Which RGB SCART cables you should buy, and how I learned to stop worrying and love the CSYNC.

The goal of this article is 1) to provide a self-contained explanation of analog video synchronization signals which includes why and how they interact with the rest of the video process, 2) to provide enough information for people to choose the right cables for their classic gaming consoles which includes addressing the reasons for the strong opinions that exist on this topic. If all you want is the TL;DR, just skip to the end of the article for the readout.

My sincerest intention is to not only represent my own opinions along with the facts that drive them but also faithfully represent the viewpoint of those who differ with me.  It seems the issue of proper construction of RGB SCART cables, and concurrently the right of RGB SCART cables to even exist has many feelings, hyperbole and misinformation entangled within it.  

While a person's reputation and expertise is definitely a good reason to take their advice, you also need to consider whether that person is motivated by the same goals as you are.  I don't have a reputation to stake against some of those who are disagreeing with me, but I make my intentions and motivations clear, and will gladly accept feedback if any of my facts are objectively wrong.

I have not been without fault in unintentionally spreading misinformation on this topic in the past, but I have endeavored to embrace correction.  There is a "right" answer to which SCART cables you should buy, but some people aren't going to like it, and aren't going to like that I say it.  And that's okay.  The great thing about discourse is hearing the ideas of the people who disagree with you and seeing if your own ideas stand up to them or fall apart.  If you feel I have misrepresented or provided misleading information, I'm interested in hearing from you (as long as you're prepared to keep emotion out of it and maintain some degree of civility).

The Traditional Recipe For Analog Video

An analog display requires (6) elements to draw a color video image correctly: Red, Green, Blue, Luminance, Horizontal Sync, and Vertical sync.  All color analog video signals carry these values.  They're not always called by these names because the different formats combine them differently for transmission, but the elements are all there.  

• The Luminance voltage tells the screen how bright (or dark) the dot is
• The Red, Green and Blue voltages tell the screen how saturated each of those colors are for each dot
• The Horizontal Sync is a timing pulse that tells the display when it has reached the end of a line and to start drawing the picture on the next line below it
• The Vertical Sync is a timing pulse that tells the display when it has reached the end of a frame (or field) and to begin drawing at the top of the screen again.

What is Sync?

So in the section above there's a little bit of explanation about what sync does, but let's break it down just a little further.  When an analog video signal is sent to a display regardless of its method of transmission (Composite, S-Video, RGBx, YPbPr) you can think of it as multiple continuous strings of AC voltages.  The voltage changes in the different signals along this string corresponds with the dots (commonly but somewhat incorrectly referred to as pixels) as they're drawn in horizontal lines across the screen.  Higher voltage on the Luma signal means a brighter dot, lower voltage a darker one. Higher voltage on the red signal means deeper saturation of the color red on that dot etc...

So really when you start putting it all together the video signal is a continuous string of colored dots which, viewed as a single line doesn't make a lot of sense.  Something has to tell the display how to assemble this string into a rectangle.  The display has to know which dot along the line to begin drawing at the top and when it has reached the end of a line and to start drawing on the next line down.  That's the function of the synchronization or "sync" pluses that we're so concerned about.

The diagram below is my crude attempt to illustrate how this works. 

The VSYNC or "Vertical Sync Pulse" tells the display to begin drawing the line in the top-left corner of the screen. The HSYNC or Horizontal Sync Pulse tells the display when it has reached the end of each line and to begin drawing from the leftmost position one line further down.  The dotted lines directly after each HSYNC pulse are called the "horizontal blanking interval" named because of the way a CRT would "shut off" the electron gun to avoid lighting any phosphors while it moved back into position to begin the next line. The solid lines between frames are called the "vertical blanking interval" which gave the electron gun a chance to get back into position at the top of the screen.

How Different Analog Formats Combine Parts of the Signal (including Sync)

Sending the six required elements of a video signal can make it complicated to connect things. The more complicated a piece of technology is to use the less appealing it is to mass market consumers, or so says conventional wisdom.  For this reason during the analog era a lot of thought and technology was put into simplifying connections.  This gives rise to an axiom: the simpler the video connection, the more complex the implementation.

