HDMI and Audio Jitter

[dlaloum]

In days of yore (circa 2008) - there were massive discussions about Audio Jitter and HDMI

Paul Miller measured Jitter on a bunch of AVR's and this list tells an interesting story:

Denon AVR-3803A
---------------
SPDIF: 560psec
HDMI: 3700psec

Onkyo TX-NR906
---------------
SPDIF: 470psec
HDMI: 3860psec

Pioneer SC-LX81
---------------
SPDIF: 37psec
HDMI: 50psec

Yamaha RX-V3900
---------------
SPDIF: 183psec
HDMI: 7660psec

This was the state of play in 2008/9

Where are we at today?

First thing I notice, is that there are few if any mentions of Jitter measured in ps...

There is an article which states that the J-Test method of determining Jitter is non viable for HDMI

https://www.bramjacobse.nl/wordpress/wp-content/uploads/2021/04/TN-118_Measuring_Jitter_with_J-Test.pdf

If this is the case - how were the Paul Miller measurements achieved?

And - of more practical import - where are we at today?
Of note: Pioneer's SC-LX81 even then was showing what is possible... all the others had 10x or more Jitter on HDMI compared to SPDIF

Key jitter symptom (often not measured separately): raised noise floor.... aka SINAD

 

[amirm]

This is an issue and is resolved in some cases using sample rate conversion (built into DAC chips like ESS). I wrote an article with a ton of measurements once about this issue with HDMI: https://www.audiosciencereview.com/...s/a-deep-dive-into-hdmi-audio-performance.56/

 

[DonH56]

See @amirm 's article. The basic answer is that today virtually all DACs do clock recovery asynchronously, with buffered data to effectively decouple the DAC itself from the data stream, making this a non-issue.

 

[mansr]

In days of yore (circa 2008) - there were massive discussions about Audio Jitter and HDMI

Paul Miller measured Jitter on a bunch of AVR's and this list tells an interesting story:

Denon AVR-3803A
---------------
SPDIF: 560psec
HDMI: 3700psec

Onkyo TX-NR906
---------------
SPDIF: 470psec
HDMI: 3860psec

Pioneer SC-LX81
---------------
SPDIF: 37psec
HDMI: 50psec

Yamaha RX-V3900
---------------
SPDIF: 183psec
HDMI: 7660psec

This was the state of play in 2008/9

Where are we at today?

First thing I notice, is that there are few if any mentions of Jitter measured in ps...

There is an article which states that the J-Test method of determining Jitter is non viable for HDMI

https://www.bramjacobse.nl/wordpress/wp-content/uploads/2021/04/TN-118_Measuring_Jitter_with_J-Test.pdf

If this is the case - how were the Paul Miller measurements achieved?

And - of more practical import - where are we at today?
Of note: Pioneer's SC-LX81 even then was showing what is possible... all the others had 10x or more Jitter on HDMI compared to SPDIF

Key jitter symptom (often not measured separately): raised noise floor.... aka SINAD

Even the worst figure in that list is still below what's audible, at least with anything resembling normal audio. There is nothing to worry about.

 

[dlaloum]

Even the worst figure in that list is still below what's audible, at least with anything resembling normal audio. There is nothing to worry about.

Perhaps - but I always found the SPDIF input on my AVR to sound better than the HDMI.... and the only thing I can attribute that to is Jitter, as it seems to be the primary difference between the two audio streams. (for identical audio streams, obviously).

And a measure of Jitter in a quick reference figure - as provided by the Miller labs measures in ps - would be useful... although some discussion of how meaningful it is is warranted.

 

[Kal Rubinson]

Here's relevant comment from a partisan: https://audiopraise.com/documents/HDMI_Audio_Jitter_v1.pdf

 

[dc655321]

Perhaps - but I always found the SPDIF input on my AVR to sound better than the HDMI.... and the only thing I can attribute that to is Jitter

That was level-matched and without peeking, right?

 

[dlaloum]

That was level-matched and without peeking, right?

