HDMI Cable Speed & Features Explained
Audioholics would like to thank Kurt Denke from Blue Jeans Cable for writing the following article:
Considering that an HDMI cable is, at bottom, nothing more than an arrangement of wire, foil and plastic, the purchase of HDMI cables have become awfully confusing. Everywhere one turns, there are all sorts of specification version numbers being tossed around, claims about HDMI cable "speed," and representations about support for 3D, 2K by 4K video, 1080p, Deep Color, and a host of other features. What does it all mean?
Lies, Damned Lies, and Video Cable
Let's start by observing that you can't believe everything you hear. In fact, if you've spent any amount of time around home audio/video and home theater, you probably already know that the sale of audio and video cable is a field in which false and fanciful claims and exaggerations are the rule rather than the exception, and that if a sensational statement is contained on the retail packaging of a video cable, it's much more likely to be false than true.
Accordingly, before we start to talk about the things that really matter, let's get rid of some of the rubbish that can simply be dismissed as false or simply irrelevant, but which finds its way onto packages and into advertising copy:
There's no such thing. We will get to the question of cable speeds later, which does relate in a way to this subject.
- Support for new audio formats,
such as Dolby TrueHD
While support for audio formats is a wonderful thing, cables have nothing to do with it. All HDMI cables support Dolby TrueHD, et cetera, and since these audio formats have no impact upon the bitrate, no cable supports them any better than any other.
- "Speed Rated" HDMI
Apart from two official speed ratings, "standard" (Category 1) and "high" (Category 2), as defined by HDMI Licensing, there are no other official speed rating standards for HDMI. Some resellers of Chinese HDMI cables at crazy-high prices (yes, you know the one I'm thinking of) mark their cables with bogus "speed ratings" for which there are no published standards or specifications. If it says "Ultra High Speed," or something like that, step slowly (or at Ultra-High Speed, if you prefer) away.
- Support for x.v.YCC colorspace
Like support for Dolby TrueHD, a good thing; but, just as with Dolby, supported equally by all HDMI cables regardless of type, spec version or anything else.
- Support for other specific
resolutions, features and protocols
With the exception of the new Ethernet and audio return channel feature, whether it's 2K by 4K video, Deep Color, or what-have-you, support for these features depends entirely on the cable's rated speed and the impact of the particular feature on the bitrate, not on the nature of the feature.
- "120Hz" or
No set-top device emits an HDMI signal at these framerates, though the Sony PS3 and any PC has the potential. Rather, a display labeled "120Hz" or "240Hz" has an internal frame-refresh rate as stated. It's completely irrelevant to the HDMI cable or to the signal the HDMI cable carries.
With all of that on the rubbish heap and burning, there's very little left in the marketing-speak of HDMI cable, and that's a good thing. As complex as HDMI standards can be, and as complex as transmission-line theory can be, buying HDMI cable actually ought to be fairly simple, in part because...
Digital Is Digital
As long as one appreciates the limits of the point, it's an important point to make: a digital signal is just a string of ones and zeros. When a digital signal gets through a cable, and is interpreted correctly at the other end with no dropped bits, the result is no loss of information, and hence no loss of picture or sound quality. The signal may have suffered a great deal of degradation along the way from multiple causes; there may have been EMI, RFI, intrapair skew, interpair skew, return loss, rounding from capacitance, attenuation, anything - but if the bitstream gets read correctly at the end of the process, none of that degradation makes one bit (either figuratively or literally) of difference.
Now, that point often gets made into something it is not. People will sometimes claim that cable quality does not matter. The truth is more like this: if a particular cable, regardless of price and internal quality, delivers the signal in condition to be accurately read, no increase in cable quality will make things any better. However, if the cable does NOT deliver the signal in good condition, it is entirely possible that a better cable (which may or may not be more expensive) may fix the problem. This is so because, while it may seem a simple matter to deliver a series of ones and zeros by switching a voltage up and down, things get pretty funky at ultra high frequencies, and electricity does some strange and not always obvious things when one tries to run high-speed signals.
