What's With These "Speed Rated" HDMI Cables?
One of the most interesting, and effective, marketing campaigns to "up-sell" consumers to expensive HDMI cables in recent times has been the effort to market different grades of HDMI cable quality as having been "speed-rated" for different HDMI resolutions. We can tell that this sales pitch has been pretty effective, because we get asked all the time whether our cables are "speed-rated."
The idea of speed-rating cables does make some sense, as we'll explain, and there are various ways of characterizing the speed capabilities of a cable. The truth of "speed-rating" for HDMI cable is a bit more complicated than marketing slogan-writers would have people believe, however. It simply isn't the case that high resolutions or special HDMI features will always (or often) require a particularly expensive HDMI cable.
Data Cable, HDMI, and Bitrate
HDMI is very unlike other methods of routing video signals between consumer devices. Most methods--component, composite, s-video, and so on--are analogue connections, while HDMI is digital. An HDMI signal rides primarily upon four twisted-pair lines, one of which carries clock pulses and the other three of which carry the colour components and sync information of the signal; as with any digital transmission, these signals are carried not as continuously varying voltages representing a range of values, but as positive and negative voltage pulses representing binary bits -- ones and zeros -- which, when decoded in accord with the HDMI signalling protocol, reveal the values of the red, green and blue components of the video signal.
For this reason, HDMI cable really has more in common with familiar data cables, such as CAT5, than with other video cables such as component video cable. And how HDMI cable behaves depends very heavily on exactly what we demand of it in terms of bitrate. The bitrate (or, to use the more colloquial terminology above, "speed") of an HDMI signal is determined by the resolution, the frame rate, and the colour depth of the signal. At the low end, a standard 480p signal, with 480 x 720 resolution, 60 frames per second, and 8-bit colour will run about 270 Mbps (Megabits per second) through each of the signal pairs of the HDMI cable. A 1080i signal (1080x1920, 30 full frames/second) or a 720p signal (720x1280, 60 full frames/second), using 8-bit colour, will run 742.5 Mbps through each of those pairs, almost triple the 480p rate. A 1080p signal, with 60 full frames per second of 1080x1920 video, at 8-bit colour, doubles that to 1.485 Gbps. But the HDMI interface now also allows for the possibility (implemented, as of this writing, on very few devices) of "deep colour," colour depths which require 12 or 16 bits instead of the standard 8. Not surprisingly, going from 8 bit to 12 bit colour increases the bitrate by 50%, while going from 8 bit to 16 bit doubles it, so that a 16-bit 1080p signal runs the signal pairs at 2.97 Gbps. The HDMI spec document calls for the interface to be able to support data rates at least up to 3.4 Gbps per signal pair; this doesn't correspond to any resolution currently in common use.
As we increase the "speed" of the signal, the degree to which the signal degrades in the cable worsens, for various reasons. Yes, it's "all ones and zeros," but the amplitude of the signal diminishes, the rise and fall times of the voltage transitions lengthen, and the quality of the signal can degrade substantially, to the point where ones and zeros do not look very different from one another. For any given cable design, if we establish a set of reasonable speed-based performance criteria, there is a threshold length of cable below which, for any given speed, the cable will pass, and above which it will fail. Ideally, these criteria should be tagged to benchmarks related to device performance, so that failure on the test is indicative of potential trouble in real-world hookups, while passing indicates that the cable should cause no problems. The threshold passing length, for any given cable design, will be longer at lower speeds and shorter at higher speeds. So the notion, fundamentally, of "speed-rating" cable is sound, and it can bear a real relationship to how cable actually behaves in the real world. That does not mean, however, that "speed-rating" as it is marketed is an accurate representation of that relationship; far from it. The language of the engineer and the language of the salesman are two very different tongues.
What Does The Rating Mean?
The problem, very simply, with any private system of "speed ratings" is that it's very hard to know what those ratings mean. There are any number of criteria one might choose to determine whether a cable meets a particular speed test; some of these criteria will match very well with real-world situations, but some will not. If the vendor will not publish the specific engineering criteria underlying the ratings, the ratings are worthless because (1) nobody knows what they mean, and (2) nobody can independently verify whether the product conforms to the standards on which the rating is based.
Contrast this with, for example, standards for professional digital video coax for SDI (Serial Digital Interface) applications. The SMPTE (Society of Motion Picture and Television Engineers) has established published standards for SDI cable which any engineer with the proper test gear can test cables against. So, when Belden or some other cable vendor says that a certain cable will carry a certain resolution of SDI a certain distance in conformity with the SMPTE standard, (1) everyone knows exactly what is meant by that claim, in terms of measurable attributes (e.g., "attenuation at a frequency 1.5 times the bitrate"), and (2) Belden's claim to conform to the spec can and will be tested by customers and competitors, and if the product doesn't live up to the promise, there'll be trouble. Standards like that are useful, meaningful, and above all, testable.
