Testing Composite Blending from PC – VideoAMP + Antonio Villena VGA → Composite/

[Rion]

Hi everyone,



I’ve been experimenting with composite blending from PC using the VideoAMP (Pro Model)   together with Antonio Villena’s VGA → Composite/S-Video transcoder (with sync combiner)  .

The setup uses the VideoAMP’s VGA passthrough to feed into the transcoder while also outputting SCART RGB at the same time. This way you can see a side-by-side comparison of RGB SCART vs Composite directly from a PC on a CRT.

I recorded two Sega Genesis games to show the effect:

🔹 Streets of Rage 2 
🔹 Shinobi III: Return of the Ninja Master

 
Many of us are used to seeing consoles through composite, with its characteristic color bleed and blending artifacts. This method shows how you can replicate that look straight from a PC, while still keeping the option of razor-sharp RGB.

Curious to hear your thoughts?
Do you prefer the crisp RGB look, or the softer composite blending that feels closer to how many of us first played these games?

Equipment used: 
- Video out: 
  • VideoAMP V5 Pro (RGB SCART + VGA passthrough) 
  • Antonio Villena VGA → Composite/S-Video transcoder (sync combiner model) 
    → Connection chain: VideoAMP VGA passthrough → HD15 VGA cable → Antonio Villena transcoder → RCA male-to-male plug straight into the CRT composite input (for the cleanest possible composite signal) 

- Audio chain: 
  • Input      → 3.5mm audio from PC → Behringer HA400 headphone amplifier (input) with ground loop noise isolator 
  • Output 1 → 3.5mm to VideoAMP V5 Pro audio input (3.5mm) with ground loop noise isolator 
  • Output 2 → 3.5mm to dual RCA (L/R) into the CRT with ground loop noise isolator 
  • Output 3 → 3.5mm to LM-40 Passive Line Level Attenuator with ground loop noise isolator 

- Recording: 
  • Samsung Galaxy A33 smartphone (video capture) 
  • UGreen USB-C sound card (audio interface) 

[Zebidee]

Interesting. Nice to see the VideoAMP put to good use. I have an older version.

Not sure I'll ever be a convert to composite generally, but appreciate that some console games actually look better like that.

Can you please provide a pic or vid of the video input chain setup? Particularly the VideoAMP/VGA cable/transcoder/scart out?

[Rion]

Interesting. Nice to see the VideoAMP put to good use. I have an older version.

Not sure I'll ever be a convert to composite generally, but appreciate that some console games actually look better like that.

Can you please provide a pic or vid of the video input chain setup? Particularly the VideoAMP/VGA cable/transcoder/scart out?

Nothing really special here.

It's a little cramped behind the CRT so here is a more detailed chain connection description

- Video out: 
  • PC (R9 380X) -> DVI-I/VGA -> VideoAMP v5 Pro - VGA Input 
  • VideoAMP v5 Pro - Scart Output -> Scart Cable -> Scart Input 1
  • VideoAMP v5 Pro - VGA Pass-through -> VGA Cable -> Mini-VGA gender changer -> VGA → Composite/S-Video transcoder -> Composite input on CRT   

- Audio chain: 
  • Input      → 3.5mm audio from PC → Behringer HA400 headphone amplifier (input) with ground loop noise isolator 
  • Output 1 → 3.5mm to VideoAMP V5 Pro audio input (3.5mm) with ground loop noise isolator 
  • Output 2 → 3.5mm to dual RCA (L/R) into the CRT with ground loop noise isolator 

Here are some picture that shows how the - VGA → Composite/S-Video transcoder is connected.




I'm planning on attaching the VGA Female to Female gender changer with some #4-40*5+L Hex Standoffs Male-Female (outside thread 10mm).
Just haven't gotten around to it at this point (read: I forget where I put them  )

[Zebidee]

Thanks! Much as I imagined.

I created a pic to roughly mirror what you are doing (below), using VideoAMP with GreenAntz. I guess I should now actually try to drive the two outputs (RGB + component/YPbPr TVs) simultaneously.

This should be sufficient to drive both TV outputs. I wonder if it is able to drive two *arcade monitor* outputs at full brightness, or even one arcade monitor + TV, as arcade monitors expect more powerful RGB signals (~2vpp or more). Reviewing the THS7375 specs (I understand this is the amp chip used), I think it might!