Weirdness in the Wire
As human beings, we have a hard time seeing or, in some cases, even understanding, things that happen really fast. That's helpful where home theater is concerned, of course, because we can watch a stream of still photos rush by and perceive the illusion of motion. It does mean, however, that some things behave rather counter-intuitively when they're running at high speed. When we switch on a light bulb, we experience the light going on simultaneously with the flip of the switch, because neither the time it takes the electricity to reach the bulb nor the time it takes the filament to heat up are long enough for us to really notice. As a matter of fact, however, when a voltage is applied to a conductor, it doesn't just instantaneously deliver itself to the other end of the conductor. It takes time--not a great deal of time, of course. When an HDMI signal is running at 3.4 Gbps, the maximum single-link speed of HDMI, (you may also see this described as "10.2 Gbps" because sometimes people sum the three data channels in stating the bitrate) the voltage is switching on and off, at a rate as high as 3.4 billion times per second, with each of those voltage changes propagating its way down the cable at nearly the speed of light, one after another. This rate is so high that at any moment, if you could freeze time and sample the voltages in the cable somehow, you'd find that the bits are literally lined up behind one another, each just a few inches long, in the cable.
So, what happens to one of these little inches-long voltage pulses? It starts out at the source with nice, relatively sharp corners, with the voltage rising or falling rapidly and then leveling off. But as it propagates through the wire, it starts to weaken and smear out a bit, for a variety of reasons. The sharp transition becomes softer; the leveling off of the voltage takes longer; the total voltage drops; and, in what has to be the most counterintuitive part of it all, bits of the pulse react to changes in the characteristics of the cable and, instead of continuing along in their original direction, bounce backward down the cable. This last phenomenon is known as return loss, and it is the worst enemy of any digital signaling process.
So, what arrives at the display is not what left the source. It is a degraded version of the original signal, with rounded transitions, noise, and a weaker amplitude. As the speed of the data increases, the degradation gets worse, so that a cable that will deliver a standard-definition 480i signal with relatively little degradation may deliver a 1080p signal in much worse condition. The challenge, now, as these bits get delivered to the display, is whether the display can reconstitute this degraded signal. If the receiver circuitry can interpret what it takes in accurately, there is no loss of picture or sound information and all is well. Because cable characteristics, including particularly the stability of the cable's impedance, affect how badly the raw signal is affected, it is quite possible that if we have two cables to try, one of them will not work correctly while the other does. It is also quite possible - likely, even - that if we were to slow the data rate down (e.g., by switching to a lower resolution, frame rate or color depth), the non-functioning cable would suddenly work just fine.
At very low frequencies and bitrates, most of these effects are minimal. Not only are effects like return loss liable to be smaller at lower frequencies, but the slowness of the data speed means there's more room for slop. At slow data speeds, it's not so hard to make a cable that performs acceptably. But higher frequencies mean more of all of the things that degrade the signal -- there is more attenuation, more return loss, more rounding from cable capacitance, as frequencies increase. The principal way to combat these problems is simply through control over manufacturing tolerances--the more consistent the wire sizes, dielectric sizes, wire spacing, shield wrapping, and so on, the less variation there will be in impedance and the less trouble there will be with skew. Two cables which appear to be of equal quality when evaluated at low data speeds may turn out to behave very differently at high data speeds.
The upshot: yes, it is "all ones and zeros," but sending and receiving a stream of ones and zeros can be a good deal less simple than it seems when they are running at the rate of 3.4 billion per second, and cable quality can be a determining factor in whether our gear works as intended.
"Speed Rating" a Cable - Separating Science from Nonsense
As we've pointed out, there are some nonsense "speed rating" systems for HDMI cable out there, which exist largely just to adorn the packages containing HDMI cable with up-selling tools for the vendor. Whether it's "Ultra High Speed," or just "Faster 'n' All Get-Out," you can safely ignore these labels--they are completely meaningless. However, there is one important, but limited, sense in which one can meaningfully and accurately talk about "speed ratings" for HDMI cables.
First-generation HDMI cables were designed with 1080i and 720p video in mind, at eight-bit color depth. Both of these resolutions require a clock rate of 74.25 MHz, and 742.5 Mbps per data channel in the HDMI signal, and originally (through HDMI specification 1.2), this is what HDMI cable compliance testing was targeted at. With HDMI specification 1.3, however, the single-link bandwidth limit per data channel was raised to 3.4 Gbps, to accommodate such things as deep color and higher framerates, and from what we've already said above it should be clear that a cable which works fine at 742.5 Mbps will not necessarily work at a data rate which is over four times as fast. To address this issue, HDMI specification 1.3 introduced two "Categories" of HDMI cable, somewhat blandly named "Category 1" and "Category 2." Ever since 1.3, all HDMI cables which are tested for compliance certification are designated either as Category 1, and tested at 742.5 Mbps/channel, or as Category 2, and tested both at 1.65 Gbps (without equalization) and at 3.4 Gbps/channel (with equalization). A cable which has passed Category 2 certification is capable of handling any data rate allowed under the HDMI specification; a cable which has passed Category 1, but not Category 2, is certified capable of handling anything up to 742.5 Mbps/channel, representing conventional 720p or 1080i HD resolutions at their normal framerates and eight-bit color depth.