And, of course, as we've pointed out, the capability of an HDMI cable to carry a particular data speed will depend very much on the cable's length. The result of any proper speed rating system will not be to say, "model A rated for bitrate X, model B rated for bitrate Y, model C rated for bitrate Z." Instead, it will be something like, "model A rated for 480p up to 75 feet, 720p/1080i up to 40 feet, 1080p up to 30 feet, 1080p/12 bit up to 20 feet," and so on, with each model being rated for higher speeds in short lengths than it is rated in long lengths. This is, in fact, exactly what one finds if one consults Belden on the suitability of cables for SDI; 7731A (RG-11 precision coax) will run any given resolution farther than 1855A (mini-RG-59 type precision coax), but 1855A will run the highest resolutions just fine, at shorter distances. If we needed to run very high-resolution HDMI for a distance of only one inch, a bundle of nineteen rusty coathangers would do the job (but would be very difficult to solder to the connectors!)despite the poor quality of the cable and the very high bitrate; if we need to run it three feet, a decently-made but unremarkable HDMI cable will work just fine; it's primarily when we need to run longer distances that the differences in cable quality start to become important.
A Speedy Sales Pitch
The one set of "speed-ratings" we know of, promoted by a prominent rebrander/reseller of Chinese-made HDMI cables found in almost every big-box retail store, fails these most basic standards.
First, the vendor does not say what each "speed" tier in the ratings means. A bitrate is given, and that's it. But what does that mean? What standard does the cable have to meet at that bitrate to earn the rating? Without that information, it's impossible to know what the meaning of the rating is, and it's impossible to take a sample of this vendor's product and test it against the standard.
Second, and quite telling: the vendor does not tie these speed ratings to length. Rather, in each case a particular model is tied to a particular speed rating, regardless of length. On the vendor's site, the highest-rated cable model is advertised with the same rating at lengths ranging from one meter to 75 feet. This is a logical impossibility, in a world where the losses in any cable are tightly tied to its length. The answer to how this can be is buried in a footnote, which reads:
"Gbps results are based on testing of 2-meter cable lengths under unequalized, worst case conditions. Actual Gbps rates may vary."1
So the rating has no relation to length; if it has any meaning, it has a meaning only at a length of 2 meters (we cannot, of course, know what that meaning is, because the testing criteria are not published), and this tells us nothing about cable performance at other lengths. All we know is that the cable will perform better at shorter, and worse at longer, distances (which, of course, is what we know about ANY cable, by the nature of cable).
The representations tied to the rating get worse than this, though. In addition to the "speed rating" claims, this vendor implies that certain resolutions, refresh rates and features are unavailable on the "lower-speed" cables, by use of little tags which label cables as being for "60 Hz" or "120 Hz" displays (or with no tag, which presumably is intended to suggest no support for 60 Hz refresh rates); similar tags which reference either 8-bit colour or 12-bit colour; and tags referencing x.v.colour and surround sound, which are attached only to the most expensive lines of cable. Let's look at these issues.
First, one thing needs to be clear: ALL HDMI cables, regardless of spec version or origin, have the same pinout and basic physical structure. What this means is that all HDMI cables are, at least in terms of being correctly wired and having all necessary parts, compatible with all HDMI features and protocols. There is no basic difference between a cable for 60 Hz or 30 Hz; no difference between a cable for RGB, Y/Pb/Pr, or x.v.colour colourspace; no difference between a cable for 8-bit colour and a cable for 12-bit or 16-bit colour.
The type of signal being carried from device to device is relevant to the cable only in one sense: bitrate. Anything which affects bitrate bears on the cable's suitability for the application, because cables built to tighter tolerances will perform better2 at higher bitrates and longer lengths than cables which are poorly built; anything which does not affect bitrate does not affect the choice of cable. So, let's go through these items:
- 60 Hz versus 120 Hz refresh rates: This is absolutely not a cable issue. The relevant refresh rate is the frame rate of the signal going through the cable, not the display's internal refresh rate. No home theater devices now on the market support 120 frame-per-second video through the HDMI interface; the highest frame rate is 60, for 1080p/60 and 720p/60 video sources. To say that one cable is suitable for a display with a 60 Hz refresh rate while another is suitable for a 120 Hz refresh rate is utter nonsense, because in both cases the signal running through the cable will be running at the same bitrate.
- x.v.colour (xvYCC), RGB, Y/Pb/Pr: Here again, this is absolutely not a cable issue. The choice of colourspace has nothing to do with bitrate. If you have two devices and a cable running RGB-colourspace HDMI signals, and you switch to xvYCC-colourspace under HDMI 1.3, the cable will handle that colourspace exactly as well as it handled the RGB. To say that one cable is suitable for one colourspace while another is suited for a different colourspace is hogwash. The cable does not know what the bits it carries mean, and the choice of colourspace does not affect bitrate.