As mentioned above, the common methods of transmitting video - RF, Composite, S-Video, RGBs, RGBHV, RGsB, YPbPr etc... transmit the same information but each one of them tries to economize bandwidth and simplify connections by combining these signals in one way or another.  RGBHV is the simplest implementation because with the exception of Luma, each signal has its own channel, whereas at the opposite end of the spectrum RF is by far the most complex implementation of video transmission.

RGBHV - Combines Luminance with each of the Red, Green, and Blue channels.  Horizontal and Vertical sync pulses each have their own dedicated channel.  Also somewhat erroneously referred to as VGA. VGA is a specific resolution and frequency transmitted over RGBHV, therefore all VGA is RGBHV, but not all RGBHV is VGA.  It should also be noted that while RGBHV can be used for 15kHz video, it generally isn't. But as a discussion of video frequencies would likely detract from the point, I'm going to save that for later.

RGsB - Combines Luminance with each of the Red, Green, and Blue channels. Horizontal and Vertical sync pulses are combined with the Green color channel. Also referred to as "Sync-on-Green". This somewhat uncommon format seems to be favored by Sony.

RGBs - Combines Luminance with each of the Red, Green and Blue channels (exactly the way RGBHV does).  Horizontal and Vertical sync pulses are combined into a single channel known as "composite sync". The meat of our discussion will be about composite sync. This method is generally what is meant when simply saying "RGB".

YPbPr - Combines Luminance with Horizontal and Vertical sync into the "Y" channel, also called "Luma".  Rather than transmitting Red, Green and Blue values directly, they are mathematically transformed into "Pb" - the difference between Blue and Luminance, and "Pr" - the difference between Red and Luminance. The value of Red and Blue can be derived directly by comparison to Luminance, and Green can be derived by comparision to the other three.  This format is also known as "YUV", "B - Y/R - Y", and "Color Difference". It is also somewhat misleadingly referred to as "Component". The term "component" simply means that Red, Green, and Blue are transmitted discretely so all of the formats that occur higher up in this list are also "component" video. For the sake of clarity, this article will use the lower-case "component" when referring to the technical definition which includes RGBx and YPbPr, and upper-case "Component" if referring to the colloquial term for YPbPr.

S-Video - Combines Luminance with Horizontal and Vertical sync into the "Y" channel (exactly the way YPbPr does). Red, Green and Blue values are combined into a chromiance or "Chroma" signal. This format is also known as "Y/C" and sometimes conflated with "Super VHS" which was an early application of the format using higher fidelity videocassette recorders.

Composite - Combines Luminance, Chromiance (which is Red, Green and Blue combined), and Horizontal and Vertical sync into a single channel.  This is sometimes referred to as CV or CVBS.

RF - the term RF broadly means "Radio Frequency" but in this context it means a radio frequency modulated to carry Composite Video (Red, Green, Blue, Luma, HSync, VSync) and an audio signal.  Generally these signals are modulated in relation to a reference or "carrier" wave.

The Quality Potential Scale

Running parallel to the axiom that video formats that are simpler to connect are more complex to implement is another general rule that the more complex a video signal is to implement the lower the fidelity of its results. Formats that combine signals to simplify connections require something at the other end to reverse the process. 

Let's look at a very high level at what has to happen to turn an RF signal into a video image:

1. The RF signal has to be de-modulated into separate (Composite) video and audio signals
2. The Composite Video signal has to be separated into Chroma (color), Luminance (brightness) and CSYNC signals.
3. The Chroma signal has to be separated into Red Green and Blue values
4. The CSYNC signal needs to be separated into horizontal and vertical sync pulses.

There are exceptions, but as a general rule each stage a video signal has to pass through has the effect of degrading the signal fidelity - that is to say it's faithfulness to the original signal before it was all combined for transmission.  Quality generally improves each time a format eliminates a step in the process.

Composite Video looks better than RF because it only needs steps 2, 3 and 4.

S-Video looks better than Composite because it only needs steps 3 and 4.

The component video formats look better than S-Video because they only need step 4.

Some transition processes are more detrimental to the image quality than others because not all degradation directly affects the image.  Separating CSYNC from CVBS or Luma has negligible impact on the image because CSYNC is so distinct, and the pulses generally occur outside the viewable image during the "blanking interval". Separating CSYNC into HSYNC and VSYNC has no discernible impact on quality because the pulses are very simple to distinguish.  On the other hand when separating Composite Video into Chroma and Luma, even the best engineered solutions still mistake part of the Luma signal for Chroma information and vice versa causing a whole list of well documented image artifacts and a very predictable loss of clarity.  The same is true for separating Chroma into Red, Green and Blue - some blending/bleeding between colors always occurs.