The "Category 1" and "Category 2" labels for these data-speed tests, not being descriptive, seem to have been a bit confusing for consumers, and accordingly, the HDMI Licensing organization has announced that they should be referred to as "Standard Speed" and "High Speed" instead. Additionally, in response to the deceptive use of bogus vendor "speed ratings," HDMI Licensing has expressly prohibited the use of variants such as "Ultra High Speed" and the like, so that with any luck the only "speeds" being talked about soon will be Standard and High.
What About "Gigabits per Second" Ratings?
For various reasons, many vendors avoid talking about the official "High" and "Standard" speed categories, and instead like to claim that a cable is rated for some particular speed which may be higher or lower than the tested speeds under Category 2. We recommend that you disregard these claims. There are no objective engineering standards against which to test them, and none of the vendors who make these sorts of claims publish the criteria by which they have allegedly rated their own cables.
One other problem with relying on these claims is that it is not always clear what is meant by a stated data rate. The use of terminology here has been incredibly sloppy. The HDMI organization will sometimes refer to the official speeds by giving the single-channel data rate (e.g., 3.4 Gbps for the higher- speed of the two Category 2 tests), and sometimes by giving the three-channel data rate (e.g. 10.2 Gbps for the same). To know what is meant when a data rate is cited, you've got to know whether it's the one-channel or the three-channel rate. To make matters worse, the HDMI organization also sometimes calls the clock speed (one-tenth the bitrate), stated in Megahertz, the "bandwidth" of the cable. This is simply an incorrect use of the term, and it means that the 3.4 Gbps speed not only may sometimes be called 10.2 Gbps, but also may be called 340 MHz. In actual fact, a cable which could only handle 340 MHz of bandwidth, as that term is ordinarily defined, could not handle a 3.4 Gbps datastream - the cable bandwidth required is somewhere in the neighborhood of 1.5 times the data rate, expressed in Hertz, or 5.1 GHz - this accommodates the maximum fundamental frequency (half the data rate) plus its third harmonic.
Ethernet and Audio Return Channel
One other characteristic feature which an HDMI cable may carry is the new "Ethernet channel," and its accompanying sub-feature, the "Audio Return Channel." To fit these features into the HDMI cable, there is a slightly revised cable structure available under 1.4 which reconfigures a couple of the miscellaneous conductors into a 100-ohm balanced data line for use in Ethernet, with one side of that line also being used to allow a display to send multichannel audio back "upstream" to an A/V receiver. It's a kludgy arrangement, and makes for a complex specification. We have our doubts that these will ever be features to see much use, but as with all future things, it's hard to say.
What About Version Numbers?
You'll notice we haven't said much about version numbers, and there's a reason for that. Version numbers have tended to confuse the issue more often than not, and HDMI Licensing has asked us to stop using them to describe cables (Editor's Note: They have also forced manufacturers to stop using them to describe electronics as well - a move we don't necessarily agree with). Why? Because a version number, by itself, tells you nothing useful about the cable. All versions of the spec permit cables to be certified compliant at the "Standard" speed data rate, and all versions of the spec permit cables to be certified compliant without an Ethernet channel. A cable tested and found compliant using spec version 1.4 does not necessarily offer any advantage over a cable tested and found compliant using spec version 1.1, and to know whether it does you need only to know two things: (1) is it certified as a "high speed" or a "standard speed" cable, and (2) does it carry an Ethernet channel? Meanwhile, HDMI Licensing has made the requirement explicit that before a cable is certified as compliant at a specific length, it must be tested and found compliant either at that length or longer, so there should be no squeezing around the issue by marketing a cable found compliant at 3 feet as though it had been found compliant at 50.
How to Tell the Compliant Status of an HDMI Cable
The advice we have been giving for a long time holds true still: if you actually want to know whether a cable is compliant, and at what standard, you need to see the vendor's compliance certificate. When a cable is tested and found compliant, a certificate which shows the length of the cable, and the nature of the testing (Category 1 or 2, with or without Ethernet) is issued. If your vendor doesn't have one for their cable, that may well be because the cable is, despite representations to the contrary, non-compliant.
