Wii Transcoding Component to RGB flicker issue?

[VEGETA]

by TI IC you mean THS7374 or LM1881? why would THS7374 suffer from csync signal? it just buffers what it gets.

as of lm1881 it is obsolete, would suggest LMH1980 instead which should have 0 problems.

anyone to test these?

Neither. The LMH1251 (component to RGB IC)

well, proper circuit should have been designed using op-amps, preferably fast ones to deal with higher resolutions.

[DrJohn]

well, proper circuit should have been designed using op-amps, preferably fast ones to deal with higher resolutions.

My first YUV to RGB converter was designed using op-amps, and had a very sophisticated method to remove the sync from the Y too. Not easy, but I did it.

The problem is that when you then start moving to the HD resolutions the colour space conversion formula changes slightly. Easy for a semiconductor manufacturer to adjust in silicon, but it does make it harder to both detect and then adjust values on an essentially discretely built analogue system. Complexity goes up, which adds a lot of cost and I've seen that people won't pay for that extra complexity.

[kitty666cats]

I haven't seen many people who need to convert 720p or 1080p component to VGA, anyway... once you hit those resolutions, you're probably using HDMI out of your source device (with a HDMI to VGA, if you're going to a monitor with no HDMI). It seems like, in 2023, the most common need for component to RGB is retro gamers wanting to convert 480p from their Wii or OG Xbox - for use on PC CRTs.

[BazookaBen]

I haven't seen many people who need to convert 720p or 1080p component to VGA, anyway... once you hit those resolutions, you're probably using HDMI out of your source device (with a HDMI to VGA, if you're going to a monitor with no HDMI). It seems like, in 2023, the most common need for component to RGB is retro gamers wanting to convert 480p from their Wii or OG Xbox - for use on PC CRTs.

The only real use case is Wii U. It's only limited RGB over HDMI, so theoretically you might get deeper color over component. But you can still get solid picture from one of the many HDMI adapters that convert limited digital RGB to full analog RGB

But for the other possible use cases, that have both analog and HDMI: PS3 has been observed to have inferior color in HD over component, so you want to use HDMI there. And Xbox 360 already has native VGA output. Everything released afterwards is HDMI only so there's not even a question.

So yeah, I'd say anyone in 2023 designing a new YPbPr converter should only focus on SD color space

[VEGETA]

well, proper circuit should have been designed using op-amps, preferably fast ones to deal with higher resolutions.

My first YUV to RGB converter was designed using op-amps, and had a very sophisticated method to remove the sync from the Y too. Not easy, but I did it.

The problem is that when you then start moving to the HD resolutions the colour space conversion formula changes slightly. Easy for a semiconductor manufacturer to adjust in silicon, but it does make it harder to both detect and then adjust values on an essentially discretely built analogue system. Complexity goes up, which adds a lot of cost and I've seen that people won't pay for that extra complexity.

you mean using rec 601 conversion standard for colors? it is for sd content but is it that bad for higher resolution? is it really detectable by users?

anyone using analog video these days is using it strictly for SD stuff anyway.

here is for example my upcoming design: https://imgur.com/a/XnPRbnB

it is RGB SCART or VGA to component, composite, and s-video converter as well as sync stripper (H, V, C). still checking it now but could be finalized soon.

I think I used conversion equations for 601 sd stuff in my component circuit despite the circuit itself using opamp rated for more than 1080i bandwidth.


it is all analog to analog conversion as seen, fairly complicated compared to similar product as it has almost 3 products in 1 (scart converter, vga converter, sync stripper) and can output all formats at the same time. of course it needs to be checked to verify all these complicated operations but eventually should work.

I am thinking of pricing it at 100$, is it good?


____

where is your design though? can it be bought?

and what is your current choice for rgb to component and vise versa if you are not using opamps?

[kitty666cats]

https://www.js-technology.com/store/ind ... er=product

I linked DrJohn's Component to RGB earlier in the thread, here it is again.

On a side note, my buddy Shank (he makes videos about CRTs and Wii mods, I think) tested the VDIGI Wii VGA cable - no sync drops on bright flashes, no purple tint to the video. DrJohn's transcoder should do the same!

[VEGETA]

https://www.js-technology.com/store/ind ... er=product

I linked DrJohn's Component to RGB earlier in the thread, here it is again.