When it comes to the transmission of component video formats the very faithful reproduction of all six requisite elements can be derived mathematically from the signals being sent.  All except for YPbPr which has to use filtering to separate CSYNC and Luminance, but again since the sync pulses generally occur outside the drawn image area during the blanking interval any degradation of the signal caused by this process is invisible.  The upshot of this is that component video formats are all basically equivalent in potential fidelity, and represent the best potential quality from analog video.

While ALL analog video formats are subject to degradation during transmission either from poor shielding, poor connection, improper implementation or strong external interference, component video formats are the only types which do not inherently degrade fidelity by the way they are processed.

To RGB Or Not To RGB

It needs to be said that formats which DO inherently degrade image fidelity by the way they are processed (RF, Composite, S-Video) aren't "bad". "Lower fidelity" is not the same as "low fidelity". If those formats were unusable, they wouldn't exist.  The only wrong answers to "what format should I use?" are those that either try to make you feel bad because you're not using the best format possible or those that try to make you feel bad FOR wanting the best format possible.

The desire for the cleanest clearest video output from retro games consoles is perhaps the most significant reason to prefer component video formats, but it's not the only one.  Historically component video formats have also been more compatible with accessible scalers and video processors that properly treat the non-standard "240p" resolution output by game consoles designed up to the mid 90's.  Put another way, it was easier to get from component video formats to something that looked good on a modern digital TV or high resolution PC monitor than it was to do the same with S-Video, Composite or RF.  Nowadays, however, there are several affordable options for being able to use those signals so it's to be expected that there is renewed interest in them.  

Aside from the fact that many consoles will not output component video without modification and many people prefer to keep their consoles as original as possible, below are some additional reasons why you might want to stick with non-component video formats.

S-Video can produce an image so sharp and clean it can be difficult to see the quality difference with RGB without side-by-side comparison.  The biggest reason to abandon S-Video for RGB isn't the quality boost, it's really the availability of equipment that accepts and properly processes the signals.  If you're playing on a CRT with S-Video inputs, you're only going to get a tiny quality bump in color separation going to RGB.  S-Video does have some huge caveat emptor surrounding it, however.  The market is flush with devices that accept S-Video but process it as Composite and cheap cables that have an S-Video connector that's just connecting Composite video output to both the Y and C pins.  If your S-Video output doesn't look significantly better than composite, you've probably been duped.

Composite Video can look very clean if implemented properly.  While it is never going to look as clear and clean as S-Video or component video, good implementations can function with a minimum of artifacts. What's more there are many games that apparently made intentional use of the inherent blurriness and inability to accurately separate colors to create visual effects such as simulated transparency.  The best example is a technique called "dithering" which outputs a pixel-by-pixel checkerboard layer over top of a background or moving object. When composite video is processed it can't clearly distinguish one dot from another and the result is a blending of the foreground checkerboard and whatever is behind it into a blurry amalgamation of the colors of both which approximates a transparent foreground object.  This is sometimes cited as evidence that these games were "never meant to be played using [component video]", however I disagree with that sentiment.  Neither format is right or wrong as evidenced by the fact that the dithering technique was also implemented in arcade games from the 90's which were hard-wired for component video where the developers would have had no expectation of Composite Video "blending".

Even RF can look quite good with certain consoles and with pre-NES/SMS consoles it's the only way to play them without modification.  This works best if you've got an old CRT Television with a Type F coax connector.  In order to use RF with a modern TV, you'll need an intermediary RF demodulator device to transform the RF signal into Composite Video.

SCART - the four letter word that has five letters

SCART or EuroSCART is a 21-pin European analog audio-video connector standard defined by EN 50049.  It was in common use in Europe and the UK but it was not adopted for use on other continents. The slight exception to this is that the physical connector was used in Japan but defined under a different and pin-incompatible signal specification called JP21.

I have to invoke the axiom again that the simplest connection is usually the most complex to implement, and the most complex to implement usually results in lower fidelity.  Well SCART was sort of an attempt to have one's cake and eat it too.  It was one cable that could carry oh so many signals.  In theory it addressed the problem of consumers being reluctant to adopt a complicated connection.