- 8-bit, 12-bit, 16-bit colour: This is a true bitrate issue. The "speed" of the data, given a fixed resolution and frame rate, is proportional to the colour depth in bits, so 16-bit colour runs twice as fast as 8-bit colour. As we have noted, for any given cable design, the higher the bitrate, the shorter the run of cable that will pass signal quality testing. But at two meters--the ONLY length for which this vendor asserts that the bitrates have been tested--almost any cable, including most very cheap cables, will pass all HDMI-supported bitrates just fine. And, of course, one cannot even determine the bitrate from the colour depth alone; it is also necessary to know the resolution and frame rate. A cable which will handle 1080p at 8-bit colour will handle 1080i at 16-bit colour; we know this because the bitrate in these two cases is exactly the same. The cable does not know what the colour depth is; the only thing which has a bearing on the cable's performance attributes is the total bitrate, regardless of which of the many possible combinations of colour depth, resolution and frame rate that bitrate represents.
In sum, this notion of "speed-rating" cables, as it is being sold, belongs in the realm of sales pitch, not engineering.
Self-Deprecation is the Sincerest Form of Puffery
The ironic thing about the "speed-rating" sales pitch is that it is a sort of self-deprecation-marketing in which the misrepresentations about the vendor's own products are mostly negative. All of this vendor's cables, if they are not very badly produced, are capable (at the tested length of 2 meters) of supporting all of the protocols, features, and bitrates which the marketing materials would suggest require the very most expensive cable. By implicitly speaking ill of its own lower-priced product and suggesting that the product won't do things it's perfectly capable of, the vendor manages to "up-sell" the customer to a costlier product. In other words, it's all about niche marketing and price-points; it is all about figuring that someone buying an $800 television might be willing to pay $75 for an HDMI cable, while someone buying a $3000 television might be willing to pay two or three times as much; it is all about the hard-sell approach of "gee, that's a nice TV; you wouldn't want to hook it up with just any old cable now, would you?" in which a salesman tries to eye-roll and shame the customer into buying costly accessories.
The HDMI Compliance Testing Spec and Authorized Testing Centers: Independent "Speed Rating" Tests
So, is there really such a thing as speed-rated cable, in a sense that's true, meaningful, and performance-based? Yes, indeed there is. Licensees of the HDMI trademarks (known as "HDMI Adopters"; Blue Jeans Cable is one) submit cable products for independent laboratory compliance testing. The HDMI 1.3 spec introduced two speed "categories" for cable testing, Category 1 and Category 2. Category 1 testing is done with the signal pairs running 742.5 Mbps; Category 2 testing is done with the pairs running 1.65 Gbps, and then again, with equalization, running 3.4 Gbps. Cables are given passing scores if they meet certain technical requirements at these speeds; and these technical requirements correspond to the spec document's minimum required performance of source and display devices, so that a "compliant" cable connected to two "compliant" devices should, in principle at least, always work perfectly.
Now, if you've been looking at HDMI cable packages and admiring the speed ratings, you may be saying, "but those Category 2 bitrates are really low. The best cable on the store shelf says it's good to 10.2 Gbps, three times the top Category 2 testing rate!" This is a common source of confusion, which unfortunately originated with some confusing use of terms by the HDMI spec authors themselves; the bitrates aren't three times as high, though it does sound that way. Some people refer to the bitrate by summing the three colour channels, while others refer to it as the rate flowing in each of those channels. So, 10.2 Gbps by one system of reference is the same as 3.4 Gbps by the other system; and similarly, 4.95 Gbps on that cable package in the store corresponds to testing at 1.65 Gbps. The better practice, in our opinion, is to use the single-channel bitrate because it has some relationship to the engineering challenges that have to be met in designing cables and interfaces; but the three-channel combined bitrate sounds more impressive, and so some like to use it.
Not surprisingly, when actual, official, independent HDMI compliance testing is done, the result is not a certification for a model of cable, but for a model of cable up to the tested length. The Compliance Testing certificate, issued by the testing center in accordance with a well-defined test spec, is the best assurance that a cable really can carry the data speeds you need it to carry. We display ours; but the vendor responsible for all this "speed-rating" nonsense not only doesn't publish its test certificates, but isn't even an HDMI Adopter and therefore isn't eligible to submit cables for Authorized Testing Center review. Are there ATC certificates for its cable? Who knows? Have they ever bothered to see how their cable performs out beyond 2 meters, which is the only length they claim to have tested and rated? They may or may not have, but you'll have a heck of a time getting a test report or a compliance certificate out of them.
__________________________________________________________1. Not long after this article was posted, the footnote was removed and replaced with one which makes no reference to the tested length at all, and then another footnote, for the "12-bit" products, stating that the cable is 8-bit only at 35, 50 and 75 foot lengths. Has the tested length changed? What are the criteria used to determine these "levels" of compliance? We have been unable to find any clue.
2. When we say "perform better," we mean this in the "signal-quality" sense: a better cable will produce less degradation in the signal itself than an inferior cable. It should be noted, however, that with digital signals, the fact that the signal is less degraded and hence "better" in that sense does not mean that the picture or sound will necessarily be any better. Rather, the difference in degradation will make no difference in many cases, and will only show up on the display in the event that the degradation crosses the threshold to the point where there are some unrecoverable data in the signal, resulting in sparkles, line dropouts, loss of sync or loss of picture. The practical difference between excellent cable and poor cable can easily be meaningless in one case and dramatic in another.