On a side note, my buddy Shank (he makes videos about CRTs and Wii mods, I think) tested the VDIGI Wii VGA cable - no sync drops on bright flashes, no purple tint to the video. DrJohn's transcoder should do the same!

what about retrotink ones?

also, what circuit does this one uses?

in my design (shown above), I am using opamp with rec601 matrix conversion using resistors to set the values.

I asked questions about sync if someone here can answer:

1- if using lmh1980, sync will be 5v ttl right? this is what i see in figures.
2- when passing sync signals through ths7374... they go as they are. meaning 75 ohm input will be 75 ohm output not ttl right?
3- when using ths7374 with 75ohm sync signals (converted from ttl by connecting 470R resistor)... should I put 75ohm termination and small ac coupling cap or just feed them directly? what about the output?
4- the final csync output in my circuit is passed through 470 ohm resistor and delivered to Y input, i think this is correct. right?

i know i can change stuff in design later on but wanted to do it properly first go.

[kitty666cats]

I think you're straying off topic a bit... this is about component->RGB transcoding, you're talking about your design which goes in the opposite direction.

RE: the RetroTink, it (and the Wakabavideo one) uses a much more simple design than the JS Technology one. DrJohn already described what makes his transcoder more complex in an earlier post in the thread.

[BazookaBen]

Here are some photos of the board of the Crescendo Systems RTC 2220.

Which definitely does not have any of the picture-loss problems described with modern transcoders

https://imgur.com/a/WwJ7exO

Spoiler

From everything I read on AVSForum and Curt Palme back in the day, this was the best YPbPr>RGB transcoder. Discontinued (and company closed) after the designer/owner passed away in 2014

I don't know if there is enough in these photos to be useful to anybody. If anyone wants to try and reverse engineer any of it, I can do a some probing with a multimeter. Like maybe with test patterns running from 240p Test Suite from a Gamecube running at 240, 480i, and 480p.

As far as I can remember, the 2220 was the last revision, with the revision before that being the 2200, and I think a 2000 before that

[kitty666cats]

Here are some photos of the board of the Crescendo Systems RTC 2220.

Which definitely does not have any of the picture-loss problems described with modern transcoders

https://imgur.com/a/WwJ7exO

Spoiler

From everything I read on AVSForum and Curt Palme back in the day, this was the best YPbPr>RGB transcoder. Discontinued (and company closed) after the designer/owner passed away in 2014

I don't know if there is enough in these photos to be useful to anybody. If anyone wants to try and reverse engineer any of it, I can do a some probing with a multimeter. Like maybe with test patterns running from 240p Test Suite from a Gamecube running at 240, 480i, and 480p.

As far as I can remember, the 2220 was the last revision, with the revision before that being the 2200, and I think a 2000 before that

A lot going on under the hood, there! Similar to DrJohn's in that regard - if you take a look inside the RetroTink COMP2RGB or the WakabaVideo component to RGB's, they are pretty bare bones in comparison.

[BazookaBen]

It also has pots for gamma control and some fine tuning h and v sync, so I imagine that explains some of the extra hardware.

Not sure what chip is actually doing the YPbPr>RGB translation

[kitty666cats]

It also has pots for gamma control and some fine tuning h and v sync, so I imagine that explains some of the extra hardware.

Not sure what chip is actually doing the YPbPr>RGB translation

I saw an LMH1251 in one of the shots

[Gunstar]

@kitty666cats - Did you notice any differences in colour space conversion with regards to John's transcoder vs OSSC or VDIGI? Having compared Wii output I much prefer the colours from RGB vs Transcoded Component (sadly I have no native Component CRTs). I tried doing the VGA Wii mod years ago but couldn't get the software to work, very tempted to try the AVE-HDMI mod (+DAC to a PC CRT) to see how it compares

[NewSchoolBoxer]

Whole circuit would cost about the same as the proper LMH1251 circuit if you limit opamp bandwidth to SD and ED

Keep in mind that OP amp bandwidth is usually specified as the bandwidth where the signal drops by -3dB, but for video you want less. For example, the filter of the THS7314 is specified with -1dB at 7MHz and that is a SD-only chip.

