gscartsw / gscartsw_lite / gcompsw switches support thread

[eric90000]

I received my gscart lite and gcompsw today...I just wanna give a huge thanks to you SuperG. I have had the gscartsw 3.4 for a good while and I didn't think it could get any better, I was wrong! The lite not only looks sharper due to the lack of LPF, but it has actually solved compatibility issues with my OSSC and my 4K TV (Samsung KS7000).

Before Linex3 was unstable and giving serious screen jitter, even with the CS toggle set to ON on the gscartsw 3.4 (Linex5 always worked). Now with the Sync regeneration switch set to ON, Linex3 is flawless, completely stable and perfect crisp image quality (same cables, consoles, power supplies etc).

Now I can't wait to send the other SCART output to a 32" consumer TV for some lightgun games! There's not much to say about the gcompsw, other than it just works flawlessly, no issues at all so far (tested XBOX & Wii).

You rock SuperG.

[DirkSwizzler]

Received my gscartsw_lite yesterday and hooked it up to my snake pit of cables. Does turning sync regeneration on/off while a signal is already active likely to cause significant hiccups?

I haven't yet diagnosed the problem. But I ran my xbox with component cable (for some reason I can get 1080 with component but not scart) into a gcompsw into a garo into the gscartsw_lite into a gscartsw 3.4 into a framemeister. I had previously used the same configuration with the garo hooked up to the gscartsw 3.4 directly.

When I turned off sync regeneration the framemeister would lose signal until I turned it back on. The previous configuration bypassing the nonexistant gscartsw_lite didn't seem to have any issues.

Will poke at it more tonight to see if I can gather more data.

[superg]

Received my gscartsw_lite yesterday and hooked it up to my snake pit of cables. Does turning sync regeneration on/off while a signal is already active likely to cause significant hiccups?

I haven't yet diagnosed the problem. But I ran my xbox with component cable (for some reason I can get 1080 with component but not scart) into a gcompsw into a garo into the gscartsw_lite into a gscartsw 3.4 into a framemeister. I had previously used the same configuration with the garo hooked up to the gscartsw 3.4 directly.

When I turned off sync regeneration the framemeister would lose signal until I turned it back on. The previous configuration bypassing the nonexistant gscartsw_lite didn't seem to have any issues.

Will poke at it more tonight to see if I can gather more data.

Shouldn't be a problem for gscartsw_lite to hot switch on/off sync regeneration but it actually depends on your TV input, if it locks onto sync and sync changes it can be a problem.
Some people had that issue with v3.4 and sync stripper.

Mm, there might me something with Garo and lite. Do you have your v3.4 with sync stripper on?
Is lite stable with Garo and sync regeneration on?

[Guspaz]

I've been talking with Ste about the GARO, and it seems there are quite a few issues with the design: apart from potential issues caused by all the extra scanliner circuitry (which shorts to ground every other scanline) and the video outputs being unbuffered, one of them is that the chip outputs separate sync which are combined by an XNOR, which he said was common, but causes issues because it's not technically accurate. This makes sense to me because the OSSC has lots of sync problem with the GARO's RGBS output, but no sync problems with its RGBHV output.

With his help, I'm going to be trying to do an RGB and sync bypass on the thing, grabbing the RGB outputs of the converter chip and the luma input and running them through a THS7374. Hopefully it will help things: all of my devices support sync on luma, so grabbing what is essentially the unmodified input sync signal should hopefully help things. I'm going to try to repurpose a Voultar SNESRGB board to do that, although it will be a while before I can get my hands on another, and it'll involve smaller soldering than I've ever attempted before

[Ward1986]

I got a gcompsw and planned to connect it to a Framemeister but the composite and component share the left and right audio out jacks.
So to solve the problem of having to swap the cable back and forth between the audio in on the D-Terminal and the front AV in on the Framemeister I bought one of these.

Just search for stackable rca audio in ebay, I got a 50cm one for £3.49.
Just thought I'd share that incase anybody else has the same problem.
Hope it helps

[DirkSwizzler]

Received my gscartsw_lite yesterday and hooked it up to my snake pit of cables. Does turning sync regeneration on/off while a signal is already active likely to cause significant hiccups?

I haven't yet diagnosed the problem. But I ran my xbox with component cable (for some reason I can get 1080 with component but not scart) into a gcompsw into a garo into the gscartsw_lite into a gscartsw 3.4 into a framemeister. I had previously used the same configuration with the garo hooked up to the gscartsw 3.4 directly.

When I turned off sync regeneration the framemeister would lose signal until I turned it back on. The previous configuration bypassing the nonexistant gscartsw_lite didn't seem to have any issues.

