RGB output with sync from composite + sync stripper is janky

Hey, I'm new here, so excuse/correct me if I am putting this in the wrong place. I'm having some issues finding information regarding my topic online and you guys seem to know what's up with hardware stuff.

Anyway, over the past week or so I've been working on RGB modding my N64. That appears to be complete and working, but then I discovered that the only RGB SCART cables I have get their sync from csync on the multiout, and that the N64 does not provide that. I went to go buy some cables that get sync from luma, but all the retailers I know of that sell RGB SCART cables appear to be out of JP21 cables, with no indication of when they would return. Rather than buy a EuroSCART cable and re-wire it for JP21 (Since that seems hard and complicated), or buy an adapter (because I'm cheap), I figured it would be easier to just modify my cable I already have to use composite video through a sync stripper (since I seem to have an adapter for JP21->XRGB Mini that contains a sync stripper anyway, for reasons I don't recall). So I rewired my cable to run composite video through pin 9 and in my connector have a sync stripper that gets it's composite input from that same pin. At first it didn't work, but I did various changes and re-soldered stuff based on this guide and I finally started getting a picture.

The main meat of the issue, though, is that after doing that, the picture is all shakey and torn. Here's a picture:


Those lines go up and down the screen over time, they are not static. My figuring so far is that this has to be a sync issue, because before, when I was using csync, my SNES was able to work without any issues like that, and now both consoles output the same tearing problem (so it's either an issue with the cable or my upscaler, and since the cable is the only one that I messed with, I'm running with that).

Here is a picture of my sync stripper circuit, which seems like the most likely culprit here:


I can't see any issues here, but I'm pretty novice at all this. The sync out is a copper wire bundle instead of a solid wire, and some of that seems to be exposed. Don't know if that's an issue or not. The wiring setup is as follows:
Red: 5v
Brown: Ground
Thick yellow: Composite video in
Thin yellow: Sync out (presumably, I didn't wire the cable, but it seems to be working kind of)

My setup is my N64 connected to an XRGB-Mini Framemeister, outputting 1080p, connected through HDMI to my HDTV.

On the Framemeister, I have tried a few things:
- Changing sync mode: Auto makes my TV not recognize that there is a signal, off is where this happens
- Changing sync level: Might be making a difference? If I increase the sync level all the way to the top, it definitely is worse, though whether that's this problem getting worse or another problem showing up I can't tell.
- I've messed with the various scaling things, to no effect.

I don't have an RGB monitor on which to do a straight signal check, the Framemeister is the best I can do in that regard.



Erm... I think that's all the information I have. Just asking to see if anyone around here can shed some light on the subject, or let me know if there's something else I didn't think of that I should check. I'm not an EE, so most hardware stuff is a little out of my depth, especially when it comes to analog signals like this.

Thanks.

 

Which yellow wire is the input signal from the console? That looks like a SCART/JP21 housing, which would suggest that the cable running out of the left side of the photo goes to the console - it that's the case, it looks like you have the input and output of the LM1881 transposed. The video signal should go through the 0.1 to pin 2 and the csync output comes from pin 1.

I would also suggest adding another 0.1 between pin 4 and 8 on the chip, although it will probably work fine without it, especially if the wires aren't too long.

 

Oh, no, sorry. That's an adapter for SCART to whatever the XRGB Mini uses, since it doesn't use a direct SCART plug. So the cable running out left connects to my upscaler.

Pin 4 and pin 8 are the bottom left and top right, respectively in my picture? So there should be a capacitor between power and ground?


Also: I forgot, I didn't select those capacitors, but those are the style that are bi-directional, right? Because I desoldered that one and put it back on, but wasn't paying attention to which direction it was going. It doesn't have any + or - on it, so I assumed it didn't matter which direction I put it. Obviously some signal is getting through, since I'm getting some kind of sync signal, whereas when I disconnected the composite signal, I was not getting any picture whatsoever.

 

OK, so the input and output are the right way around in that case. Yeah, the cap should go across the power pins - it's just to smooth out the power.

The caps are ceramics, and can be connected either way around.

Another thing to try is terminating the video signal - put a 75R resistor between the console video out and ground.

 

Does anything happen using composite video for sync the ordinary way, without a sync stripper?

 

Do you have a 680 Ohm resistor in series with the sync stripper output?
It outputs full TTL level sync and many devices can't work with that.
The 680 Ohm together with the termination in the TV / Upscaler will fix it.

 

I don't have any .1s around, so I'll go pick some up later today. Basically the idea is to clean out any AC?

I'll give that a shot later tonight and report back.

That was actually one of the first things I tried. Someone said the Framemeister should work with composite video for sync, but I just get a blank screen when using sync mode off, and a "No input signal" from my tv when I use sync mode auto.

By in series, you mean from the sync output to ground, right? If so, that does not work, or seem to have any effect whatsoever (I clipped it while I was testing and nothing changed).

 

In series means inbetween, not to ground. It can either be "in series" or "to ground"

 

... Well that was stupid. I got all the names mixed up. I was thinking the options were "series" and "sequence" for some reason, so I was thinking series was parallel, and then I wasn't sure where that was supposed to go and guess I decided that was the right way to go. I actually should have known that.

