|
 |
UPDATE AND WARNING: The article below was written in 2017, and is badly out of date. We've kept it here as a kind of "archive" piece, but it should not be relied upon for current information.
HDMI 2.1 and the HDMI Cable: an Early Look
The HDMI 2.1 spec, which was expected to be issued in the early 2nd quarter of 2017, has just now (November 28, 2017) been published. Blue Jeans Cable is an HDMI Adopter, and so we now finally have our hands on a copy of the specification and are reviewing it in detail. Our initial review has answered a few questions and raised some interesting flags, however, and so we thought we'd write a bit to let you know what's up vis-a-vis HDMI 2.1 and cables.
When Will 2.1 Be Implemented? When Will Products Be Available?
There are no products yet which implement any of the new features or use any of the new bandwidth, resolution, etc, of 2.1. The specification was published November 28, 2017; that's a sort of a "go" signal for developers of source, display and other devices, but it will take quite a while before products implementing 2.1 are on the market. In fact, at the moment even if someone were to come up with a new device implementing 2.1, there'd be no way to certify it, because HDMI has published only the master specification document. In order for products to be evaluated for conformity to 2.1, there will need to be Compliance Testing Specifications, and these are not likely to be out until deep into 2018. The upshot is that it's fairly likely that you won't see products until 2019, so don't fret about not having 2.1 devices in your system today, or in the coming year or so. After that, we'll see how long it takes for 2.1-specific features, protocols and resolutions to be implemented.
Connectors:
Under HDMI 2.1, cables will continue to use the same connector styles which are currently in use; the pinout and the relationships between conductors have remained the same, so that any cable made for the new spec will be backward-compatible with HDMI 2.0 and all prior versions of the specification. The HDMI Dual-Link connector, type B, still remains in "orphan" status, with no HDMI Dual-Link mode specified, and we suspect that there never will be such a thing as HDMI Dual-Link.
Pinouts and Basic Cable Structure:
The roles of the wires in the HDMI cable do not change under 2.1, with one exception: the pair which carried the clock signal in all prior versions is converted to a full-bandwidth data pair, carrying video and audio signals in parallel with the other three video-and-audio data pairs. This, by itself, would require no cable redesign, as the clock pair is normally made identical with the other pairs.
Bitrate:
We've maintained for a long while that there's just too much demand placed upon the four main pairs of the HDMI cable -- that twisted pairs aren't really capable of handling the data rates that are being pushed through them, at least not without some bandwidth-reducing device such as multilevel encoding (used on such things as Ethernet connections). But HDMI 2.1 sticks with binary encoding, so it literally is all "ones and zeros," and now the data rate in each of the four data pairs runs up to 12 Gbps, for a total throughput of 48 Gbps (compared to a maximum of 18 Gbps in the HDMI 2.0 spec). This, of course, means that cables will need to meet new testing in order to keep up.
New "Category 3" Cable:
For some time now, there have been two bandwidth-specific testing tiers for HDMI cable -- "Category 1," or "Standard," and "Category 2" or "High Speed." Additionally, there is the optional "Premium HDMI Cable" certification testing, in which we've had most of our Series-FE cables certified. The 2.1 spec introduces a "Category 3" testing tier, which will be more stringent than Category 2. The details as to exactly how much more stringent this testing will be are a bit tricky for us to work out on first read, as there are several elements to the testing, including some S-parameters the exact nature of which we haven't quite worked out.
Forward Compatibility:
The first question everyone will want to know is, of course, whether any of the cables available today will meet this new Category 3 testing, and to that we have to admit we simply don't know, apart from the general observation that cables carrying active circuitry are unlikely to do so as the chipsets will need updating. There is an interesting clue of sorts in the spec, though, and that is a sort of advisory drawing/spec for a Category 3 cable. Note the word "advisory" -- the features shown aren't mandatory, and the spec remains a "performance" spec and not a "construction" spec: it doesn't matter how you go about meeting the test criteria, it only matters whether you do meet them or not.
Will our Series-FE cable do the trick? Again, we don't know, but we do have some historical precedent on our side. Those who have been our customers for a long time now will recall that when we first introduced bonded-pair HDMI cable, there was only one "speed" rating, for a total throughput of about 2.25 Gbps. But bonded pairs are designed for high-frequency performance, and when "high speed" HDMI at 10.2 Gbps came along, these bonded-pair cables still met spec even though they hadn't been designed with these higher bitrates in mind. Then, when 2.0 came along, at 18 Gbps, all of our high-speed certified Series-FE cables were tested, and the same design STILL met the new standards and qualified for "Premium HDMI Cable" certifications. 2.1 represents a doubling of the data rate on each pair (from 6 Gbps/pair to 12 Gbps/pair), and we do not have the test data, at this time, to predict whether our existing bonded pair designs will satisfy the 2.1 standard or whether something new will be required.
We will begin assessing, immediately, what to do; we have to determine what the likelihood of any of our existing products meeting the new Category 3 testing will be, and then work out what, if any, changes we will need to make. But understand: this is all at the conceptual stage right now. There isn't even a testing specification in print, so until that comes out, nobody can definitively say anything about what will and won't pass Category 3.