[Rion]

Thanks! Much as I imagined.

I created a pic to roughly mirror what you are doing (below), using VideoAMP with GreenAntz. I guess I should now actually try to drive the two outputs (RGB + component/YPbPr TVs) simultaneously.

This should be sufficient to drive both TV outputs. I wonder if it is able to drive two *arcade monitor* outputs at full brightness, or even one arcade monitor + TV, as arcade monitors expect more powerful RGB signals (~2vpp or more). Reviewing the THS7375 specs (I understand this is the amp chip used), I think it might!

Just to clear this up 🙂

The standard versions (v1–v4 and the normal v5) can only run one output at a time so it's either SCART or VGA passthrough. The reason you’ve seen me running both SCART and VGA together is because I have one of the very few v5 Pro units, which supports true dual output. That version was never released widely and only a handful were built, and mine came out of some specific testing and feedback discussions with njz3.

Over time I also suggested a few smaller things that ended up being useful for everyone:

  • Adding a DIP switch to convert RGBHV → RGBS, so you don’t need an external sync combiner.

  • Adding stability tabs in the 3D case so the ports are held more securely.

  • Writing a wiki page for the VideoAMP when using CRT EmuDriver setup to make it easier for new users.

I only mention this because it helps explain why my setup doesn’t match what you’d normally expect from a stock VideoAMP since only the Pro model will do SCART + VGA simultaneously.

So in short:

  • Normal units = single output at a time.

  • Pro unit = special small batch with dual output.

  • Some features (like the RGBHV→RGBS switch and case reinforcement) came from ideas shared during testing.

I also don’t think njz3 or Bandicoot would be against making more Pro units if someone really needed that functionality. It never became the standard model, but asking them directly might be all it takes.

Edit:

Just as an aside, I’ve also tried to get a bit more attention on the VideoAMP outside of our usual circles.
Over the years I shared info, demo videos, and wiki links with Bob from RetroRGB, hoping he or one of his contributors might give it a look.

We talked about:

  • How the VideoAMP acts like a “UMSA on steroids” with SCART, VGA passthrough, audio input, sync filtering, and even GunCon support.

  • The usefulness of features like the built-in RGBHV to RGBS switch and the dual-output on the small Pro run.

  • How it connects to the wider CRT PC workflow, including CRT EmuDrivers, GroovyMAME, and other CRT focused distributions.

Bob was always polite and said he was really busy, and even mentioned passing it along to his Linux and MAME contact Dan who could better cover it.
So it is not that there was disinterest, but more that it never quite fit into his usual coverage focus, which makes sense given how niche CRT and PC setups can be.

I only bring this up to underline that I did try to push the VideoAMP into more mainstream visibility, but at the end of the day it mostly remains something shared within the PC CRT setups.

[Zebidee]

Just to clear this up 🙂

The standard versions (v1–v4 and the normal v5) can only run one output at a time so it's either SCART or VGA passthrough. The reason you’ve seen me running both SCART and VGA together is because I have one of the very few v5 Pro units, which supports true dual output. That version was never released widely and only a handful were built, and mine came out of some specific testing and feedback discussions with njz3.

...

I only mention this because it helps explain why my setup doesn’t match what you’d normally expect from a stock VideoAMP since only the Pro model will do SCART + VGA simultaneously.

Thank you. This explains a lot. The question about whether VideoAMP supports multiple outputs has been kicked around on BYOAC before, but never really answered properly.

I bought a VideoAMP v3.3 a couple of years ago, mostly for the EDID feature (forcing the GPU into 240p). However, I never got around testing for various reasons/excuses, including: my available PCs already run CRT_emulator so no driving need, GreenAntz transcoder development, TV repair/modding, and serious injury.

Anyway, I think VideoAMP is a great thing, and dual output widens it's appeal and usefulness.

Right now I have two CRT TVs, a component TV (Sony flat 20") and a modded dual RGB SCART/VGA input TV (JVC curved 20"). I'd love to be able to test them with VideoAMP, both running the same output simultaneously. I've added a pic below as a teaser.

Over time I also suggested a few smaller things that ended up being useful for everyone:

  • Adding a DIP switch to convert RGBHV → RGBS, so you don’t need an external sync combiner.

This is also a very good feature.

I have managed to get GreenAntz auto-detecting composite sync input. It works great when csync is the only input. No need for a switch.