SCART, like any physical connector has it's pros and cons, and some of those cons are at the heart of the issue I hope to tackle with this article. 

Pros:

- A generous number of pins/channels allows the same connector to carry multiple formats

- Had the equivalent of Audio Return Channel (ARC) 30 years before HDMI was invented.

- A single plug greatly simplifies the complexity of connecting a source to a display

- Supports component video formats like RGB

- Carries audio and video in a single cable

Cons:

- The connectors and sockets were made to varying degrees of quality. Many cheaper sockets wear out and lose retention force after a relatively low level of insertions and removals.

- The standards were, in some cases, too broad and led to confusion with implementation (more on this in a bit).

- An RGB SCART and Composite SCART cable can look exactly the same from the outside.

- No commercially sold North American televisions have SCART inputs

Why Did SCART Become A Thing During the Retro Gaming Revival of the last Decade and a Half?

While I can't answer that question definitively, I can share my experience and what I think.  In North America while Europe and the UK had been enjoying RGB capable sources and displays for decades, we were lucky if our televisions even supported Composite Video.  S-Video was designed in 1987, but I never even saw a TV capable of using it until the mid-90's and even then it was financially out of reach.  When the NTSC was trying to figure out how to deal with the imminent introduction of EDTV (480p) and HDTV (720p/1080i) I remember reading (I think it was Popular Science) that there was a debate over whether to adopt the EuroSCART connector, or to just add four more RCA plugs to allow cables to carry RGBS the way professional broadcast equipment (i.e. PVMs and BVMs) did with BNC connectors.  However they finally settled on YPbPr - one less RCA plug but totally incompatible with RGBs signals.  

So in North America, even though our game consoles had the capability of RGB video output basically as early as the Sega Master System (1986), by the time we started getting TVs with discrete color inputs they weren't compatible with our consoles.  If memory serves the Playstaton 2 (2000) was the first console released in North America that supported YPbPr video output.

So retro game enthusiasts constantly reaching for the next rung on the ladder of graphical fidelity were faced with some choices.  On the one hand they could cut up their consoles doing component video mods (I saw quite a few Neo Geo AES systems sold this way), OR they could try to convert the existing RGBs signals from the console into YPbPr using an external transcoder.  If they chose the latter they were forced to solve the problem of how to physically connect them - create brand new one-off custom cables for every console, then find/make some generic connector to a transcoding device (like an XRGB), OR use RGB cables that already existed with transcoder connectors that already existed.  It's not hard to see why the latter was more appealing.

 

My experience suggests SCART was adopted by the retro gaming scene outside of Europe/UK because it was there and it was already 95% of what we needed. SCART carried both audio and video. Official cables already existed for many consoles. If you standardized on SCART you could use commercially available switches to connect multiple consoles.  All the parts required to make your own cables were readily available.  Devices like XRGB had SCART adapters already.  There weren't really any alternatives that ticked that many boxes.

 

Could it have gone a different way?  Sure, if HDRetrovision fell through a time warp and started selling Genesis and SNES cables in the mid-90's, SCART would never have gotten a foothold in the scene.  It's also possible that we could have standardized on another similar connector like DB15 used by PCs for analog video at the time, but considering it had no provision for carrying audio, Composite Video, or S-Video it would have resulted in a completely out-of-spec implementation and arguably would have led to even more complaints than what's currently going on with SCART.

 

So Why Is Sync Source Such a Big Deal With SCART?

When I listed the "cons" of SCART in general, I left off a couple of juicy nuggets that apply specifically to the retro gaming scene.  

So the first and biggest reason many people detest SCART is a lack of consistency of implementation that can cause anything from the cable just failing to work up to actually damaging your console, scaler or display.

 

SCART can carry RGBs, S-Video, and/or Composite Video signals.  A single SCART cable can carry up to two of these signals simultaneously. Pin 20 can either be used exclusively for a composite sync signal, or it may be wired to a CVBS (Composite) or Luma (Y) signal.  This means that the sync pulses for an RGB video signal can either be exclusively sync (CSYNC/composite sync), derived from Y (Sync-on-Luma) or derived from CVBS (Sync-on-composite).   