Will poke at it more tonight to see if I can gather more data.

Shouldn't be a problem for gscartsw_lite to hot switch on/off sync regeneration but it actually depends on your TV input, if it locks onto sync and sync changes it can be a problem.
Some people had that issue with v3.4 and sync stripper.

Mm, there might me something with Garo and lite. Do you have your v3.4 with sync stripper on?
Is lite stable with Garo and sync regeneration on?

I've been talking with Ste about the GARO, and it seems there are quite a few issues with the design: apart from potential issues caused by all the extra scanliner circuitry (which shorts to ground every other scanline) and the video outputs being unbuffered, one of them is that the chip outputs separate sync which are combined by an XNOR, which he said was common, but causes issues because it's not technically accurate. This makes sense to me because the OSSC has lots of sync problem with the GARO's RGBS output, but no sync problems with its RGBHV output.

With his help, I'm going to be trying to do an RGB and sync bypass on the thing, grabbing the RGB outputs of the converter chip and the luma input and running them through a THS7374. Hopefully it will help things: all of my devices support sync on luma, so grabbing what is essentially the unmodified input sync signal should hopefully help things. I'm going to try to repurpose a Voultar SNESRGB board to do that, although it will be a while before I can get my hands on another, and it'll involve smaller soldering than I've ever attempted before

I gathered more data. The TLDR is that the gscartsw_lite is super awesome and the Garo is outputting something weird.

I narrowed down the problem to be a bad interaction between the Garo and the original gscartsw 3.4 when the xbox is outputting 1080i, but not 480p. But ONLY if the Garo is plugged into any port other than 8 on the gscartsw. I must have had it plugged into port 8 before I hooked up the gscartsw_lite. When it's plugged into port 8 everything works just fine. If it's a port other than 8, it works until the xbox switches to 1080 and then it appears to get a garbage sync signal where I can make out parts of the menu but it's moving around a lot and switching back and forth to screen garbage. So yeah. I'm chalking it up to the Garo being goofy as Guspaz mentions and the gscartsw switching back and forth to port 8. Which just happens to be "good enough" if the Garo is already on port 8.

If I plug the gscartsw_lite into port 8 on the gscartsw 3.4 , the signal is stable with sync regeneration on and off.

If I plug the gscartsw_lite into any other port of the gscartsw 3.4, the signal is stable if I have sync regeneration on, and it completely drops the input if it's off.

I was already planning on hooking up all my non-csync sources to the lite and leave regeneration on all the time. So I'll just have to write down a note that the Garo is weird if I have issues again.

[DirkSwizzler]

BTW. If you ever make a gcompsw that outputs scart RGBS. Or just a gcompsw with a special power port near the output such that a transcoder box could be plugged into it directly without extra wires. I'd buy that in a heartbeat

[Guspaz]

Unfortunately it seems like chips that do the conversion aren't cheap: it looks like the TI LMH1251 used by the Garo goes for ~$11 just by itself (unless you buy in huge quantities), and you'd need other supporting circuitry (to buffer the output, or to combine the sync, and so on)... This would probably increase the cost of the gcompsw too much for something most people would not use. And as soon as you're using a pin header for plugging in a transcoder box to the gcompsw, you might as well just have a separate box with RCA inputs that you connect the gcompsw to.

Of course, that's not to say that I wouldn't buy a gcompsw with good quality RGB SCART output! It seems like there isn't really any solution available that works properly with all component sources and supports both SD and HD without changing some knob or dip switch or something.

[Dopefish]

I'd be interested in purchasing a gcompsw with 8 inputs. If it had 2 outputs aswell, that would be an added bonus. Either way, I'll make a purchase when the next batch becomes available

[DirkSwizzler]

...And as soon as you're using a pin header for plugging in a transcoder box to the gcompsw, you might as well just have a separate box with RCA inputs that you connect the gcompsw to.

Eliminating wires is something I'm interested in. With 5 devices in my chain leading up to HDMI, wire management is becoming an issue. And The Garo's use of all sides for wiring makes it's placement a pain.

[Guspaz]

I think you could fit the required components for a transcoder into a SCART head, if you weren't doing all that other stuff the Garo does, and you could have component leads coming out of it, which means the number of wires ends up being the same as if you had a transcoder in the gcompsw and ran a wire from that to the gscartsw.

The only thing extra would be power: I don't think the gscartsw provides any power to the SCART ports (the typical 5v would be *from* the scart cable, not *to* it), so you'd need a power lead coming off it too, maybe a short female micro USB cable if you want to skip the linear regulator and just feed 5v directly in.

