How does jitter affect HDMI?
Posted 22-10-2008 at 10:39 PM by welwynnick
There's a fundamental difference in the way that spdif and HDMI handle digital audio, and unfortunately it seems to cause more of a problem with HDMI. Most audio systems screw up digital audio in one way or another, but HDMI adds a new way to do it - plus a possible way out. Digital audio carries not one, but two, information streams - amplitude information and timing information. The amplitude info is the data in the ones and zeros, and the timing info is the synchronisation between the sampling (ADC) and the reconstruction (DAC) of the analogue output stream.
SPDIF carries both streams on the same channel - the timing info is embedded into the data stream, which are recovered and separated by the receiver for sending to the DACs. Sources and sinks have gotten better over the years, but as you might expect, this is a compromise, and the data tends to corrupt the timing.
HDMI does it quite differently, and carries data (amplitude info) and clock (timing info) on physically separate channels. Although the audio data is embedded into the video data on HDMI, this is still a good thing in principle, as it should help to maintain the integrity of the clock. Unfortunately (and this is why audiophile manufacturers criticise audio on HDMI) the HDMI clock is a video clock, not an audio clock. The receiver has to reconstruct an audio clock out of the video clock by down-sampling it internally under continuously varying decimation commands sent over the CEC by the transport. This DOES work (well... function) but nothing like as well as a dedicated audio clock.
Hence the problem with HDMI audio, and it is a real problem, nomatter what some manufacturers may say. Its not a simple matter though. HDMI seems to perform much worse when there is no video being carried. Playing a CD over an HDMI connection can be very poor, and spdif can sound significantly better. The difference can only be attributed to jitter on HDMI - either because of the link, or because of the receiver processing. LPCM with video on HDMI seems to be a different matter though, and can sound very good. I have a good understanding of the causes and effects of jitter, but things start to get difficult here. I think that audio jitter may be related to the video clock that is used for each signal transmission format. HDMI has a very wide range of transmission formats, which have to cover everything from NTSC to 1200p video. There is a correspondingly very wide range of video clock frequencies to cover these, and some of them don't seem to work well with audio. CD audio seems to be a particular problem, but HDMI does have a trick up its sleeve to get round this very effectively, and I'll come back to that later.
SPDIF carries both streams on the same channel - the timing info is embedded into the data stream, which are recovered and separated by the receiver for sending to the DACs. Sources and sinks have gotten better over the years, but as you might expect, this is a compromise, and the data tends to corrupt the timing.
HDMI does it quite differently, and carries data (amplitude info) and clock (timing info) on physically separate channels. Although the audio data is embedded into the video data on HDMI, this is still a good thing in principle, as it should help to maintain the integrity of the clock. Unfortunately (and this is why audiophile manufacturers criticise audio on HDMI) the HDMI clock is a video clock, not an audio clock. The receiver has to reconstruct an audio clock out of the video clock by down-sampling it internally under continuously varying decimation commands sent over the CEC by the transport. This DOES work (well... function) but nothing like as well as a dedicated audio clock.
Hence the problem with HDMI audio, and it is a real problem, nomatter what some manufacturers may say. Its not a simple matter though. HDMI seems to perform much worse when there is no video being carried. Playing a CD over an HDMI connection can be very poor, and spdif can sound significantly better. The difference can only be attributed to jitter on HDMI - either because of the link, or because of the receiver processing. LPCM with video on HDMI seems to be a different matter though, and can sound very good. I have a good understanding of the causes and effects of jitter, but things start to get difficult here. I think that audio jitter may be related to the video clock that is used for each signal transmission format. HDMI has a very wide range of transmission formats, which have to cover everything from NTSC to 1200p video. There is a correspondingly very wide range of video clock frequencies to cover these, and some of them don't seem to work well with audio. CD audio seems to be a particular problem, but HDMI does have a trick up its sleeve to get round this very effectively, and I'll come back to that later.
Total Comments 0