The opamp's filter fixes the realized bandwidth to 7MHz at -1db and -3dB at 8.5MHz like you're saying but the true opamp bandwidth must be closer to 50 MHz like its THS cousins at the same price point or it wouldn't be able to handle SD. The pricier THS7376 used in OSSC has -3 dB filters at 9MHz (SD), 16MHz (ED) and 35MHz (720p/1080i) and 150 MHz bandwidth for 1080p with filters bypassed. We might say it's true bandwidth is 150MHz. Maybe there's a better technical term.

With thanks to kitty666cats for the mention and also messaging me.

Yes, I designed a circuit using the LMH1251, actually this is the second circuit designed using this. First version didn't cover SD because of the Macrovision problem, but I already produced another product which would convert YUV to RGB SCART for interlaced video. It used discrete op-amps and was absolutely superb.

This new version takes some of the technology from the YUV to RGB SCART converter I did originally and also part of the YUV HD to VGA converter and combines them.

Why is this important? The LMH1251 simply can't handle video with Macrovision on it; it doesn't like the bogus syncs. Splicing those bogus syncs out and providing a clean sync to the LMH1251 is the best solution.

Secondly, there's a video above about the LMH1251 blooming in colours occasionally. Using the test film on DVD, Stargate, I've seen this too. It's to do with the input filter and it's own DC bias to the IC. After trying several different add ons and variations it seems the recommended values for the input filter are the best, as adjusting them then creates more problems elsewhere.

Welcome! That's pretty sweet you already completed my thought experiment, in both composite and separate sync, and defeating Macrovision is a whole other use I wonder if can be marketed. Thanks for the LMH1251 tip. A +12V supply versus USB surprises me but price is good.

I asked questions about sync if someone here can answer:

1- if using lmh1980, sync will be 5v ttl right? this is what i see in figures.
2- when passing sync signals through ths7374... they go as they are. meaning 75 ohm input will be 75 ohm output not ttl right?
3- when using ths7374 with 75ohm sync signals (converted from ttl by connecting 470R resistor)... should I put 75ohm termination and small ac coupling cap or just feed them directly? what about the output?
4- the final csync output in my circuit is passed through 470 ohm resistor and delivered to Y input, i think this is correct. right?

i know i can change stuff in design later on but wanted to do it properly first go.

All sync stripper chips expect to drive computer or professional equipment so all output TTL no matter what you pass in. THS7374 expects 1Vpp Composite from SCART and can't handle TTL input. It doesn't output TTL, it level shifts by 150mV to avoid saturation then doubles the voltage to give amped Composite/Luma or 75 ohm Csync our world invented. It expects to drive SCART televisions from RGB. You would need 470R just on THS7374 input to reduce TTL to 75R level. Output voltage is fine.

DC coupling is better from not creating another pole but it's a risk having no form of isolation from antique consoles. I'd add the small coupling cap. The amp has DC restoration circuity anyway. Output, you want 75R in series to get 75 ohm impedance followed by big capacitor like on page 26 of the datasheet. At least for most consoles. The big capacitor is already inside PS2. The capacitor forms a ~5 Hz high pass filter with the x2 75R inside the console and you want that high uF value to get below the 25 Hz PAL / 30 Hz NTSC frame rates to avoid field tilt. DC coupling naturally avoids this but you don't know what people connect your device to. Have to AC couple input and output when you don't control the full video chain. I think a few computer monitors do need DC coupling.

[BazookaBen]

It doesn't output TTL, it level shifts by 150mV to avoid saturation then doubles the voltage to give amped Composite/Luma or 75 ohm Csync our world invented. It expects to drive SCART televisions from RGB. You would need 470R just on THS7374 input to reduce TTL to 75R level. Output voltage is fine.

Unrelated question to the topic, but I built a buffer circuit for RGB for a Neo Geo MVS using a THS7374 on this board someone designed on a German arcade forum (values for components are listed on back side of PCB): https://oshpark.com/shared_projects/jzNmi8o1

Sync was way too weak with the values he listed. I didn't know what to change them to, so just bypassed the buffer and sent sync directly to the 220uf cap with a 470ohm resistor

But it sounds like, from what you're saying, it sounds like this PCB, on sync specifically, doesn't need 75ohm on output, and also is using way too high of a resistance before the buffer? And I'm also wonder what the resistor to ground should be. It almost looks like he got it backwards. Too much resistance on the sync line to the THS7374, and too little going to ground.