However CRT_emulator, when generating composite sync, does not switch off the vertical sync output! These "ghost" signals overwhelm the XNOR logic, creating an inverted vertical sync pulse. This problem affects all the transcoders I have tested it with (not just GreenAntz), it seems to affect all XNOR-based logic Csync. So the vertical sync still needs to be "blanked" when using CRT_emudriver Csync, thus a manual switch is still useful.

This might just be a problem with my setup (Win7, HD6450), so interested if this problem (vertical sync still output when in csync mode) is universal or just affecting me. I posted about it here, but only heard crickets chirping. Maybe I should just ask Calamity directly.

  • Writing a wiki page for the VideoAMP when using CRT EmuDriver setup to make it easier for new users.

Great idea. Link?

I also don’t think njz3 or Bandicoot would be against making more Pro units if someone really needed that functionality. It never became the standard model, but asking them directly might be all it takes.

IMO these features should be in the standard model. Is there some reason to exclude them, like cost?

Over the years I shared info, demo videos, and wiki links with Bob from RetroRGB, hoping he or one of his contributors might give it a look.

We talked about:

  • How the VideoAMP acts like a “UMSA on steroids” with SCART, VGA passthrough, audio input, sync filtering, and even GunCon support.

  • The usefulness of features like the built-in RGBHV to RGBS switch and the dual-output on the small Pro run.

  • How it connects to the wider CRT PC workflow, including CRT EmuDrivers, GroovyMAME, and other CRT focused distributions.

Bob was always polite and said he was really busy, and even mentioned passing it along to his Linux and MAME contact Dan who could better cover it.
So it is not that there was disinterest, but more that it never quite fit into his usual coverage focus, which makes sense given how niche CRT and PC setups can be.

I only bring this up to underline that I did try to push the VideoAMP into more mainstream visibility, but at the end of the day it mostly remains something shared within the PC CRT setups.

Thanks again. I agree with these points and none of it really surprises me. VideoAMP is great, but you need to have a driving reason to use it. He gave some valuable feedback though, and all those points you listed are good ones. Partly mirroring Bob's comments, I feel that:

• The Csync switch is essential.
  
• Boosting video output to support two outputs simultaneously would greatly improve the appeal and "need" for VideoAMP.
  
• A better understanding of how people would likely use VideoAMP in their PC/CRT/arcade/retrogaming setups is a good thing.
  

I have a few ideas and insights that could help njz3 and Bandicoot with their development. For example, how to boost the THS7375 outputs, and/or tweaking the outputs for different situations (e.g. output to two arcade monitors vs to just one TV). I might also be able to help with the Csync.

[Calamity]

This might just be a problem with my setup (Win7, HD6450), so interested if this problem (vertical sync still output when in csync mode) is universal or just affecting me. I posted about it here, but only heard crickets chirping. Maybe I should just ask Calamity directly.

Sorry, I probably missed that or got me in the middle of some extra busy period. AMD's native c-sync is barely used by a few users as far as I know. I found it problematic to setup. E.g. if things don't go well it's hard to tell if the lack of sync is caused by c-sync or something else. If things break eventually and you need to do some configuration, you'll be back to separated sync without the proper wires. If you want to dual boot on Linux, no c-sync there. Etc. That's why I find a much better solution to handle c-sync externally.

Anyway, it's interesting what you mention of vertical sync not being turn off. I didn't know that. That's probably because it was assumed that v-sync wouldn't be wired. In fact that's how I connect it to my BVM, leaving the vertical bnc disconnected.

These are things worth investigating, having the right equipment, skills and time.

[Calamity]

I only bring this up to underline that I did try to push the VideoAMP into more mainstream visibility, but at the end of the day it mostly remains something shared within the PC CRT setups.

I think VideoAMP is fantastic as it solves several issues at once that no other device combines: amplification + detection + edid emulation.

The only thing I missed in the test unit that njz3 gently sent me is amplification for the vga passthrough. I don't know if this was included in later versions, but looked like great addition for my use case.

The usefulness of hardware EDID emulation as an alternative to software based custom video is way more questionable, but that's another topic.

It was also suggested to provide support for VideoAMP through Switchres, which I really would like, if only I had the time.

[njz3]

Hi,

Rion ping me to jump to this topic and answer about the simultaneous output feature.