HDMI Licensing has issued a new set of logos which are, in future, to be used to label HDMI cable assemblies. There are five of these, only four of which are applicable to conventional HDMI cables (the last is for automotive use), and which cover the four possible answers to the two questions: standard speed or high? Ethernet or not?
Now, the clever fellows who think up ways to sell cables are no doubt working out just how they're going to make new specious claims about HDMI cables without running afoul of the trademark licensing guidelines, and it will be interesting indeed to see what they come up with. When you run into new, strange and interesting ways of rating HDMI cables, be alert: more likely than not, the purpose is not to teach you something about HDMI cable but to pry more money out of your wallet.
Many thanks to Kurt Denke of Blue Jeans Cable for contributing this article to our site.
Confused about what AV Gear to buy or how to set it up? Join our Exclusive Audioholics E-Book Membership Program!
Recent Forum Posts:
bandphan, post: 699514
Here is a company that is trying to rate and certify them.
I know those guys–nice fellows, but I think their methodology is wrong. Among other things, they're not testing cable at high speeds with equalization as the spec would require, and that distorts the results. In particular, a cable with higher capacitive losses and lower return loss will EQ out better than a cable with higher retun loss and lower capacitive loss, and if you fail to EQ, you can't tell the difference. In official HDMI Compliance Testing, the highest bitrate (3.4 Gbps/channel, 10.2 total) is always EQ'd, and a cable like ours actually performs much, much better at that extreme high bandwidth test than it does at the highest non-EQ'd compliance test tier (1.65 Gbps/channel, 4.95 total).
HDMI Cable at Blue Jeans Cable
As long as the spec remains what it currently is, it's going to be very hard for anybody to greatly exceed the lengths which are now doable on conventional copper cable. The problem basically is that there's only so much permissible attenuation. The spec allows you to either pass a set of attenuation benchmarks or pass an eye-pattern test; a pass of either, even with the other failing, will pass the overall test and be compliant. At 23.5 AWG, the Series-1 fails the attenuation benchmark tests well before it fails the eye-pattern, but the eye-pattern itself, indirectly, imposes limits on attenuation, too, because if the amplitude of the signal drops too low, it'll start to impinge on the eye mask.
We have been back and forth with Belden on some development ideas to try to further extend the reach. The problem is that, with the bonded-pair design already having the return loss well under control, the only way to improve distance performance on the eye-pattern is to either reduce attenuation (which would require bigger wire, and correspondingly bigger dielectrics–and the cable is big enough already!) or to reduce capacitance. Capacitance is tightly tied to impedance, which is required to be 100 ohms on these data lines, and the only way to improve capacitance while holding impedance steady is to go to a different dielectric material–but, without going into too much detail there, it's fair to say that when you change dielectrics from good old solid PE to something with a lower dielectric constant, you start to face a new set of problems with dimension control and manufacturing consistency.
So, while I wouldn't be surprised if we're able to put out a 30-foot Category 2 cable sometime (we're about to start ATC testing on our new Ethernet designs), I'll be very surprised if, barring a complete change in the spec, we're ever able to put out a 40-foot Category 2 cable. Likewise, we might stretch the Series-F2, our 28 AWG cable, from 15/25 in Cat 1/2 to 20/30. We'll see.
The existing design unfortunately bears the stamp of HDMI's lowly origins. It was made backward-compatible with DVI, which was designed to be a computer-monitor interface, and which therefore was not usually needed to run over distance. Computer engineers seem to have a thing for twisted-pair data cables, and I think that the spec was written without a strong appreciation for just what kinds of problems a balanced-line configuration would create. The cable's got to have very high manufacturing consistency to avoid skew (intrapair and interpair) and return loss, and if HDMI had been run unbalanced (i.e., on coax) things would be easier.
Blue Jeans Cable
Do you thinks it's even possible to go longer distances with HDMI cables?
My real wish is for someday fiber optics to get down to a reasonable cost. But considering the cost regular ethernet fiber media converters, I doubt it will ever happen. The dual Cat5e sollution is nice, but still seems like it would end up hitting the bandwidth wall in the future.
I guess you can tell I'm not a fan of the current HDMI cables and specs. It was a good idea when first conceived, but went to poop shortly there after.