Basically, you could have one "cable" going between the gscartsw and the gcompsw (plus a power lead), like you would anyhow, and it wouldn't require extensive changes to the gcompsw, so people who buy the gcompsw wouldn't need to pay extra for something they wouldn't use.

What do you say, superg? I know you've got plans for that PS2 cable to solve the SoG issue, how about throwing a YPbPr-to-RGB transcoder into a SCART head to let people hook the gcompsw directly to the gscartsw?

[FBX]

Just search for stackable rca audio in ebay, I got a 50cm one for £3.49.
Just thought I'd share that incase anybody else has the same problem.
Hope it helps

This reminds me, I don't have one to test, but is it possible to pass composite video through the luma inputs on the gcomp? If not, a really cool revision in a future version would be to auto-detect and pass composite video through the luma inputs to the luma output.

[Ward1986]

Just search for stackable rca audio in ebay, I got a 50cm one for £3.49.
Just thought I'd share that incase anybody else has the same problem.
Hope it helps

This reminds me, I don't have one to test, but is it possible to pass composite video through the luma inputs on the gcomp? If not, a really cool revision in a future version would be to auto-detect and pass composite video through the luma inputs to the luma output.

Yeah like on some TVs

[Milspex]

Picked up my gscartsw at the post office today and wow it is an amazing product. Works right out of the box

[DirkSwizzler]

I think you could fit the required components for a transcoder into a SCART head, if you weren't doing all that other stuff the Garo does, and you could have component leads coming out of it, which means the number of wires ends up being the same as if you had a transcoder in the gcompsw and ran a wire from that to the gscartsw.

But adding cable to the device introduces new branches in design that quickly segregate the buyer pool. How long should the cable be? Should it have individually shielded wires? Not to mention that cable manufacture seems to be hard in general.

If it was a small box that plugged directly into the gcompsw output layout. Then you'd just need to solve power which seems easy enough to with a dc barrel splitter for the supply going into the gcompsw. And if people wanted to use the box without a gcompsw. It'd be easy enough to put rca female to female couplers on the component end.

[Guspaz]

Why would it segregate the buyer pool? It's a standalone cable with a SCART head on one end and RCA plugs (male or female) on the other end. It does not require any modification to either the gscartsw or gcompsw, and so does not segregate the buyer pool at all, people who want to connect their gcompsw and gscartsw together can simply get the cable and plug them together.

[DirkSwizzler]

Why would it segregate the buyer pool?

How long should the cable be? Should it have individually shielded wires?

If you expect that everyone would agree on both points. Then I'm wrong.

It does not require any modification to either the gscartsw or gcompsw

My second suggestion doesn't require modification of either existing unit. It doesn't even require a power adapter if it comes with a dc barrel plug splitter. In my head it would have 5 RCA male plugs on one side that perfectly line up with the gcompsw RCA female jacks for component output. And on the other side it has a female scart socket.

[Guspaz]

I thought about that (trying to put the conversion on the gcompsw side instead of the gscartsw side), but I don't think anybody makes board-mount male RCA connectors, only female. You could, of course, put the same kind of connector as the gcompsw uses and then just put RCA couplers on each one. The advantage of putting the converter on the other side is that getting a PCB connected to male or female SCART connectors is easy, because you can get a thick PCB made with pads where the internal SCART pins will land. That's how the NESRGB official SCART adapter works, and how most of RGC's SCART cables are made (they use a PCB for the components). If you could fit the entire thing inside the SCART head, you could also have a much cleaner solution than a bare PCB plugged into the gcompsw.

In terms of cable length or shielding, I'm not really proposing building a full length component cable in there, but something like short female RCA leads. That lets you use whatever cable you want. I think trying to build RCA jacks into a SCART connector is probably too hard.

Maybe your original idea could work, if the gcompsw could have some pin headers for a daughterboard to be attached with the transcoder, and maybe it could have some solder bridges so that you can break the original connection to the outputs... Then you could add the daughter board to the gcompsw and convert the outputs to RGBS (using yellow for sync), and then you could use a standard SCART-to-BNC cable to connect the gscartsw to the gcompsw (using some BNC to RCA adapters).

[superg]

I gathered more data. The TLDR is that the gscartsw_lite is super awesome and the Garo is outputting something weird.

I narrowed down the problem to be a bad interaction between the Garo and the original gscartsw 3.4 when the xbox is outputting 1080i, but not 480p. But ONLY if the Garo is plugged into any port other than 8 on the gscartsw. I must have had it plugged into port 8 before I hooked up the gscartsw_lite. When it's plugged into port 8 everything works just fine. If it's a port other than 8, it works until the xbox switches to 1080 and then it appears to get a garbage sync signal where I can make out parts of the menu but it's moving around a lot and switching back and forth to screen garbage. So yeah. I'm chalking it up to the Garo being goofy as Guspaz mentions and the gscartsw switching back and forth to port 8. Which just happens to be "good enough" if the Garo is already on port 8.