The "standard" board has only a single THS7375 (gain ~x5) and there are 3 pots directly placed on the RGB lines of the amplifier.
Then, we added different impedances to the traces routed to each respective type of output (VGA, Scart, or RGBS) so that the final output amplitude match what is expected by the load.
For example, the VGA and Scart outputs have ~200ohm so that, when you plugged a 75ohm load, the final amplitude of the output signal is matching the expected 0 .. 0.7V range (you can actually boost a little bit the output signal to 110% of the input, sometime that can help).
If you plugged VGA and Scart outputs simultaneously, they will share the same video smplifier THS7375+3x pots, but then you have 2x75ohm load, dividing the output amplitude by 2x.

The "Pro" board has 3x THS7375, each output type has its own amplifier and get a separate output impedance.
I kept the 3 RGB pots only to tune the gains for the RGBS output (arcade monitor). VGA and Scart outputs have fixed impedance to get 1:1 amplitude with the VGA input.
Because it adds 2x THS7375, I cannot maintain the same price as the standard board (each chip already add ~2€ to the BOM + increased manufacturing costs/time).
At the end, the board price would be globally more expensive (+10€) while not everybody is interested by having simultaneous outputs.

I did a small batch of "Pro" boards, so if people wants they can ask me to get one.
Lately, I spent most of my time finalizing the jamma low-latency add-on for the videoamp (JammaMia board), so the VideoAMP design is almost frozen since a year or more.
We still welcome request for changes, provided they are simple enough to not "revolutionize" the board.

[Zebidee]

Hi njz3 

The "standard" board has only a single THS7375 (gain ~x5) and there are 3 pots directly placed on the RGB lines of the amplifier.
Then, we added different impedances to the traces routed to each respective type of output (VGA, Scart, or RGBS) so that the final output amplitude match what is expected by the load.
For example, the VGA and Scart outputs have ~200ohm so that, when you plugged a 75ohm load, the final amplitude of the output signal is matching the expected 0 .. 0.7V range (you can actually boost a little bit the output signal to 110% of the input, sometime that can help).
If you plugged VGA and Scart outputs simultaneously, they will share the same video smplifier THS7375+3x pots, but then you have 2x75ohm load, dividing the output amplitude by 2x.

A reasonable approach to sharing a single amp - but your impedance matching is gone and still only one output at a time.

The "Pro" board has 3x THS7375, each output type has its own amplifier and get a separate output impedance.
I kept the 3 RGB pots only to tune the gains for the RGBS output (arcade monitor). VGA and Scart outputs have fixed impedance to get 1:1 amplitude with the VGA input.

Because it adds 2x THS7375, I cannot maintain the same price as the standard board (each chip already add ~2€ to the BOM + increased manufacturing costs/time).
At the end, the board price would be globally more expensive (+10€) while not everybody is interested by having simultaneous outputs.

I did a small batch of "Pro" boards, so if people wants they can ask me to get one.

I imagine that the video output quality is improved as well, even for a single output - I'd pay a little extra for that alone. So I may just have to get one from you!

Lately, I spent most of my time finalizing the jamma low-latency add-on for the videoamp (JammaMia board), so the VideoAMP design is almost frozen since a year or more.
We still welcome request for changes, provided they are simple enough to not "revolutionize" the board.

I know just how you feel     Regardless, here are my suggestions (a little long, sorry).

I believe you can reduce the number of THS7375 amps required. Maybe just two are needed. The datasheet states in many places that the THS7375 can support two 75R outputs simultaneously. You'll just need a few more caps and resistors.

The "secret sauce" is to AC couple the inputs (put 0.1uF caps in series) and add "pullup" resistors from voltage supply (Vs). This allows you to set a reference DC voltage bias on the inputs and outputs. Mostly we're interested in the outputs for now. The amp will now try to maintain output voltage at that reference level (with 5.6V/V gain) even if you add a second output. This is covered p.23 of the datasheet (Figure 51).

For the VGA/SCART/SDTV 75R terminated connections, and assuming a Vs=5v, try pullups of 16.7M ohms.

"Quoting" myself as an aside to say: the datasheet authors did a crap job of explaining how they come up with their numbers. They provide an equation from nowhere but no explanation, provide pullup values without going through the steps, and you are left to reverse-engineer calculating the pullup values for yourself. The info is all there, if you read it about 100 times, look at it backwards and perform some algebra on a napkin. Anyway, back on topic.