No - just long term listening, and experimenting... but I may have been influenced by knowing the specs to be:

SPDIF: 470psec
HDMI: 3860psec

Still - like anything - I would really like to have the Jitter specs identified with every AVR review...

 

[dlaloum]

Here's relevant comment from a partisan: https://audiopraise.com/documents/HDMI_Audio_Jitter_v1.pdf

Thanks for that Kal - I had actually just read that one a couple of hours back!

Trouble is - with AVR (or prepro) based setups, I am not sure where a device like the VanityPro fits in (if at all)... does it decode all the current AV Audio streams? - Atmos, Auro3d etc?

Or is it an Audio only tool.... in which case, much of the benefit can be gained by the simple expedient of bypassing HDMI altogether and resorting to SPDIF (for the 99%+ of the Audio streams that are supported)

 

[Kal Rubinson]

Trouble is - with AVR (or prepro) based setups, I am not sure where a device like the VanityPro fits in (if at all)... does it decode all the current AV Audio streams? - Atmos, Auro3d etc?

Nope. No decoding at all.

Or is it an Audio only tool.... in which case, much of the benefit can be gained by the simple expedient of bypassing HDMI altogether and resorting to SPDIF (for the 99%+ of the Audio streams that are supported)

None of the streams that I really care about can be handled by S/PDIF; they're lossless hi-rez multichannel.

 

[DonH56]

Jitter articles on ASR: https://www.audiosciencereview.com/forum/index.php?threads/digital-audio-jitter-fundamentals.1922/ and follow-on https://www.audiosciencereview.com/...igital-audio-jitter-fundamentals-part-2.1926/

These days all DACs I have heard about will reclock/resample and ignore any jitter on the HDMI wires, at least up to the point that it is bad enough to lose bits. I do not know what sort of error correction HDMI uses. Ten or fifteen years ago this was an issue, and "asynchronous" DACs were just starting to appear. Now they are ubiquitous.

 

[dc655321]

No - just long term listening, and experimenting... but I may have been influenced by knowing the specs to be:

SPDIF: 470psec
HDMI: 3860psec

Still - like anything - I would really like to have the Jitter specs identified with every AVR review...

I searched for thread titles containing AVR and posted by @amirm. Nearly all of the first 6 I looked at showed jitter spectra. Are you looking to have that distilled to time figure?

Do you understand how inconsequential 4 billionths of a second is here?

 

[dlaloum]

Do you understand how inconsequential 4 billionths of a second is here?

Yes a spectrum provides one form of valuable information .... but in the same way that a SINAD figure provides an easily handleable data point, (seperate and different from a noise spectrum chart) it would be useful to have Jitter as a single comparable data point.

Leaving aside the audibility of jitter wars....

 

[dlaloum]

Jitter articles on ASR: https://www.audiosciencereview.com/forum/index.php?threads/digital-audio-jitter-fundamentals.1922/ and follow-on https://www.audiosciencereview.com/...igital-audio-jitter-fundamentals-part-2.1926/

These days all DACs I have heard about will reclock/resample and ignore any jitter on the HDMI wires, at least up to the point that it is bad enough to lose bits. I do not know what sort of error correction HDMI uses. Ten or fifteen years ago this was an issue, and "asynchronous" DACs were just starting to appear. Now they are ubiquitous.

That's all good and well in theory - but the actual measurement spectra are showing all the signs of Jitter...

So, clearly, this is NOT a resolved issue.... (at least, not for AVR's & PrePro's) it may be a different issue however, the cause of that Jitter could be different - but it clearly is there, and should be measurable and quantifiable.

 

[amirm]

Yes a spectrum provides one form of valuable information .... but in the same way that a SINAD figure provides an easily handleable data point, (seperate and different from a noise spectrum chart) it would be useful to have Jitter as a single comparable data point.

Leaving aside the audibility of jitter wars....

