Page 12 of 13

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Jan 11, 2014 9:26 pm
by blargg
NightWolve, when you measure component values, are you removing the components or just putting the probes to them while in the circuit?

Re: YPbPr ouput mod for SNES consoles

Posted: Sun Jan 12, 2014 3:01 am
by NightWolve
blargg wrote:I have a specific question for the person who measured all the component values to map out the SNES video output circuitry (which BTW is a very good step to become more grounded in reality). Did you desolder the components before measuring, or just measure them as soldered in the circuit? ... NightWolve, when you measure component values, are you removing the components or just putting the probes to them while in the circuit?
So you were asking me about this (I was actually in the middle of answering you when you made your 2nd post):
Image

OK, short answer is no, you're dealing with surface mount components, so it's not easy to desolder, but this concern is also unnecessary. I've heard it before, but yeah.

Full/Long answer: We have the raw SNES schematics and that also played a role in creating the pretty schematic that I made. Let's take just the Luma/Y circuit and I'll explain:

Image

Resistors: Every resistor value was verified with the DMM check as matching the SNES schematic. Take the three 39 Ohm resistors, they came out exactly as 39 on the DMM. Also, with a magnifying glass, you could see 39 printed on them. So you have 3 forms of verification: 1) Schematic, 2) Label, and 3) DMM test. 470 Ohm resistor checked out as 470 with the DMM test as well. The 10k resistor checked out as 99XX Ohm, etc.

Capacitors: Well here there were a few problems. The SNES schematic is wrong about the 220 uF. That DC filter cap is composed of two 100 uF caps soldered in parallel on the top of the motherboard, so the intent was 200 uF, otherwise you could've used a solid 220 uF cap. Therefore, my pretty schematic reflects my actual motherboard of 200 uF. The tiny 10 pF ceramic capacitor, well, my VC99 DMM cannot handle capacitors smaller than 1000pF or 1nF, so I have to trust the SNES schematic in reporting 10 pF. Finally, what the SNES schematic didn't clearly show was what happens right before the signal leaves the Multi-AV out on its way to the TV. Well, there was a tantalum pulldown capacitor, and my DMM measured it exactly at 199 uF, meaning it was a matching 200 uF. This is new information and I know I didn't see it elsewhere.

Well, there you go. That was the process. You can see that I left extra height in that pretty schematic at the bottom because I was to add Credits, details about exact resistance measurements, like the 450 Ohm resistor for Chroma, the preferred value stated is 470, but I reported that actual resistance, etc. Details like that needed to be added. Anyway, I lost interest, got distracted, and to be honest, the heckling heckler here really pissed me off as well, so I never got back around to things. Everybody else in this thread was cool, MikeMoffit, ziggy, etc. but there just had to be one amazingly arrogant know-it-all that likes to f--k with people cause of some sort of elitist scum attitude - he's butt-hurt cause most of us interested in this are not formally trained electrical engineers or something like he apparently is! There's something going on with him, and I don't quite get it, but yeah.

Memo: Ultimately, the purpose of my pretty schematic is to help lead to a more perfect transistor-based amp for Red-Y and Blue-Y by allowing others to clearly see what ROHM/SNES engineers did. It's out of that hope that an engineer with a scope, SNES hardware and who IS hands-on will show up and be able to model the best circuit design by having all available information upfront and in a very clear manner, etc. That's the idea.

Re: YPbPr ouput mod for SNES consoles

Posted: Sun Jan 12, 2014 12:13 pm
by blargg
Thanks, that helps the schematic, which is valuable to have with the confirmed component values. You might notate two 100uF caps in parallel on the schematic, because non-capacitance aspects might be the reason for the pair (ESR for one). I'm pretty ignorant on that but it could be important information.

Re: YPbPr ouput mod for SNES consoles

Posted: Tue Jan 14, 2014 4:28 am
by kyuusaku
Image

Re: YPbPr ouput mod for SNES consoles

Posted: Fri Jan 31, 2014 5:58 pm
by SkinnyV
I was revisiting this page but still not sure if any circuit is now considered the correct one to use. Which one is now deemed the optimal one? I read almost all the pages but still not sure since most of the posts are about adding a removing a part by trial and error or why the previous amp is wrong. I'm looking at trying it on my snes.