For two simultaneous outputs from one THS7375, you'll want to AC couple each RGB output with a separate 75R and capacitor as well. Six sets, 3x2. This will reduce cross-interference and the cap will help maintain consistent voltage at the output. This is covered on p.25 (Figure 53).

For the arcade monitor, again AC couple the THS7375 inputs with 0.1uF caps, but this time try pullups of around 5.1M ohms. This should give you a reference DC-bias of about 3.5v for the output! This time there is no real need to AC-couple the output as it is for a single device. I'm not sure what output impedance you should set here, but I just went and measured one of my arcade monitors and the RGB inputs had 1K terminators. So, maybe pots (~2k to 5K?) in-series, only two pins connected (wiper + one, I think no ground is needed), on the RGB outputs. This would allow you to vary the impedance and voltage levels to suit.

An alternative, or additional, way to "tune" the amplifier gain is to put pots to ground right against the RGB inputs. Quite large pots, like 5 or 10M+, with a 1.5M-2M resistor as a base/floor value (you don't want it shorted direct to ground, and pulling the termination resistance too low will lose filtering benefits). The idea is for this to run this in parallel with the built-in 800k resistor to ground on inputs. Reducing this 800k value will affect the reference DC-bias voltage calculation (refer equation p.23), reducing the output gain without having to sink current at the output (it will sink a tiny bit more at the input, but this is small by comparison). These would be true "RGB gain" controls!

In terms of my own experience with GreenAntz and the closely related THS7374, AC coupling the inputs (adding caps and pullups) made a huge difference to the video outputs. I went from barely having enough for one 75R output at 0.7vpp, to having to scale it back significantly. In particular the voltage referencing makes the colour outputs much more *consistent* and *full* (but not oversaturated!) as varying demands are placed on the amp.

So, I'm not really an amplifier expert or guru. Everything I know about electronics is self-taught. I've just been working on/thinking about this over a long time. Writing all this down gives me some ideas about making a better video amp.

Please understand that I really don't know much about VideoAMP specifically either, except what I've been told and what I can physically see on the unit (v3.3) I have! I see the THS7375 has a couple of bypass caps for the Vs, but that is all. In fact I am just assuming this TSSOP-14 IC I've been looking at is a THS7375 (the numbers have been scratched off).

Hope this helps. Make a local copy if you think this may be useful. I am aware of sensitivities and may delete all this in a week or two to save it being "harvested" by unwelcome players.

EDITED: corrected pot values for reducing amplifier "RGB gain".

[Zebidee]

So, I went and did it. Then I took a video!





Using the VideoAMP v3.3, I took video outputs from the VGA passthru and SCART heads.

• The VGA output (via VGA M-M cable and GreenAntz v1.94 RGB-YPbPr transcoder) goes to Sony CRT TV with component inputs
• The SCART output (via home-brewed VGA-SCART cable) goes to a recently RGB-modded JVC

Even though VideoAMP v3.3 does not technically support 2x outputs, the picture looks great.

The Sony/component TV looks full, very sharp picture and colours clear and strong. Obviously the GreenAntz has its own internal video amp (THS7374) with the forementioned DC-bias voltage referencing of outputs, which means it will boost the signal even if the inputs are a bit low. The colours may actually be a little stronger than usual.

The JVC/RGB TV looks great too, colours clear and bright. The picture is perhaps slightly less sharp and bright, very slightly, but difficult to notice unless you hot-swap it back to single video output. EDIT: the sharpness and brightness came back fully after some minor adjustment to the color and brightness via normal TV menu.

Why? The JVC likes lower input voltage anyway, and I built 68R resistors into the RGB inputs to bring voltage down to the 0.5vpp it prefers (the JVC's jungle RGB input pins are internal, not standard for external input). These extra 68R resistors on the JVC's RGB may affect the "balance" of outputs from the VideoAMP, giving the GreenAntz/Sony a slightly larger "share" than the JVC. This explains why the Sony is slightly brighter. The two video outputs look great next to each other anyway. Any video output "balance" affects are small, very hard to discern, probably indistinguishable if the the JVC did not have those 68R resistors. The issue should disappear with VideoAMP "Pro" v.5 as it uses separate amps on the outputs.

I call this a success!





Some further explanation of the SCART:

I don't have a SCART-SCART cable, so improvised and used a VGA-SCART (now SCART-VGA) I already have, but in reverse. The VGA then connects to an alternate input I'd already prepared for the JVC TV.