I could never figure out how Paul Miller was assessing the single digit value for Jitter. It could have been a jitter meter of sorts in his analysis software that computed it. What it was based on, I couldn't see based on values for each jitter component. Maybe it was RMS, or peak or something else.

 

[dlaloum]

I could never figure out how Paul Miller was assessing the single digit value for Jitter. It could have been a jitter meter of sorts in his analysis software that computed it. What it was based on, I couldn't see based on values for each jitter component. Maybe it was RMS, or peak or something else.

I might have to go back through my J-test reading - I think the ps figure was based on an RMS reading of some sort - one of the articles I read over the last day or so mentioned it... but I skimmed past!

 

[DonH56]

That's all good and well in theory - but the actual measurement spectra are showing all the signs of Jitter...

So, clearly, this is NOT a resolved issue.... (at least, not for AVR's & PrePro's) it may be a different issue however, the cause of that Jitter could be different - but it clearly is there, and should be measurable and quantifiable.

Yes, Amir's tests show plenty of jitter in AVR/AVPs, but I don't know it is HDMI per se. I mean that literally -- I do not know either way. I am not sure @amirm is even testing them with HDMI inputs? Have not looked, other things going on today...

 

[dlaloum]

Yes, Amir's tests show plenty of jitter in AVR/AVPs, but I don't know it is HDMI per se. I mean that literally -- I do not know either way. I am not sure @amirm is even testing them with HDMI inputs? Have not looked, other things going on today...

Yes Amir has been testing with both HDMI and SPDIF inputs - although there seems to have been some variation (the Denon X3700 didn't get a HDMI test, but instead had its pre tests via coax/SPIDF?... would have liked to have seen the HDMI results!)

Various tests have also shown that merely having HDMI connected, even while using the SPDIF input, can generate noise.... so things can clearly get complicated!

 

[DonH56]

Yes Amir has been testing with both HDMI and SPDIF inputs - although there seems to have been some variation (the Denon X3700 didn't get a HDMI test, but instead had its pre tests via coax/SPIDF?... would have liked to have seen the HDMI results!)

Various tests have also shown that merely having HDMI connected, even while using the SPDIF input, can generate noise.... so things can clearly get complicated!

I can well believe that, especially for an AVR or AVP. For that matter, just connecting USB to some components increases jitter, mainly because of lack of ground isolation and poor circuit design and layout (IMO, having had to deal with isolating ADC/DAC digital and analog portions in various designs through the years). Then add video processing and there is potentially a lot of noise in the box that could corrupt the DAC in numerous ways. I do not disagree with HDMI problems at all, just not sure that is the largest and certainly not the only problem.

I have not kept up with the HDMI spec. Last time I dug into it was ~1.4, and I only glanced at 2.0. Key specs seemed to be lacking or poorly defined, and interoperability left to the whim of the manufacturers. More or less. I am not an HDMI expert, not even terribly knowledgeable, though a few Ive know who are up on the spec tended to agree.

 

[DonR]

I can well believe that, especially for an AVR or AVP. For that matter, just connecting USB to some components increases jitter, mainly because of lack of ground isolation and poor circuit design and layout (IMO, having had to deal with isolating ADC/DAC digital and analog portions in various designs through the years). Then add video processing and there is potentially a lot of noise in the box that could corrupt the DAC in numerous ways. I do not disagree with HDMI problems at all, just not sure that is the largest and certainly not the only problem.

I have not kept up with the HDMI spec. Last time I dug into it was ~1.4, and I only glanced at 2.0. Key specs seemed to be lacking or poorly defined, and interoperability left to the whim of the manufacturers. More or less. I am not an HDMI expert, not even terribly knowledgeable, though a few Ive know who are up on the spec tended to agree.

As of 2.1a implementing features is now optional. You can call your interface HDMI 2.1 but implement only a subset of 2.1 features... so much so that HDMI 2.0 can be now called HDMI 2.1 ... it's getting as bad as USB.

https://www.theverge.com/2021/12/29...ess-cables-source-based-tone-mapping-ces-2022