable.

For video purposes, opamps have two pertinent properties:
1- Whichever is worse of gain-bandwidth and slew rate. A gain of only 2 is a little low, so slew rate is more likely to be a problem.
2- Current drive capability.

You also need to match the "impedance" of the video cable that is used. In most video applications this is 75Ω. That's why there's a 75Ω resistor connected to Vin in your picture - to prevent reflections - as well as why there's two 499Ω resistors on both sides (a gain of 2) and a 75Ω resistor on the output stage.

For standard consumer 0.7Vpp video into a 75Ω load with proper impedance matching, you need an opamp that can drive 1.4Vpeak into a 150Ω load, or 9.3mA. (For Sync-on-luma/green or Composite, peak voltage and therefore peak current may be higher)

For baseband SDTV timing, a bandwidth around 3MHz (i.e. gain-bandwidth of ≥(3MHz·gain of2=6)MHz; slew rate of ≥(3MHz·1.4Vp·2 transitions per oscillation=9)V/µs) will be a little smeary but not too bad. Using my US TV's s-video input, I can see a maximum luma bandwidth around 7-8MHz (i.e. gain-bandwidth around 14-16MHz; slew rate of 19-22V/µs); anything more would be wasted. Your TV's SCART input may provide more bandwidth ... or it might not, since square pixels on PAL are roughly at 15MHz anyway, a bandwidth of 7.5MHz.

Wow, that's a lot of useful information, thanks!

My interest has mainly been in audio so i've never really had to think about bandwidth & slew rate, haha. Started to realize it was going to be a problem when trying to read up on video signals...

Another question and some musings if you're not bothered...
it's quite common for audio circuits to have single supplied opamps and then bias the input to a value between Vcc and gnd, which also requires placing some coupling caps here and there to filter out the DC bias voltage (well, that's my limited understanding of it anyway).

Been trying to read up on video signals and they appear to have a DC component as well? So coupling caps would be no good I'm guessing? Perhaps the easiest approach there is to split supply the opamps to work around that.
The LT6206 example doesn't have coupling caps and is using a single supply though, and the text that goes with the example says "careful input biasing is required (or a negative supply as suggested previously)", so it appears to be doable but I just don't know how.

Officially, baseband video is defined in terms of absolute voltage. Blanking is supposed to be 0V; sync tip at -0.3V (sync-on-green) or -0.286V (sync-on-luma or composite), full scale at 0.7V (red/green/blue on VGA/SCART; other standards exist including 1Vp and 5Vp) or 0.714V (luma) or 1V (composite).

A bipolar supply for the op-amp might be necessary even without negative sync voltages: many opamps cannot drive the output all the way down to 0V if they're not supplied a voltage below 0V.

In practice, many TVs let you get away with an AC-coupled signal, as long as the corner frequency is low enough. Bad AC-coupling produces something known as "field tilt" (previous mention). Too small of an AC coupling capacitor (i.e. meaningfully changing reactance over the frequencies used for video) will cause unwanted reflections, too.

Oh, so it's not an AC signal on paper, but you can sort of-kind of treat it as one. I think i understand that.
For the opamp, I'll go with split supply to make things simpler then.

Thank you so much for the help! I'll try to get a prototype going and see what happens haha.

Haven't gotten the parts required to build a test yet. In the meantime i found another video line driver from a opamp datasheet, this time AC coupled with a biased input. Could be extended to a splitter rather easily i'd think. I'll try the inital schematic first though.

Alright, the first round of testing failed but not for the reasons I would have assumed haha.
I took two scart ports and wired the R, G, B and composite signals (for sync) pins through the opamp circuit. The other necessary pins (like audio and the various gnd pins) I just connected between the scart ports with a wire.
This didn't send anything, not even audio... so we disconnected the opamp circuit and tried getting just the scart-scart connection to work.
Eventually we noticed that plugging in the snes cable in one end and a scart->3xRCA breakout on the other side would transmit both audio and video. So the wiring was clearly not entirely wrong.

I must be overlooking something about connecting two scart cables... i'm at a loss there. Back to the drawing board...

Audio and video lines are crossed on SCART cables, out on one end goes to in on other. RGB, blanking and grounds are straight.

_________________
http://www.tmeeco.eu

Oh... yes, i see now what you mean... haha. Thanks for pointing that out!

Crossed the audio & composite wires and... well, got something!
Here's the 240p test suite: https://i.imgur.com/4FmmRRX.jpg
It's black & white, but at least the image is quite crisp! I had expected it to look worse.

Tried disconnecting some color wires and actually got color output, albeit a bit dark.
Only red: https://i.imgur.com/uuLqXDs.jpg
Red+green: https://i.imgur.com/ysgyZ6D.jpg

So at least it's sorta working! I'd love to keep going but I think this is as far as I can take it right now. I don't really have the test equipment (or even a snes/tv, this was my friend's) to diagnose why it's not working properly. The opamp I got isn't quite up to spec (GBWP=16mhz, 7V slew rate) so might be that, prices ramped up pretty quickly for these things and I just wanted to give it a try. Or, well, could be just about anything else. Haha!

Anyway, thanks for helping out! This is definitely doable, might revisit it later if I get better test equipment!

Your photo of red+green is all yellow - there's no shades of red or green within it. This implies that you've somehow accidentally made multiple copies of the same signal, instead of one input per output...

Hmm, yes, you're right. Not sure why i didn't think of that. I can't see how that's the case, but i'll examine the circuit again.