The VGA-SCART cable is one I "rolled" myself years ago. It is extremely simple, all the connections are straight passthroughs, no resistors except a 150R on the blanking. It was designed for use as VGA to SCART, but works just as well for SCART to VGA! I had to swap the sync from pin 20 to pin 19 to match SCART-OUT pinout, but otherwise everything the same. Swapping the sync pins was super-easy because I build my SCART cables using female crimps and slide these onto the flat blades or pins used in SCART headers.


 


VideoAMP v3.3 does not support Csync input, which makes hot-swapping the unforgiving JVC back to single video output configuration potentially complicated. The JVC accepts pure csync input only, no "smushy" sync from twisting H&V together. I'd have to exit games, swap around displays and reset CRT_emudriver to Csync.

To completely bypass this potential sync-swap drama, I used one of Buttersoft's csync 15khz protection dongles. H+V IN, Csync OUT. Very useful device. Even though it is not used in the final setup, it helps occasionally for troubleshooting, so I included it in the pic below. The dongle is wrapped in paper tape as the 3D-printed enclosure falls apart after a while! Poked holes for the LEDs. I also needed two VGA gender-changers to match my cables on this occasion. Gender changers are attached in the pic.

[Zebidee]

This might just be a problem with my setup (Win7, HD6450), so interested if this problem (vertical sync still output when in csync mode) is universal or just affecting me. I posted about it here, but only heard crickets chirping. Maybe I should just ask Calamity directly.

Sorry, I probably missed that or got me in the middle of some extra busy period. AMD's native c-sync is barely used by a few users as far as I know. I found it problematic to setup.

Oh, truly sorry, I really don't expect you to read all my posts and respond personally. Rather I was surprised that nobody responded at all, lol! I'm sure you are busy enough already. My own bad, posted on the main GreenAntz thread, so it was a bit buried. I should put up a new topic on the GM forum branch. Thank you for responding.

Even so, I'd argue that the issue must affect a fairly reasonable number of people, at different times, even if they don't understand it. If you are outputting to a CRT TV then an option for native c-sync output is better than smushy sync (twisting H+V wires) or some external device, like an extron, or a csync circuit built into a SCART header (advanced!). JPACs invisibly solved the problem for arcade monitor users (you need the JPAC for the JAMMA interface anyway), but not for TV users.

E.g. if things don't go well it's hard to tell if the lack of sync is caused by c-sync or something else. If things break eventually and you need to do some configuration, you'll be back to separated sync without the proper wires. If you want to dual boot on Linux, no c-sync there. Etc. That's why I find a much better solution to handle c-sync externally.

Anyway, it's interesting what you mention of vertical sync not being turn off. I didn't know that. That's probably because it was assumed that v-sync wouldn't be wired. In fact that's how I connect it to my BVM, leaving the vertical bnc disconnected.

Yes - when using Csync, disconnecting the Vsync, to avoid random interference, seems a sensible precaution. I've given such generic advice many times before. With your BVM & BNCs you'll probably never notice the vsync issue. For my own part, I was mostly using Buttersoft's VGA dongle (outputs csync with no vsync) when testing H+Vsync vs Csync autoswitching on GreenAntz, so I missed it for a long time too.

For me, the issue became obvious when I was using native csync from CRT_emulator for the JVC TV RGB mod, and swapping back to the GreenAntz/Sony TV for troubleshooting. The GreenAntz lost sync! At first I thought I'd done something wrong, or some extreme interference on the Vsync wire. Only after hooking up the oscilloscope did I realise that it was more than just random interference, which shouldn't normally bother GreenAntz, but rather a full Vsync signal.

The point I trying to highlight, and you've already mentioned it, is that there are many different circumstances: sometimes we want Vsync connected, sometimes not, but it is not always easy to disconnect like a BNC. You can't just pull the V wire out of the VGA cable, or clip off the pins, or haul out and plug-in a different monitor. At first my only solution was going back to an older GreenAntz unit with a sync switch, until I figured to just "blank" (pull high) the GreenAntz Vsync signal with a wire (really just another kind of switch). Not everyone has those options.