If I plug the gscartsw_lite into port 8 on the gscartsw 3.4 , the signal is stable with sync regeneration on and off.

If I plug the gscartsw_lite into any other port of the gscartsw 3.4, the signal is stable if I have sync regeneration on, and it completely drops the input if it's off.

I was already planning on hooking up all my non-csync sources to the lite and leave regeneration on all the time. So I'll just have to write down a note that the Garo is weird if I have issues again.

Thanks for your thorough testing!
One thing to mention is that v3.4 multiplexer IC's don't support HD signal and I have reports that some native VGA output with sync combiner circuit isn't forwarded by v3.4. I didn't look into it though. That said lite has HD video multiplexers which forward everything.

Eliminating wires is something I'm interested in. With 5 devices in my chain leading up to HDMI, wire management is becoming an issue. And The Garo's use of all sides for wiring makes it's placement a pain.

I absolutely agree on a cable management thing, my setup is a mess and I am trying to find a way to minimize cable count (combine power, reposition some consoles etc.)

The only thing extra would be power: I don't think the gscartsw provides any power to the SCART ports (the typical 5v would be *from* the scart cable, not *to* it), so you'd need a power lead coming off it too, maybe a short female micro USB cable if you want to skip the linear regulator and just feed 5v directly in.

What do you say, superg? I know you've got plans for that PS2 cable to solve the SoG issue, how about throwing a YPbPr-to-RGB transcoder into a SCART head to let people hook the gcompsw directly to the gscartsw?

Actually it does, you can easily get up to 500mA from output SCART pin 8. I did that just to be able to power some external devices.

This reminds me, I don't have one to test, but is it possible to pass composite video through the luma inputs on the gcomp? If not, a really cool revision in a future version would be to auto-detect and pass composite video through the luma inputs to the luma output.

Yes, it's possible. Yellow / Green / Blue / Red video connectors are physically equivalent and passed through the same circuit but the sync-based signal detection is performed only on Green / Yellow connector. You can even use gcompsw as RGBS switch given that you'll come up with your own wired SCART->RCA RGBS adapters.

Why would it segregate the buyer pool?

How long should the cable be? Should it have individually shielded wires?

If you expect that everyone would agree on both points. Then I'm wrong.

It does not require any modification to either the gscartsw or gcompsw

My second suggestion doesn't require modification of either existing unit. It doesn't even require a power adapter if it comes with a dc barrel plug splitter. In my head it would have 5 RCA male plugs on one side that perfectly line up with the gcompsw RCA female jacks for component output. And on the other side it has a female scart socket.

That's the reason I don't provide any cables with neither of my switches: quality varies, everybody wants different lengths and it's a nightmare for me to manage all this stuff, I'm already too busy with just the switch case assembly and unexpected soldering.

I soldered more lite units, currently I'm testing them, tomorrow assembly and these will be shipped on Monday.
So tired.

[RyuJose]

got my gscartsw_lite finally. I connected my Super Famicom, Sega Genesis, and PC-Engine scart cables to the switcher, then a male-to-BNC cable to my PVM. Only the SF and SG worked fine. Unfortunately, the PC-Engine didn't work.

PC-Engine set-up goes like this: PC-Engine -> db-grafxbooster -> genesis model2 scart cable -> gscartsw_lite -> male-to-female BNC -> PVM 20M2MDU.

I confirmed there's nothing wrong with neither the model2 scart cable nor the db-grafxbooster by not using the gscartswitch_lite, but rather connecting like this: PC-Engine -> dbgrafxbooster -> genesis model2 scart -> female-to-BNC -> PVM.

It seems to be the switcher Any Ideas?

[zegota]

This may have been answered, but it's hard to look through so many posts!

My consoles use a variety of different sync options -- some CSYNC, some Luma. If I plug these all into the lite, and then connect that to my xRGB with, say, a sync-on-luma cable, will it cause any issues?

[Guspaz]

On the subject of the Garo's wonky sync, and why I'm planning the sync (and RGB) bypass, here are some scope captures from Ste.

The yellow graph is the original sync signal that is being input to the Garo (the luma sync signal), the blue graph is the output of the Garo, which is an XNOR of the separate horizontal and vertical sync signals that are output by the LMH1251 chip (the chip on the Garo that converts YPbPr to RGBHV). The Garo is not correctly generating a sync signal.