A pic of the completed board to his directions: https://imgur.com/6oHh3Nl

And after I modified it with the 470ohm resistor to bypass: https://imgur.com/vOsw5au

[VEGETA]

All sync stripper chips expect to drive computer or professional equipment so all output TTL no matter what you pass in. THS7374 expects 1Vpp Composite from SCART and can't handle TTL input. It doesn't output TTL, it level shifts by 150mV to avoid saturation then doubles the voltage to give amped Composite/Luma or 75 ohm Csync our world invented. It expects to drive SCART televisions from RGB. You would need 470R just on THS7374 input to reduce TTL to 75R level. Output voltage is fine.

DC coupling is better from not creating another pole but it's a risk having no form of isolation from antique consoles. I'd add the small coupling cap. The amp has DC restoration circuity anyway. Output, you want 75R in series to get 75 ohm impedance followed by big capacitor like on page 26 of the datasheet. At least for most consoles. The big capacitor is already inside PS2. The capacitor forms a ~5 Hz high pass filter with the x2 75R inside the console and you want that high uF value to get below the 25 Hz PAL / 30 Hz NTSC frame rates to avoid field tilt. DC coupling naturally avoids this but you don't know what people connect your device to. Have to AC couple input and output when you don't control the full video chain. I think a few computer monitors do need DC coupling.

so I better put 470R in series (terminated by 75R to ground, or else won't be a voltage divider?) to a 220uF cap to THS7374 amp. As for output, 75R in series with a 220uF cap to output pin itself or any route needed. can it be something like 10-22uF ceramic cap instead of big 220uF elec. cap?

instead of ths7374, one can use a good opamp. anyone tried it? i mean in voltage follower configuration which could amplify the signal to be exactly the same as input... so that I can deliver the same signal to multiple other circuits.

Unrelated question to the topic, but I built a buffer circuit for RGB for a Neo Geo MVS using a THS7374 on this board someone designed on a German arcade forum (values for components are listed on back side of PCB): https://oshpark.com/shared_projects/jzNmi8o1

assuming TTL sync level... he is feeding it to 430R terminating resistor and 6800R series resistor and this will make a voltage divider like so:

V_out = V_in * (430/(430+6800) which means V_out = 0.059 * V_in.
if sync is 5v then output sync will be 0.297v. is this 300 mV sync the so called "75R sync level"?

maybe if it is using a similar divider like RGB signals it would have been ok?

what issues did you suffer from using it?

can you do some measurements or better yet oscilloscope captures?

[kitty666cats]

Finally got around to it, tested out DrJohn's Component to RGB transcoder with a Nintendo Wii:


- No sync dropped sync during bright to dark / dark to bright screen transitions.

- No purple-tinted image


If you're interested in a transcoder that FINALLY alleviates these issues that plague all the other component->RGB transcoders, the link is here below!

https://www.js-technology.com/store/ind ... er=product

[BazookaBen]

Very cool!

One thing I'm curious about is his Component>SCART vs Component>VGA. Are they both Rec 601? The VGA one mentions "HD" so I don't know if that means it's tuned for rec709 or what

[Rulumi]

Having compared Wii output I much prefer the colours from RGB vs Transcoded Component (sadly I have no native Component CRTs).

The digital output signal the GameCube and Wii uses to convert it to analog is YCbCr 4:2:2, so Component will have good colors if handled correctly.

[kitty666cats]

Checking in with John. I'd imagine that 480i/240p/480p will be treated as Rec 601 & any higher resolutions as Rec 709. The VGA output can also output SD signals, btw. There is a dip switch for composite sync on the VGA out.

[kitty666cats]

John’s response (RE: me asking if it outputs Rec 601 or Rec 709 depending on which resolution it's being fed):

“Yes it does the switch of colours because of IC being used, one of the features. From memory of the data sheet it gives the equation rather than the standard. But certainly referring to making the switch.”


Again, I’m not lyin’… this is absolutely the best component to RGB transcoder readily available nowadays

[BazookaBen]

Again, I’m not lyin’… this is absolutely the best component to RGB transcoder readily available nowadays

Sure seems that way. Is he still making them or is this all just old stock that's going to run out eventually?