An excerpt from the EN 50049 official SCART definition showing all of the signal types the specification allows on "terminal" 20.  While the terminal was clearly meant for carrying Composite Video or Luma (Y) as a full signal not merely as a sync source, it is also designated as a source for pure CSYNC (implicit with RGB). "When the signal on this terminal is exclusively a synchronization signal" is a clear description of CSYNC - "exclusive" means no other signal is present.

Sync-on-composite is a common source of RGB sync in analog consumer goods in the UK and Europe. It would probably be easier to understand if we called it “Sync from Composite Video”. It's the most prone to cause video noise - which can exhibit itself as a "checkerboard" pattern.  The composite signal is strong and carries a lot of high-frequency waves which can cause parasitic capacitive coupling with the Red, Green and Blue lines. In laymans terms some of the electrical energy from Composite Video extends like a radio wave beyond the cable and into the other nearby cables. This tendency of sync-on-composite to produce a degraded picture can be greatly reduced by properly shielding the Composite Video signal to isolate the electromagnetic waves and diminish their ability to interact with the other signals. However the ENTIRE path of the Composite Video signal needs to be shielded - if you plug a fully shielded cable into a switch with poor shielding, the Composite Video will likely couple within the switch and produce interference.

Sync-on-luma exists as a sort of middle ground between sync-on-composite and pure composite sync. Again it’s maybe more easily understood if you think of it as “Sync from Luma”. The Luma channel carries the brightness along with the composite sync signal, it's similar to CVBS but it lacks the chroma subcarrier that gives Composite Video its color.  In fact if you plug a Luma signal into a Composite Video input, you'll get a fully functional black-and-white picture. The chroma subcarrier is what is responsible for most, if not all, of the potential of CVBS to cause interference. While it is possible for a poorly shielded Luma signal to capacitively couple the Red, Green, and Blue lines, since those lines all carry the same Luma signal such coupling would theoretically be reinforcing rather than destructive and would result in a marginally brighter image.

Composite Sync (CSYNC) is a pure sync pulse sent along a dedicated channel.  To keep this one separate in your mind, think of it as “Sync Composed of Horizontal and Vertical”. Its frequency is too low to pose a serious risk of coupling the video channels even if it's completely unshielded, and as the pulses occur outside of the visible area of the screen you wouldn't be able to see them if they did.  CSYNC is generally compatible with every scaler and display out there.  Even if the scaler or display is expecting to derive sync from a Composite Video source, CSYNC will still pass through its filtering or "sync stripping" stage and work just fine.  The opposite does not hold that devices or displays expecting CSYNC can handle sync-on-composite or sync-on-luma.

If the sync source is sync-on-composite, or sync-on-luma, the extraneous information needs to be removed from the signal somewhere in the chain. Sometimes that's done in the SCART cable itself with a "sync stripper", sometimes the scaler/display has the necessary circuitry built in. Sooner or later, though every sync source becomes CSYNC.

 

Because not all sources (consoles) output CSYNC, and not all displays/scalers are capable of separating CSYNC from Composite/Luma there is no such thing as a one-size fits all approach.

The (Potential) Dangers of using CSYNC

Firstly there is no danger whatsoever from properly implemented CSYNC.  All of the danger with regards to CSYNC involves improper implementation and improper application.  There are essentially two things that can go wrong which can damage equipment.  The first is sending TTL sync to equipment made for 75 Ohm sync, and the second is using the wrong region cables on a couple of very specific consoles.

Sending TTL CSYNC (High Voltage) Instead of 75 Ohm CSYNC

Unlike CVBS and Y, CSYNC has more than one common voltage level - namely TTL sync and "75 Ohm" sync.  TTL sync is also called "logic-level" sync and it is generally in the 5 Vpp range.  That's read as "5 Volts peak-to-peak" in reference to the way the waveform appears when visualized on an oscilloscope.  The difference between the highest and lowest point on the waveform is is 5V.  TTL sync is the level used by PCs to output RGBHV.  It's also used with certain professional broadcast equipment and in arcade cabinets.  However it is not generally appropriate for consumer TVs or most scalers.  If you refer back to the SCART standard, CSYNC is only supposed to be transmitted at 300 mVpp.  

The SCART standard defining 300 mVpp (.3 Vpp) as the voltage standard for CSYNC.