Take SCART CRT TVs, for example. Probably 90% of CRT TVs will work OK with raw smushy sync (twisted wires). Problem with smushy sync is that, during the vertical sync interval, it doesn't invert certain pulses and completely loses many/most of the timing and information pulses (this is also where data like teletext lives!). Even so, most CRT TVs are clever (a better word maybe is "forgiving") enough to deal with this and give you a clear picture anyway. Affects vary from model to model, but many TVs will show a telltale little zigzag near the top of the screen. Anyway, if it works for you, and you're OK with the limitations, twisting H+V together and running that through a resistor is OK. Heck, you don't even need the resistor much of the time.

However, if you are fussy, or have a TV that is fussy, you may need to run with "native csync" and physically disconnect that Vsync line. The only other alternative is to run it through some external device (Buttersoft's convenient dongle to the rescue!). Never say never, but it is something I prefer to avoid, at least for any kind of long-term arrangement.

With a SCART or VGA cable input, disconnecting Vsync can be a lot more difficult than a BNC terminal. If you have a soldered cable, it could be very messy or even impossible. This is one of the reasons I make different VGA-SCART cables for different occasions (with/without resistors, Vsync plugs, auto AV-switching, external power input, audio, for example).
 
It's also a reason why I prefer to use crimp terminals on the SCART pins - no soldering, more flexible.

Back to the current example, this JVC TV is one of those 10% of TVs that is very fussy about sync! It demands good clean csync. Right vpp levels and no vertical sync weirdness. So I enabled Csync with CRT_emulator and disconnected the Vsync wire. In this instance I just pulled it off and re-crimped the Csync wire. Not a big deal, but no way back without soldering or re-crimping. On some other occasions I've made connectors using dupont (arduino) wires to make reversal easier.

Many of these situational dramas could be avoided if Vsync switched off when Csync is enabled. Most of the time I'd be happy to leave the Vsync wire connected so long as there is no signal on it.

[njz3]

Hi Zebidee,

The "secret sauce" is to AC couple the inputs (put 0.1uF caps in series) and add "pullup" resistors from voltage supply (Vs).

.. snipped your precious explanations that for sure I will save ...

First an introduction: Bandicoot and I are not professional electronics engineers. Worth pointing out that what we came with is purely done based on example (manufacturer datasheets), trial and errors.
We are simply two hobbyists that were trying to solve minor issues we have seen with other existing solutions, specifically the fact that no solution provides VGA input monitor detection. For arcade cabinets, there are known solutions like jammaSD, jpac or ultimarc amplifier, and maybe a few other solutions that I don't know of. For TV with RGB Scart inputs, there are UMSA and many DIY solutions which offers more or less the same level of functionalities. The VideoAMP was only a way to "unify" all those functionalities in a single board.

Regarding AC operation mode of the THS: we actually use the DC input/output mode of operation with 5VDC power, as explained in the datasheet p.21, figure 52 p.24 and described on the schematics of the evaluation board Figure 55 p.29 when you set it in DC operation.
We tried the THS7374 first (which is what almost everybody in our community use), and also I remember asking myself whether to use decoupling input (AC input operation). We picked the DC mode because it was simple and does not require noise rejection or any EMC filtering (see p22 about "hum" 50/60Hz noise rejection in AC mode) that might filter the video signal, loosing the "quality" of the edges that you can get from a purely RGB signal. The result was so good compared to the THS7374 that we kept the solution we have tried.

Please understand that I really don't know much about VideoAMP specifically either, except what I've been told and what I can physically see on the unit (v3.3) I have! I see the THS7375 has a couple of bypass caps for the Vs, but that is all. In fact I am just assuming this TSSOP-14 IC I've been looking at is a THS7375 (the numbers have been scratched off).

You are right, the caps are mainly for voltage input and filtering of sync signals.

Hope this helps. Make a local copy if you think this may be useful. I am aware of sensitivities and may delete all this in a week or two to save it being "harvested" by unwelcome players.

Ok, will do.

I imagine that the video output quality is improved as well, even for a single output - I'd pay a little extra for that alone. So I may just have to get one from you!

I can send one "Pro" board to you if you want (I have a few built available) at same price as a standard one.
If we could produce 1000 boards in a single batch, then the price decrease on the BOM would be so high that having 3x THS in DC mode, instead of 2x in AC mode would not even be a question ;-)
So far (and even if I try to make it more PCB assembly friendly), we only buy components by 10x or 20x and each board is still partially assembled by hand which takes time.