And if he is still making this stuff, he should produce a RGB>YPbPr converter as well. That's a much more commonly needed conversion stateside

[kitty666cats]

Again, I’m not lyin’… this is absolutely the best component to RGB transcoder readily available nowadays

Sure seems that way. Is he still making them or is this all just old stock that's going to run out eventually?

And if he is still making this stuff, he should produce a RGB>YPbPr converter as well. That's a much more commonly needed conversion stateside

He is still making these - they released quite recently and not many people have been aware of them! I helped talk him into making them. I think there's ample RGB to component transcoders today which do the job well, but I didn't think there were many component to RGBs that were truly good quality... that's what led me to suggesting John design one. I also convinced him to make the 1:2 SCART buffer the other year, for people who want to stream but don't want to get a expensive gscartsw just for 2 SCART outputs.

The word just needs to be spread, really - if only word could get over to RetroRGB or somewhere that someone who has a lot of reach could vouch for the unit's superior quality!

[BazookaBen]

The only issue is the website is a little slow. People might think the site's down or something and click out.

The word just needs to be spread, really - if only word could get over to RetroRGB or somewhere that someone who has a lot of reach could vouch for the unit's superior quality!

Bob has an account here, doesn't he?

[NewSchoolBoxer]

Sorry for brigading the topic but most of the circuit design discussion besides sync applies to Component to RGB. I think a legitimate reason for a Component setup is you don't have to screw with external sync and its many flavors.

It doesn't output TTL, it level shifts by 150mV to avoid saturation then doubles the voltage to give amped Composite/Luma or 75 ohm Csync our world invented. It expects to drive SCART televisions from RGB. You would need 470R just on THS7374 input to reduce TTL to 75R level. Output voltage is fine.

Unrelated question to the topic, but I built a buffer circuit for RGB for a Neo Geo MVS using a THS7374 on this board someone designed on a German arcade forum (values for components are listed on back side of PCB): https://oshpark.com/shared_projects/jzNmi8o1

Sync was way too weak with the values he listed. I didn't know what to change them to, so just bypassed the buffer and sent sync directly to the 220uf cap with a 470ohm resistor

But it sounds like, from what you're saying, it sounds like this PCB, on sync specifically, doesn't need 75ohm on output, and also is using way too high of a resistance before the buffer? And I'm also wonder what the resistor to ground should be. It almost looks like he got it backwards. Too much resistance on the sync line to the THS7374, and too little going to ground.

A pic of the completed board to his directions: https://imgur.com/6oHh3Nl

And after I modified it with the 470ohm resistor to bypass: https://imgur.com/vOsw5au

I'm not very good at working out PCB diagrams but this one is a design I understand and uses the other side for ground. Every hole (via) is a ground connection. Super easy then to work backwards to circuit diagram:

Spoiler

So it's an MVS Jamma board. I'm not familiar with arcade voltages other than they are high. 10 minutes of internet research led me to believe the R, G, B and probably Sync voltages are 4.1Vpeak. 4.1V in, 0.7V out with the voltage divider for R, G, B and sync is 243mV out. 4.1 x (1400/(1400+6800)) = 0.7. Yes, sync should be 300-450mV and I would say the 6800-1400 values are too high except if you need the current in the uA range. With common resistor values, I'd use 1% tolerance 443 and 91 instead for 4x less thermal noise on R, G and B and Rth of 75.

The 75 ohm output is because the output impedance from THS7374 is deliberately low, 0.7 ohm or even less in bypass mode, so the series 75 ohm resistor brings the impedance up to match the cable's 75 ohms and prevent reflections. I certainly don't think it's helpful on the csync line in the kHz audio range unless it has an awful lot of MHz video crosstalk. Nor do I think amping the already high sync makes sense, so I agree with your single resistor bypass strategy to avoid the opamp's distortion. The 220uF cap is a safety measure, up to you to avoid it or not. I read that arcades are DC coupled, which explains lack of input capacitors but could then debate the 220uF output ones being there. If THS7374 is getting destroyed from an arcade voltage transient, may as well DC couple the output too and save 50 cents.