Not only does sending TTL Sync over SCART a violation of the standard by being fully 15x the expected voltage, it is also very likely to damage the image processing hardware on the receiving end. Even at this multiple of expected voltage, damage may not occur immediately and instead only become evident after a period of use.

So how do people "accidentally" send TTL Sync over SCART?  Let's look at the example of the Sega Genesis/Mega Drive console.  That console outputs TTL CSYNC at the multi-out port.  The official RGB SCART cables made by Sega contain the necessary capacitors and resistors to lower the CSYNC level to the 75 Ohm range to comply with the SCART standard.  Sega's design choice was probably motivated by the cost savings achieved by placing fewer components on the Mega Drive PCB.  They considered RGB a very niche use case during the Mega Drive's commercial life.  It made financial sense to place those components in the cable itself since the vast majority of customers who would never use RGB would never benefit from having those components inside the console.  Where this becomes a problem is when someone who is trying to make their own RGB SCART cable for the Genesis isn't aware of the different voltage levels and just wires up the signals directly which outputs TTL CSYNC from the cable.

Another place where this has happened historically is with arcade conversion hardware such as "superguns".  Because Arcade hardware generally operates on TTL sync and some of these conversions are targeting computer monitors rather than commercial televisions and scalers, they directly output TTL CSYNC.  Again, people wiring custom cables and unaware of the output levels and the need for 75 Ohm sync voltages have inadvertently sent TTL CSYNC into equipment that can't handle it.

This confusion and ignorance about the proper voltage levels for CSYNC can also lead to the opposite, but not dangerous, problem of lowering the CSYNC voltage too far by inadvertenlty attenuating the signal twice.  If, for example, your Genesis has an RGB mod that's already outputting CSYNC at 75 Ohm levels and you attach a SCART cable which is designed to lower a TTL CSYNC signal, the level can be dropped so low that the scaler or display can't detect it and your display can't properly synchronize the image.  Because consoles like the Sega Genesis also output color lines at higher-than-spec voltage (with the expectation that components in the cables will attenuate those as well), the most obvious sign that you've double-attenuated the signal is a very dim or dark picture.

Somewhat adjacent to the problem of directly sending TTL CSYNC to equipment that can't tolerate such relatively high voltage is the problem of improperly made cables which include attenuation but possibly not enough.  For example, it's possible that the wrong resistor value was used so that the cable is outputting 3Vpp instead of 5, but 3Vpp is still too high and could damage equipment.  

Based on longstanding practices, chip tolerances in datasheets and comparison to official SCART cables, as long as the CSYNC peak-to-peak voltage is between 300mV and 700mV it's generally considered "safe".  All reputable cable makers design their cables to respect this safe range.

Using The Wrong Region Cable With Specific Consoles

This is a particular issue with the way Nintendo implemented RGB output on certain of it's PAL consoles vs NTSC consoles.  In PAL regions, Nintendo consoles do not output CSYNC, but their NTSC counterparts do. This means that for a European SNES, the best option available is sync-on-luma, but for a North American SNES, you can use a CSYNC cable.

 

Here's where it gets dangerous: On NTSC Nintendo consoles pin 3 of the Nintendo multi-out is the CSYNC pin which is wired to pin 20 of the SCART plug.  Remember CSYNC signals are ideally around 1/3 of a volt.  On PAL Nintendo consoles instead of CSYNC, pin 3 is the "Aspect Ratio" signal - which sends 12V to the display to tell it that the SNES picture should be displayed as 4:3. On a PAL cable, pin 3 of the Nintendo multi-out is wired to pin 8 of the SCART plug.  If you accidentally plug a CSYNC RGB cable made for an NTSC Nintendo console into a PAL Nintendo console, it's going to dump 12V into an input that's only meant to handle 1/3 of a volt.  This is enough of a differential to usually do instant and permanent damage.  

This is a corner case to be sure as it only applies to certain Nintendo consoles (SNES and Gamecube, I believe) and is an inconsistency that Nintendo created rather than anything to do with the SCART or CSYNC standards.  However the danger is so profound that this is often used to declare CSYNC cables categorically unsafe. I can understand the emotional momentum of that idea, but I disagree with it's logic.  Clearly Nintendo never expected anyone to make SCART cables for NTSC consoles and therefore weren't worried about re-purposing pin 3 on the multi-out.  It was a reasonable assumption on their part as SCART was non-existent in NTSC regions during the SNES's commercial life.   However reputable sellers all include some kind of explanation that you should only use Nintendo NTSC CSYNC cables for NTSC Nintendo consoles and never use those cables on their PAL equivalents.  Furthermore very few people have any reason to mix PAL and NTSC Nintendo consoles in their retro gaming setup and the ones that do can reasonably be expected to be savvy enough to know about this danger and how to avoid it.