A Note on Twisted Pairs, the "Model" Category 3 Cable, and Crazy Bitrates:
Those who've been following our writing on HDMI for a while will recall that, long ago, we argued that because of the inherently poorer impedance stability of twisted pair cable and the high-frequency demands of HDMI signaling, the writers of the original spec ought to have built it around a coaxial cable geometry rather than a bundle of data pairs (which was done to maintain backward compatibility with the DVI specification). This, which we felt from a wire-and-cable perspective was rather obvious, led to a surprising number of people letting us know that, in their view, we had no idea what we were talking about. It was therefore a bit amusing to see that the HDMI 2.1 spec's example of a compliant Category 3 cable design replaces the four data pairs with eight coaxes -- still running differential signals, but in twinned coaxes rather than in conventional twisted pairs.
While we suppose we ought to feel somewhat vindicated there, it does point out some unfortunate features of the HDMI and HDMI 2.1 worlds. HDMI ought to have been run, originally, in coaxes without differential signaling; coax, if properly constructed for the application, has the bandwidth to run these signals -- even the signals called for in 2.1. We are, in fact, now selling a full line of 12G SDI coaxial cables for professional video camera feeds and video production. But a proper coax-based HDMI design would have had the video signals simply running unbalanced, not in a differential signaling configuration. Doing it this way opens up a range of difficulties in cable construction, including:
(1) Even the most minimal differences in characteristics between two coaxes being run as a differential pair will cause huge problems with skew. Intrapair skew is limited to 30 picoseconds; this equates to a physical cable length disparity, in a foamed PE dielectric, of about a third of an inch, and unless dielectric quality is exquisitely well controlled, it's easy to wind up with differing electrical lengths even when the physical length is identical.
(2) The limits on permissible connector overmold size mean that large cables are very difficult to use; this means we are looking at micro-coaxes, with high attenuation, imposing inherent limits on length. Worse, as dimensions shrink, dimensional control, which is critical to maintaining impedance, becomes more difficult. Meanwhile, the resistance of the cable to physical distortion through flexing diminishes. The coaxes ought to be reasonably protected from one another by inner jacketing; but there won't be room in a bundle of this size for anything more than a mylar film, which will provide insulation between pairs but will not help physically stabilize them.
(3) Control over crosstalk will require good shield performance, because it will not be possible to twist the coaxes into pairs with each other; crosstalk will be presented to the "pairs" very asymmetrically. But effective shielding relies upon being able to shunt noise to ground at low resistance, and foil shields simply aren't very good for this. To make them more effective, traditional cable design would call for a braided shield over the foil, as we see in SDI coax. But braids take room, and the cable drawing here calls for a spiral serve shield instead. While a spiral serve shield will take less room and be simpler to apply, it also is more likely to contribute periodicity to the cable, and to open up asymmetrically during handling. There are good reasons why even cheap CATV coax doesn't rely upon spiral serve shields, despite their being cheaper and quicker to apply.
These are just a few of the issues. Getting 12 Gbps performance out of micro-coaxes, or out of conventional twisted pairs, is going to be troublesome; it's easy to make a sketch of the cable, and much harder to make the actual cable stock. Whatever you may hear casually said about it being "just ones and zeros," there is nothing easy about shoving twelve billion of those ones and zeros down a length of cable every second.
"What about this 2.1 Cable I'm seeing ads for?"
Just in the last few weeks, we've been bombarded with questions about cables which are already claiming 48 Gbps bandwidth and HDMI 2.1 compatibility. Any such claims are dishonest; those who make them should be viewed as highly suspicious. Until yesterday (11/28/17) there wasn't even a master specification setting out what the criteria for meeting the 48 Gbps requirement were or what other characteristics a cable must have in order to meet Category 3 testing. There still isn't a Compliance Testing Specification, and won't be one for months at least.
On this, a note: one of the odd features of HDMI cable marketing the last few years has been that vendors will advertise a cable as having some particular bandwidth, often one which is many Gbps above the highest tier in existing specifications. One thing everyone should understand is that such claims are not so much false as they are entirely meaningless. Until the specification for some bitrate is released, one CANNOT know what a cable would have to do to be regarded as offering X amount of bandwidth. That's because "bandwidth" isn't some sort of an attribute with a fixed meaning, but depends upon what's in the specification, e.g., how many dB of attenuation are permitted at particular frequencies, and under what test conditions. One could write a specification which places relatively little burden on the quality of cables, but a greater burden on the quality of sending and receiving circuits; or one could write a specification which places relatively little burden on sending and receiving circuits, and heavy burdens upon cable. Until there's a spec, nobody knows what the criteria will be, so claiming a bandwidth beyond anything provided for in existing specs is fluff -- it bears, and can bear, no relation to any real standard.
The upshot: until there are certifications being issued for compliant cables, anybody who tells you that his cable supports 36 Gbps, or 48 Gbps, or 17,000 Gbps (we haven't seen that claim yet, but hey, it's only a matter of time...), should be assumed to be either dishonest or ignorant. We won't make those claims; when we have new products with certifications, or new certifications for existing products, we will tell you exactly what our certifications are and what they mean in accordance with the official HDMI specifications. HDMI 2.1 products are surely coming to market, but they're not coming particularly soon.