Thanks

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Feb 01, 2014 8:16 pm
by Ziggy587
SkinnyV wrote:I was revisiting this page but still not sure if any circuit is now considered the correct one to use. Which one is now deemed the optimal one? I read almost all the pages but still not sure since most of the posts are about adding a removing a part by trial and error or why the previous amp is wrong. I'm looking at trying it on my snes.

Thanks
Honestly, save your time and money and disregard this mod. I wish I had. I know the SNES directly outputting YPbPr seems awesome, but if functionality is all you're after, an external device is a better option in my opinion.

From my own personal findings, YPbPr offers no visual improvement over S-Video on a CRT TV. And no, I DON'T mean that as an end-all statement for everything, but that has been my experience with retro consoles. If you're gaming on a CRT, just stick with S-Video where possible. Forget about YPbPr entirely.

For HDTVs, the only benefit I see with YPbPr (again, for retro consoles) is the simple fact that newer TVs and AVRs don't have S-Video inputs anymore (and if they do, they're super rare). If you fall into this boat, then yes, by all means, try and use YPbPr since your only other option is composite video. But I still recommend forgetting this mod and using an external solution like a transcoder or (gulp) the XRGB-mini or similar.

Now let me brace for all the people that'll tell me I'm an idiot for not thinking that YPbPr is superior to S-Video. edit: That's sarcasm btw, I know what the advantages are.

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Feb 01, 2014 8:44 pm
by tepples
One advantage of S-Video is that some TVs can sync to 240p S-Video but not 240p YPbPr. My Vizio VX32L is one of them; PlayStation and GameCube games that use 240p video don't display correctly through the YPbPr input.

I agree that YPbPr isn't a noticeable improvement over S-Video at 240p or 480i. Other than support on certain TVs that lack S-Video inputs, all YPbPr buys you are 480p, 720p, and 1080i video (which the Super NES doesn't support anyway) and more chroma bandwidth. S-Video's 3.58 MHz QAM chroma carrier has enough theoretical bandwidth for about 370 chroma samples per scanline. Normally, the Super NES outputs 280 chroma samples per scanline (256 picture and 24 border), and S-Video is just as good as component. Even DVD uses only 352 chroma samples per scanline. In hi-res background mode, there are 560 samples, which exceeds S-Video's theoretical chroma bandwidth, but developers of games that use hi-res mode expect chroma to smear anyway.

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Feb 01, 2014 9:06 pm
by blargg
S-Video may have a theoretical horizontal resolution exceeding that of low-res SNES graphics, but I can see some chroma smearing on SNES around ~180-degree hue shifts. I'm skeptical that it has the bandwidth for worst-case chroma signals.

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Feb 01, 2014 9:29 pm
by lidnariq
I think that S-video, unlike NTSC or component, never had an associated formal standard that specified exactly how the chroma channel's bandwidth (and other properties) should be treated. So I think you could find technically-compliant S-video receivers that had bandwidths all the way from the NTSC minimal and strictest 1.5MHz I and 0.6MHz Q, to ones that had sharp cutoffs at exactly the modulation frequency, to ones that use comparatively sophisticated upper side band demodulation techniques to get arbitrary bandwidth.

In any case, 3.6MHz of bandwidth will only provide 227.5×¾, or about 171 visible chroma periods per scanline. I have no idea if any TVs might use DSP to further exchange vertical chroma resolution for horizontal chroma resolution.

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Feb 01, 2014 10:40 pm
by NightWolve
Well now, lemme be the first to thank you for all of your contributions to this thread, and most especially this latest and very, very curious one right here... I think, however, it would strike some, myself included, as a little, oh, I dunno, racist, RACIST, RACIST!!! ;) As to the content, that's also curious, your attempt to generalize me (and still from a position of arrogance), either from just this encounter or I wonder perhaps because you also had a hard-on for me before I even got here; it did make me wonder, but no matter.