Now where bypassing could matter is the group delay on the R, G and B, which is a negligible 3 ns in bypass mode but 70 ns with the opamp's lowpass filter on. This means video is arriving 70 ns after sync. Still within analog video tolerance but maybe an HDMI scaler's sampling phase gets tweaked. I'm glad you pointed out the questionable parts of the design. Too bad it got spread around in English too. Just cause an adapter board has been around for 5-6 years doesn't mean it gets a pass. Better amps than the cheapest 50 MHz tier but that's another discussion.

[NewSchoolBoxer]

so I better put 470R in series (terminated by 75R to ground, or else won't be a voltage divider?) to a 220uF cap to THS7374 amp. As for output, 75R in series with a 220uF cap to output pin itself or any route needed. can it be something like 10-22uF ceramic cap instead of big 220uF elec. cap?

instead of ths7374, one can use a good opamp. anyone tried it? i mean in voltage follower configuration which could amplify the signal to be exactly the same as input... so that I can deliver the same signal to multiple other circuits.

You can go the 1 resistor route or the 2 resistor route. Giving up 75 ohm matching when you need attenuation is justified with pure sync in 15 kHz range. The cable is too short to have kHz reflections. If you suspect MHz video crosstalk then 2 resistors can get the match pretty close. (R1*R2)/(R1+R2) = 75, [TTL sync]*R2/(R1+R2)=[75 ohm sync], where larger R1 is in series before smaller R2 in parallel to ground. At common values, you'd probably want 1240 and what's cheaper between 78.6 and 80 ohm at 1%.

You use a very small cap like 100nF on the input to THS7374 to AC couple since it has a small parasitic R and L. Too small a capacitor and you start attenuating low frequencies. At PAL 15.625 kHz sync frequency, the 100nF has -108.9 ohms impedance. Z = sqrt(470^2 + 108.9^2) = 482. If R is 470, now you want reduce it to be 457, so I suppose common 453 at 1%. I think 470 is still close enough given analog tolerance unless you decrease the capacitor.

The capacitor screws up the two resistor strat though. Could bypass THS7374 with sync like BazookaBen did and go straight to the 220/330uF cap, which is fine for any resistance. Then I suppose you could reduce the 470 to give its amped value? The optimal 2 resistors change to R1 = 625 and R2 = 85 but analog tolerance gives you leeway.

Oh and you want the AC coupling capacitor closest to THS7374, so after the resistor(s) to have it be most effective.

[BazookaBen]

'm glad you pointed out the questionable parts of the design. Too bad it got spread around in English too. Just cause an adapter board has been around for 5-6 years doesn't mean it gets a pass..

I don't think it was ever really intended as a mass market thing. It was kind of an experiment developed between a few users on a German forum who were trying to make an impedance-matched board for SCART TVs. But it works well for my consolized MVS as I can send it straight to my component converter that I use for all my other RGB systems

You can read the original thread here, Google translate handles it well: https://circuit-board.de/forum/index.ph ... t-THS7374/

So you see the user n00b made a revised version. I'm not exactly sure what the differences would be in the final video output, so I just went with philenotfound's board because it had fewer components to solder, and it wasn't clear from the conversation that n00b's looked any better for the final video output.

n00b's board: https://oshpark.com/shared_projects/R8NJ9UQ4

Then I believe n00b is also on this English forum as SVM and discusses it a little more. Seems his revision didn't work out for one user at least

https://www.arcade-projects.com/threads ... -rgb.6469/

But the output of the philenotfound board looks pretty good as far as I can tell, after I made the change to sync. Low pass filter off since I'm on a CRT

Twinkle Star Sprites looks oversaturated but I think that's just Twink Star Sprites. Other games look great, I can't spot any flaws. Very rich. Gamma seems in the ballpark of 2.2

[jd213]

Would be curious to see how an Electron Shepherd HDMI adapter + HDMI-VGA adapter looks. Might be good enough for some people.

https://electron-shepherd.com/products/ ... 51tOxZrM42

[BazookaBen]

Would be curious to see how an Electron Shepherd HDMI adapter + HDMI-VGA adapter looks. Might be good enough for some people.

https://electron-shepherd.com/products/ ... 51tOxZrM42

I'm not a fan of adding an extra analog-digital-analog step when it's avoidable.

[jd213]

Right, it wouldn't be ideal but the results might be acceptable. I only care about a handful of Wii games, and if I didn't have the Crescendo 2200 myself I might settle for something like this, unless it had some really bad flaws.