#1 Most Important Advice

Do not get swept up in sensationalism or fads when it comes to advice on which cable to buy or use.  Not everyone giving you advice has the same motive for deciding which method is "best".  At very least you should make sure the people you listen to are solving for the same priorities as you are. 

For the record my priority is getting the best image quality with the lowest potential for degradation.

My Advice: Use CSYNC Wherever Possible

Without exception anyone who tells you which sync type to use is offering an opinion, not a fact.  My recommendation is to use CSYNC wherever possible.  Instead of just saying everyone who disagrees with me is wrong, I'm going to attempt to justify my opinion with facts and repeat the strongest arguments I've heard against it so you can decide for yourself.

It cannot be stressed enough that there is no one-size-fits-all solution. Not all consoles output CSYNC, not all consoles output a Luma (Y) signal, not all devices receiving the connection have "sync stripping" circuitry to separate composite or luminance video from the sync signal.  In short you need to know about your equipment in order to make a good decision.

Reasons why CSYNC is the best option:

• It's clean - there's comparatively no chance of it interfering with image quality whether the cables are properly shielded or not.
• It's universally compatible - any scaler/processor/display that accepts sync-on-composite or sync-on-luma will also, by definition, work with CSYNC - the opposite is not true.
• Some devices/displays REQUIRE pure CSYNC in order to function, no scaler/display requires sync-on-composite or sync-on-luma to function.
• Consoles generate CSYNC because they're designed to use CSYNC.  Official Mega Drive cables sold by SEGA used CSYNC.
• All sync sources are eventually broken down into CSYNC anyway - using it directly reduces processing requirements.

Reasons others cite for avoiding CSYNC (rebuttals in red)

• Incorrectly made CSYNC cables can cause damage because of improper (TTL) voltage, while improperly made sync-on-composite or sync-on-luma cables cannot. 
• You are at no risk if you buy your CSYNC cables from a reputable seller.  This argument hinges on on the idea that poorly made or miswired CSYNC cables are the only ones that represent a danger of potential overload.  Really any cable that is mis-wired or poorly made has the potential to damage a console. ALL SCART cables carry a 5V line that far exceeds the tolerance of all of the other signals. If 5V from a sync-on-composite cable comes in contact with one of those other signals whether because it was wired to the wrong place or shorted because it was poorly insulated, or insecurely soldered on, it absolutely can cause damage just as surely if it was a CSYNC cable. If buying cheap, poorly made or otherwise mystery cables you always take this risk no matter what sync type they're wired for.  If you buy CSYNC cables from reputable places there's almost no chance of getting something that is mis-wired, improperly insulated or insecurely soldered. Reputable sellers know how to make these cables so they properly attenuate higher voltage signals to safe levels.
• Incorrectly used cables can cause damage.  "Plug an NTSC CSYNC cable into a PAL SNES and have fun!" 
• If you don't have PAL Nintendo consoles, this is not a risk to you at all.  If you have PAL Nintendo consoles but not NTSC consoles, this is only a risk to you if you buy the wrong cable because you have declined to read - if you're smart enough to stick your credit card number into a computer, you're smart enough to read the warnings that say "NTSC ONLY". If you DO mix PAL and NTSC consoles using RGB through the multi-out and don't trust yourself to read, comprehend or keep track of what you're plugging in where, then for this very specific case sync-on-luma is definitely a better option for your Nintendo consoles.
• CSYNC isn't even part of the SCART specification! 
• This is an urban myth that I admit I fell for and also propagated right up until I saw the actual SCART standard documentation EN 50049 which clearly describes CSYNC (page 5, Table 1, "Contact Number" "20") and gives parameters for its implementation.  Even if CSYNC wasn't part of the SCART specification this isn't much of an argument and somewhat of an irony considering that 95% of all video signals being sent over these cables are "240p" which is definitely not part of either the NTSC or PAL video standards.  That lack of adherence to the standard is the entire reason we have to use specialized scalers to treat video game signals in the first place. If you can't make allowances for de facto standards you're going to have a rough time reconciling with this hobby in general.
• Sync-on-composite doesn't cause interference as long as the cable is properly made (i.e. shielded against parasitic capacitive coupling). Since there are properly shielded cables being made nowadays there's no longer any advantage to using CSYNC. 
• I have strong doubts that any amount of shielding can completely prevent CVBS from coupling with the RGB lines to degrade image fidelity. In every case where I've compared them I was able to see interference with sync-on-composite and not with CSYNC cables of equivalent build quality. However I freely admit the sample size of my personal experience is far too small to be  definitive.  What I will say is that insofar as the shielding is effective it has to be present along the entire signal path to work. This means that if you use one of the many SCART switches that don't shield the signals within the switch, even with the best cable in the world you're still going to wind up with checkerboarding if you use a sync-on-composite cable. Not every device that accepts SCART can handle having CVBS connected to the sync line which means if you're using a sync-on-composite cable, you'll have to complicate your video chain by adding a sync stripper to supply pure CSYNC.