Anyway, I do humbly apologize for not noting that the 80 Ohm resistor was actually a "preferred" labeled 82 Ohm type from any general resistor kit, but understand that I just didn't realize nor ever expect that I'd have a nitpicking nutjob that would catch something so trivial like that, and harp/heckle me down over it and pretend that it was such a big deal, etc. I sure did learn my lesson though, I'll have to be WAY more careful on the ole Internet... Heh. Well, once again, thanks for making this thread a tad more "entertaining" and "informative" than what one might've expected to find having clicked in it! It's been charming to have made your acquaintance here.
blargg wrote:Thanks, that helps the schematic, which is valuable to have with the confirmed component values. You might notate two 100uF caps in parallel on the schematic, because non-capacitance aspects might be the reason for the pair (ESR for one). I'm pretty ignorant on that but it could be important information.
Could be, but I think the reason(s) for it is on the top of the motherboard, the manufacturer usually goes cheap with aluminum electrolytic capacitors, while on the bottom (less accessible) they go expensive with tantalums, and aluminums come in their standard (or "preferred") sizes, going from 100 uF to 220 uF. In general aluminum cap kits, there's no solid 200 uF cap, so the only way to get that value if you really don't wanna use a solid 220 uF, is to take two 100 uF caps and solder 'em up in parallel. You can find ANY size I believe if you shop at places like mouser.com/digikey.com, but with general aluminum capacitor kits, yeah, you won't see a solid 200 uF. Anyway, I'm a layman in this, so just an observation.
SkinnyV wrote:I was revisiting this page but still not sure if any circuit is now considered the correct one to use. Which one is now deemed the optimal one?
I wouldn't say we reached an "optimal" or "most correct" one just yet, really. But you have multiple options, so depends what you want.

a) Do you want cheap and basic ?? Then a simple transistor-based circuit is the way to go as you can walk into your local Radioshack and buy everything that you need.

b) Do you want to experiment with these op amp chips from these various online suppliers ? That's an option, if you're feeling up to it. They are said to amplify better than the basic transistor.

c) Do you want to avoid hardware-modding altogether, the fastest option, and use something that was designed by a professional electrical engineer and are willing to pay for it ?? Then Ste's cable is the way to go, a plug'n'play solution, pretty much. Since these SNES units were providing 75 Ohm ready RGB signals at the Multi-AV out, a special cable with general RGB-to-YPbPr circuity inside it was feasible. See: http://www.hdretrovision.com/ - Not sure if he's been selling these yet, though. So maybe some wait on this option will be needed.
Ziggy587 wrote:From my own personal findings, YPbPr offers no visual improvement over S-Video on a CRT TV. And no, I DON'T mean that as an end-all statement for everything, but that has been my experience with retro consoles. If you're gaming on a CRT, just stick with S-Video where possible. Forget about YPbPr entirely.
Well, to this I would ask, how many CRT TVs in your lifetime with both S-Video and YPbPr have you observed this finding and thus base this generalization from ? 3 ?? 5 ?? 10 ??

When it comes to my 32" Panasonic CRT TV with both S-Video and YPbPr, the YPbPr inputs are clearly far superior. The picture is sharper, there's less color bleed, and there is a lot less shimmering than with S-Video when there is scrolling/motion, etc. It's how I connect my BluRay player to it, and when I tried my SNES, it really did look like I was playing with an emulator. I was certainly happy with the results. There is the "green" problem though, but I already talked about that way back in this thread. Composite of course is just plain horrible once you see the differences, and add in the dot crawl effect, etc.

Now, unfortunately, I only have ONE CRT TV to base this finding on, or rather, "observation." The other cases are a Samsung digital TV which I can only compare Composite to YPbPr, and the winner is gonna be clear with that every time, and the other case is a YPbPr-to-USB device that also has S-Video, and in that case, the YPbPr is superior.

Anyway, I understand, I've seen this debate wage on, is YPbPr really better, but I wonder just how many TVs were sampled by the people who say YPbPr isn't even worth it compared to S-Video, etc.

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Feb 01, 2014 11:13 pm
by blargg
NightWolve wrote:When it comes to my 32" Panasonic CRT TV with both S-Video and YPbPr, the YPbPr inputs are clearly far superior. The picture is sharper, there's less color bleed, and there is a lot less shimmering than with S-Video when there is scrolling/motion, etc.
I'm surprised you saw shimmering during scrolling. I don't understand how that's even possible, unless the TV internally just mixed the S-video into composite before processing.

Re: YPbPr ouput mod for SNES consoles

Posted: Sat Feb 01, 2014 11:27 pm
by NightWolve
Yeah, actually, about the same as what my EE friend 'thesteve' would say :
S-Video is a type of Component and should perform close.
It does depend on how the TV implemented it.
If the TV mixes S-Video to Composite first, it will look like Composite.
If the TV processes the Luma and Chroma separately it should be really sharp.
Hence, if true, you'll want to use your YPbPr connections more times than not! In other words, the circuitry of the YPbPr processing is cleaner/better, while the S-Video circuitry might be pretty shitty on some TV brands, etc. This is all talking about the same TV that has both S-Video and YPbPr mind you, etc. Not comparing a Sony TV using S-Video and a shitty brand using YPbPr and saying, hey, the Sony looks better! That's not a fair comparison.