So the objective is just to add resistance to the signal, right? So it's not as strong? I thought it would make more sense to connect it in line to achieve that goal, which is why I brought it up. But, just as an intellectual exercise for me, connecting a resistor to ground would still reduce the signal, wouldn't it? Just by a significantly larger amount? Assuming grounding that pin reduces the signal 100%, adding a 680ohm resistor would cause it to ground like 98% or something, right? Or is there something there about taking the path of least resistance that makes that not happen?

 

The series resistor forms a voltage divider with the input impedance - in most video circuits this input impedance is 75R, so the divider ratio is 75/680 = about 0.11 so this would attenuate a 5V signal to about 550mV. You could also work it out in the opposite direction.

 

Alright, so I added a 680ohm resistor actually in series this time and it almost, but not completely, solves the problem. The problem is probably reduced by about 8x. There's usually no more than 2 or 3 fairly minor tears on the screen at any given time now, and they don't last long. It's still there, though.

Following the "more is better" principal, I tried using a 1k resistor. That made the sync stop working entirely. Then, following the "well, that didn't work" principal, I tried a 470ohm resistor, which made it worse, and then tried a 680ohm + 100ohm resistor in series, which also made it worse. So 680 does appear to be a kind of a sweet spot, it's just not working perfectly. I don't feel like just randomly adding resistors until I get something good is going to get me any meaningful result, so I haven't tried that. If fiddling with resistors seems like the thing to do, though, let me know. According to TriMesh, there should be some math here that is somehow different for me than usual? Is the 550mV an incorrect value for use with the Framemeister? I can't find any info about what it will take. Does that 5V come from the 5V into the sync stripper? Or is it something different?

According to this page, the SCART cable may not provide enough power to run the sync stripper. Could that be the issue here? Measuring with my voltmeter, I'm getting 4.97v, which seems like it would probably be within tolerance, but I really don't know.

In other news: I updated the firmware on my XRGB Mini, and it started getting sync from pure composite, but the sync looks just like it did before I added the 680ohm resistor to the sync stripper. However, I can now use sync_mode auto with the sync stripper enabled and get the same display I had before with sync mode off on the old firmware. I don't really know what that means.

 

It sounds like you're good with the 680 Ohm and your problem is now with the Framemeister alone.
Have you tried tuning the sync level and could you try a different console please?

 

I have tried tuning the sync level, reducing it from 10 to 0 seems to make little difference, increasing it makes it worse, eventually causing breakage. I've been testing by switching consoles between the N64 and the SNES. They both appear to have the same issue, except the SNES works fine with csync. Neither of them have any issues outputting with S-Video or composite, it's just the RGB output.

 

Does the SNES still work if you pass it through your 680 Ohm modded sync stripper?
If it does, the sync stripper works well and your problem is before it (somehow).
That or your Framemeister is damaged..

Edit: Wait, where are you getting your 5V and Ground for the LM1881 from?

 

Do you mean running csync through the sync stripper? I haven't tested that, but basically the idea is that the lm1881 takes the composite in and creates csync, so if you run csync in it shouldn't change anything, and if you get clean out, then the sync cleaner is not affecting the csync signal, so the issue has to be in the original composite signal going into the cleaner? I'll test that a bit later.

The power and ground are from pins 10 and (I think) 5 on the multiout from the console. Ground is hooked up multiple places, the 5v in is connected to either pin 13 or 9 (directly to the right of the sync pin, if you look at it with the sync pin on the side facing you). There's a resistor from that pin to ground, I don't remember the size or colors of it, I can check later. I believe it was in the 1k range, but I don't remember.

 

Yes, run SNES csync though the sync stripper. It will act as a buffer to 5V TTL and otherwise work the same as with a composite source.
It just doesn't strip anything. This will test whether the output side of the circuit works fine.

Measure your 5V that go into the sync stripper please. It will only work reliably at ~4.5V or above.

 

I measured it earlier, I'm getting 4.97V.

 

Alright, reconnected csync, and now not only does it have issues when run through the sync stripper, it has issues when I run it through without the sync stripper. Which means I probably fucked something up on the original cable with my soldering skills that are on par with those of a drunk 3 year old, but I don't really know what. There was a ton of hot glue in the original cable that I peeled out, which I assumed was for the purpose of insulation, but it doesn't seem any of the connections are bridged. I've attached some pictures of the cable just in case:



Grey is sync, the yellow cable going off into nowhere is composite video.


So it does seem the original problem is fixed, but then I somehow introduced another one...



In good news, I tested my genesis (which uses csync) with both my clean cable and my sync stripper one and it works great, which means I definitely either fucked up my cable, or fucked up my SNES and N64 in the same way. Occam's razor says probably just my cable, so that's good. I just don't really know how.

 

Check your cable against the NTSC SNES cable schematic:
http://members.optusnet.com.au/eviltim/gamescart/gamescart.htm#snes

 

Is that Euro SCART? All of my cables are JP21, and don't seem to have any similarities to start you posted besides wire names. My cable is mostly accurate to this schematic. Whether that's accurate or not I can't say. The only difference is that everything labeled "shield" appears to just go to ground. I don't know if that's on purpose or not. I did desolder the one labeled sync shield once, but I put it back.