How can HDMI avoid jitter?
Posted 22-10-2008 at 10:40 PM by Welwynnick
There's a little known feature of HDMI that was introduced in V1.3a last year. HDMI effectively superceded i-Link as the high definition / multi-channel digital audio interface, but for purely audio purposes it was a great step back. I-Link was effective because it neatly side-stepped the problems associated with getting digital audio from a player to an amp. It did this with a new Audio & Music protocol that kept the digital audio amplitude info transfer from the player to the amp, but now had the timing info transfer the other way - from the amp to the transport.
This was quite revolutionary, and meant that the transport could now become a source of digital ones and zeros, just as many people imagine digital audio does work in its "perfect sound forever" way. Jitter is an unwanted and insidious attribute of the timing info - not the amplitude info. Since this new digital audio replay architecture had the timing info going TO the player, so that the player could be kept in time with the amp, rather than the other way round, there could be no timing info to carry jitter from the player. Does that make sense? We now have an entirely jitter-less transport.
The master audio clock is now in the amp, where it should have been all along (which is why I said "Most audio systems screw up digital audio in one way or another"). The timing info from the master clock is needed at the DACs to regenerate the orignal analogue audio, and if the clock can be co-located with the DAC, and connected with a robust timing interface that doesn't allow data to get anywhere near it, the better chance we have of getting accurate timing to the DAC. Of course, the clock itself still has to be good quality and reasonably jitter-free, but simply not having to make the perilous journey from transport to amp is a huge advantage. There were a number of players and amps released by the major manufacturers a few years ago that had i-Link or (similar) Denon-link interfaces, and these were a great success. Unfortunately, HDMI effectively replaced i-link in the last few years, and that great quality was lost.
The key though, was not the physical interface itself (HDMI TMDS, IEEE 1394 or ethernet or whatever) but the protocol that was run on it. I called it the A&M protocol, which comes from the IEC standard, but each manufactuer had his own name for it, like HAT or PQLS eyc. However, there is no reason in principle why those protocols cannot be run on HDMI as well, and HDMI V1.3 section 7.11 introduces Audio Rate Control (ARC). Different manufactuers still call this HATS or PQLS as before, but the important thing is that this allows the amp to control the transport in the same way that i-Link did. I was the first lay-person to highlight this, in July last year, and now its being introduced by Sony and Pioneer in their latest players and amps. If anyone is interested, there's a good read on this thread, and it covers a few other areas too:
Prospect of better audio with HDMI V1.3a? Pay attention audiophiles - could be good!
Quote:
This was quite revolutionary, and meant that the transport could now become a source of digital ones and zeros, just as many people imagine digital audio does work in its "perfect sound forever" way. Jitter is an unwanted and insidious attribute of the timing info - not the amplitude info. Since this new digital audio replay architecture had the timing info going TO the player, so that the player could be kept in time with the amp, rather than the other way round, there could be no timing info to carry jitter from the player. Does that make sense? We now have an entirely jitter-less transport.
The master audio clock is now in the amp, where it should have been all along (which is why I said "Most audio systems screw up digital audio in one way or another"). The timing info from the master clock is needed at the DACs to regenerate the orignal analogue audio, and if the clock can be co-located with the DAC, and connected with a robust timing interface that doesn't allow data to get anywhere near it, the better chance we have of getting accurate timing to the DAC. Of course, the clock itself still has to be good quality and reasonably jitter-free, but simply not having to make the perilous journey from transport to amp is a huge advantage. There were a number of players and amps released by the major manufacturers a few years ago that had i-Link or (similar) Denon-link interfaces, and these were a great success. Unfortunately, HDMI effectively replaced i-link in the last few years, and that great quality was lost.
The key though, was not the physical interface itself (HDMI TMDS, IEEE 1394 or ethernet or whatever) but the protocol that was run on it. I called it the A&M protocol, which comes from the IEC standard, but each manufactuer had his own name for it, like HAT or PQLS eyc. However, there is no reason in principle why those protocols cannot be run on HDMI as well, and HDMI V1.3 section 7.11 introduces Audio Rate Control (ARC). Different manufactuers still call this HATS or PQLS as before, but the important thing is that this allows the amp to control the transport in the same way that i-Link did. I was the first lay-person to highlight this, in July last year, and now its being introduced by Sony and Pioneer in their latest players and amps. If anyone is interested, there's a good read on this thread, and it covers a few other areas too:
Prospect of better audio with HDMI V1.3a? Pay attention audiophiles - could be good!
Quote:
Quote:
I stumbled across a small new section that I hadn't seen or heard of before. This described, in not very much detail, a new and optional facility for the audio clock in the processor / amplifier to control the clock in the source using a CEC Set Audio Rate command back the source. Here's what the spec says:
Quote:
So what does this mean for us? I'm not entirely sure, as the new capabilty doesn't look closely defined, but to me it reads very much like Sony and Pioneers' descriptions of iLink operation. That's a rate adaptive connection, where the sink (amp) has the master audio clock, rather than the other way round.
It's not a synchronous connection, but I think it's a good thing because the timing information doesn't have to come from the player. That means it's less susceptible to the jitter from (usually) a video clock, a less-than-stellar connection, and an audio clock regenerator in the amp that is slaved to the player. In this configuration, the timing information is generated in the amp, leaving the source to supply only the digital amplitude information. So bits really will be bits for once. Timing info is sent back to the source to maintain flow control.
I don't know where this came from, and I don't know why I never noticed it before. Maybe the audiophile manufacturers, so dismayed by the disappointing jitter performance of HDMI, have been lobbying HDMI for something better. If this is it, then I think we have a great deal to look forwards to. Note that this facility was not incorporated in HDMI V1.3, though, only 1.3A.
Quote:
Quote:
7.11 Audio Rate Control Overview
The Audio Rate Control feature allows a Sink to slightly and continuously adjust the audio clock rate of the Source in order to match the Sink’s crystal-based audio clock. The Sink controls the Source’s audio clock rate with the CEC <Set Audio Rate> command. See CEC Supplement section CEC 13.16 for details. Source ACR behavior is not affected by Audio Rate Control. When Audio Rate Control is enabled the Source shall continue to generate correct ACR packets that accurately reflect the current (possibly adjusted) audio clock rate.
The Audio Rate Control feature allows a Sink to slightly and continuously adjust the audio clock rate of the Source in order to match the Sink’s crystal-based audio clock. The Sink controls the Source’s audio clock rate with the CEC <Set Audio Rate> command. See CEC Supplement section CEC 13.16 for details. Source ACR behavior is not affected by Audio Rate Control. When Audio Rate Control is enabled the Source shall continue to generate correct ACR packets that accurately reflect the current (possibly adjusted) audio clock rate.
It's not a synchronous connection, but I think it's a good thing because the timing information doesn't have to come from the player. That means it's less susceptible to the jitter from (usually) a video clock, a less-than-stellar connection, and an audio clock regenerator in the amp that is slaved to the player. In this configuration, the timing information is generated in the amp, leaving the source to supply only the digital amplitude information. So bits really will be bits for once. Timing info is sent back to the source to maintain flow control.
I don't know where this came from, and I don't know why I never noticed it before. Maybe the audiophile manufacturers, so dismayed by the disappointing jitter performance of HDMI, have been lobbying HDMI for something better. If this is it, then I think we have a great deal to look forwards to. Note that this facility was not incorporated in HDMI V1.3, though, only 1.3A.
Total Comments 0
E. & O.E.