Note: The S-Video connection is still much better than the Composite one on this CRT TV of mine. When I got my S-Video SNES cable, it was a great improvement, but still, with YPbPr, it got rid of the shimmering entirely, well, mostly, and looks a lot sharper, etc.

Re: YPbPr ouput mod for SNES consoles

Posted: Sun Feb 02, 2014 6:28 am
by Ziggy587
NightWolve wrote:Well, to this I would ask, how many CRT TVs in your lifetime with both S-Video and YPbPr have you observed this finding and thus base this generalization from ? 3 ?? 5 ?? 10 ??
Enough, but if you want a honest answer, 3 CRTs plus 2 HDTVs. Most of the comparison was done on my Sony Wega KV27FS120 (CRT) and my Samsung LN40E550 (HDTV).
NightWolve wrote:When it comes to my 32" Panasonic CRT TV with both S-Video and YPbPr, the YPbPr inputs are clearly far superior. The picture is sharper, there's less color bleed, and there is a lot less shimmering than with S-Video when there is scrolling/motion, etc.
Sounds like your TV sucks. :wink:
NightWolve wrote:Composite of course is just plain horrible once you see the differences, and add in the dot crawl effect, etc.
You should consider picking up a Sony Wega. The one I have, composite looks amazing (compared to composite on other CRTs). I've never seen an NES look so good on a CRT before. In fact, composite looks so good on this TV, it led to some confusion. I had to pull apart my SNES composite/s-video cable to confirm that it was wired for true s-video and not just composite over S-video (as I've read some do).

I know, I know. Blasphemy! Ziggy said YPbPr doesn't look any better than S-Video and that composite looks amazing, burn him at the stake!

If you frequent local thrift stores and check eBay/Craiglist for local listings, you'll find a Wega in no time. Often times for really cheap, if not free, since most people don't want bulky and heavy CRTs anymore.

Re: YPbPr ouput mod for SNES consoles

Posted: Sun Feb 02, 2014 11:40 am
by blargg
Do your Sony TVs digitize the image? I've avoided those digital-on-CRT TVs like the plague. If yours is analog, I'm wondering how the NES (or SNES) look good on composite without turning off the CTRP in the service menu (which annoyingly it doesn't remember on power down, so you have to turn it off, then adjust another setting to hide the service menu, and don't adjust volume or it re-enables it, at least on the older TVs). Without it off, you get hanging dots. With it off, it looks like a smaller TV expanded.

Re: YPbPr ouput mod for SNES consoles

Posted: Sun Feb 02, 2014 12:52 pm
by tepples
lidnariq wrote:In any case, 3.6MHz of bandwidth will only provide 227.5×¾, or about 171 visible chroma periods per scanline.
Minus a bit for horizontal blanking; the active scanline is 52.148 microseconds long. There is actually 7.16 MHz of bandwidth on the chroma carrier (DC to twice color burst). And though AM uses twice the bandwidth of baseband, reducing modulator bandwidth to 3.58 MHz, QAM fits two channels (U and V) onto the same frequency. So you end up with 3.58 MHz bandwidth for U and 3.58 MHz for V. Now the Nyquist theorem states that a bandwidth of R corresponds to a sample rate of 2R. Thus 3.58 MHz bandwidth means 7.16 MHz sample rate, or 373 samples per active scanline.

I don't think sideband demodulation works with QAM. One sideband ends up carrying the sum and the other the difference, as I understand it.

That said, there are pathological signals for encoders that don't do proper filtering of U and V before modulation. This is easy to see on composite: my "tvpassfail" test ROM for NES has a worst-case example. I also have evidence that the Super NES's encoder uses shortcuts that produce artifacts similar to those of the NES. Such a pathological signal might even shimmer during scrolling. YPbPr might appear clearer on pathological signals because it doesn't need filtering, at the cost of HDTV compatibility.

Anyway, I have a Sharp CRT SDTV with both S-Video and component inputs and a Magnavox DVD player with both S-Video and component outputs. Want me to try to look at fine details through both?