Whether you agree with my rebuttals or not, there is one thing that should be obvious.  No one is disagreeing that CSYNC is the cleanest highest quality method to use.  If you read between the lines when it comes to image quality, what the opponents are all trying to argue is that sync-on-composite or sync-on-luma if fully and properly shielded can be as good as CSYNC in terms of not introducing picture-degrading interference.  While recommending against it they're indirectly acknowledging that CSYNC is the real standard by which they're measuring "good".  We're all dealing with the same facts here, those who disagree with my conclusions just have a different priority than I do. Those advocating for sync-on-composite and sync-on-luma instead of CSYNC are defining "best" as the cables with the lowest risk of causing damage, and I'm defining "best" as the cables with the lowest risk of producing a degraded picture.  

The anti-CSYNC sentiment appears to be driven primarily by people who make cables, mods and devices rather than individual players or retro game enthusiasts.  Those individuals making devices and cables have a much stronger technical understanding to lend weight to their recommendations, but they also have countervailing priorities.  It's not unreasonable to conjecture they are fed up with customers not paying attention and either ordering the wrong thing, or connecting the wrong type of cable to a mod/device and then creating support hassles for them when things don't work or get damaged. From their perspective the best treatment is the one that avoids this hassle altogether.

They're downplaying the risk to image quality because they, no doubt, personally have device chains with good shielding and filtering throughout so they aren't dealing with serious image degradation from sync-on-composite and the constant customer bumbles are a far more real and painful problem for them and the customers themselves.  I'm downplaying their "safety" arguments because I took the time to learn before buying so I've never been confused about the proper configuration or application of CSYNC cables, but image degradation from using sync-on-composite cables is a very real and painful problem for me.

TL;DR

Yes, CSYNC is the cleanest way to transmit sync and least likely to cause interference.

Yes, CSYNC is part of the official SCART standard.

Yes, CSYNC can be dangerous if you don't follow two simple rules:

1. Buy your cables from reputable sellers
2. Don't try to use Nintendo NTSC cables on PAL Nintendo consoles

Get RGB SCART cables wired for CSYNC if:

• The best image output is your first priority
• You're using a SCART switch that didn't cost > $200USD
• Any part of your video chain is using "SVGA" cables to transmit RGBs
• You are going to buy your cables from a reputable seller (not Amazon/AliExpress/Ebay)
• You are NOT mixing PAL and NTSC Super NES's, or Gamecubes in your setup
• You can read and understand simple English and accept responsibility for the consequences if you choose not to 

Get RGB SCART cables wired for sync-on-luma if:

• You're confused by all this "sync" crap and just don't want to mess anything up.
• You ARE mixing PAL and NTSC Super NES's or Gamecubes in your setup and you don't trust yourself not to mix the cables up even though they're clearly labeled

Get RGB SCART cables wired for sync-on-composite if:

• You're confused by all this "sync" crap, just don't want to mess anything up, AND your console doesn't output a Luma signal.
• You are NOT going to be OCD if the noisy composite video finds a way around the shielding and  causes checkerboarding.
• You've got a shiny new RetroTink 5x and want to be able to get both RGB and composite video on the same cable.