Genesis 2:


Genesis 3:


Quite different, as you can see. I verified that bypassing the Garo's sync circuitry (pulling sync off luma instead of the Garo) solved any sync issues, although obviously without an amplifier the voltage was much lower from doing that. Ste confirmed that the RGB video output lines up to the YPbPr input, so there won't be any alignment issues from the bypass. Anyhow, maybe this will be useful to superg to see what the Garo is doing. It would seem to explain why the Garo is causing issues with the gscartsw lite sync regeneration.

[andykara2003]

Apologies, I've been out of the loop - will the gscart lite look better/sharper on a CRT than a gscart 3.4?

[FBX]

Apologies, I've been out of the loop - will the gscart lite look better/sharper on a CRT than a gscart 3.4?

Unless I'm mistaken, there's no LFP on the lite, so it will look pretty much as sharp as a direct connection would.

[DirkSwizzler]

It would seem to explain why the Garo is causing issues with the gscartsw lite sync regeneration.

That's the thing that perplexes me a bit. Sync regeneration on is a perfectly good signal. So the GARO must be outputting a signal that's at least detected by the gscartsw_lite. But I don't understand why sync regeneration off does something other than act as though I bypassed the lite entirely and got a garbage (but still visible) signal passing through the gscartsw 3.4

The only conclusions I can draw are either:
1. Sync regeneration off still messes with the signal a tiny bit.
2. Sync regeneration on is a factor on whether an input signal (on the lite itself) is deemed valid.

EDIT: And now that I think about it. I don't believe it would be #2 if sync regeneration off into port 8 of the gscartsw 3.4 works fine. So sync regeneration off must push it just over the edge of valid to invalid signal to the gscartsw 3.4

[superg]

got my gscartsw_lite finally. I connected my Super Famicom, Sega Genesis, and PC-Engine scart cables to the switcher, then a male-to-BNC cable to my PVM. Only the SF and SG worked fine. Unfortunately, the PC-Engine didn't work.

PC-Engine set-up goes like this: PC-Engine -> db-grafxbooster -> genesis model2 scart cable -> gscartsw_lite -> male-to-female BNC -> PVM 20M2MDU.

I confirmed there's nothing wrong with neither the model2 scart cable nor the db-grafxbooster by not using the gscartswitch_lite, but rather connecting like this: PC-Engine -> dbgrafxbooster -> genesis model2 scart -> female-to-BNC -> PVM.

It seems to be the switcher Any Ideas?

Some people are having problems with TurboGrafx / PC-Engine and my switch due to incorrect mods and incompatible sync level. I don't own db-grafxbooster and I can't say for sure what's going on there. I have Voultar kit for TurboGrafx though and once I ship everything I'm going to look what's going on.

[andykara2003]

Unless I'm mistaken, there's no LFP on the lite, so it will look pretty much as sharp as a direct connection would.

Damnit - I'll swap it out at some point..

[superg]

That's the thing that perplexes me a bit. Sync regeneration on is a perfectly good signal. So the GARO must be outputting a signal that's at least detected by the gscartsw_lite. But I don't understand why sync regeneration off does something other than act as though I bypassed the lite entirely and got a garbage (but still visible) signal passing through the gscartsw 3.4

The only conclusions I can draw are either:
1. Sync regeneration off still messes with the signal a tiny bit.
2. Sync regeneration on is a factor on whether an input signal (on the lite itself) is deemed valid.

EDIT: And now that I think about it. I don't believe it would be #2 if sync regeneration off into port 8 of the gscartsw 3.4 works fine. So sync regeneration off must push it just over the edge of valid to invalid signal to the gscartsw 3.4

I'll have to look with the scope Garo->gscartsw_lite to tell the reason what exactly is wrong and why it's not forwarded.
But based on Guspaz scope captures I see a lot of potential issues: timing problems, TTL output amplitude and shifted output.

Damnit - I'll swap it out at some point..

Or you can just disable LPF, but it all depends on your setup, LPF helps in a number of situations where you have worse cables etc.

[andykara2003]

Ahh thanks - how would I disable that? I only use official cables and good quality scart to scart cables..

I have version 3.0 - can I ask what the differences are between that and 3.4? And is the signal quality just as good/sharp with LPF disabled?

[superg]

Ahh thanks - how would I disable that? I only use official cables and good quality scart to scart cables..

I have version 3.0 - can I ask what the differences are between that and 3.4? And is the signal quality just as good/sharp with LPF disabled?

The difference would be power polarity change, overvoltage protection and DC restore circuit, maybe some other minor stuff. Disabling the LPF filter involves unsoldering the resistor and soldering one wire jumper, if you're comfortable with that let me know.