Kudos to Amir

[jkeny]

Believe or not I saw this thread as quite a concession by Ethan. He has been much more moderate of late.
HiI statement that designers should aim for more than borderline acceptable performance imo represents progress. See OP.

Nothing wrong with that - it's the rest of the post that is misleading & incorrectly tries to rope Amir into his stated position on jitter.
If Ethan would only follow his own ironic advice

"Exactly. Misinformation become even more dangerous when people who are successful and highly respected experts in some other aspect of audio make statements about things they don't understand. Then people who don't know how to tell what's real or fiction for themselves only have "reputation" to go by as they pick a side for who to believe. ..............

--Ethan"

 

[Gregadd]

OK

 

[NorthStar]

Hmmm... What I get for writing tired after a long day. I meant it literally; I was not sure what you were looking for, and was not sure if it fit in this thread. I only skimmed the thread and looking back realize how harsh/obnoxious I sounded myself, sorry!

---- Don, I truly appreciate the fact that you did put yourself in my shoes,
and that you just mentioned it above, thank you. :b

* Not everyone will admit their 'small mistakes', but you just did; the sign of a real man.

Amir, while the benefit of a separate clock line is debatable, would you care to discuss what in the protocols themselves might make one or the other better in terms of jitter (or anything else)? Data encoding and such that could lead to lower/higher jitter, BER, etc. I don't know the standards off-hand and a quick summary might be useful (?)

---- This one's for Amir. :b

JKenny just to add as totally agree,
and I guess for Ethan that also means Paul Miller does not what he is talking about reporting various types of jitter and how it is audible (depending upon the behaviour-trait involved of course)
Cheers
Orb

---- Orb, many readers here don't know who Paul Miller is (I do myself no problem),
and just a simple link would help for these people. ...Simple suggestion. :b
=> http://www.milleraudioresearch.com/aboutus/about.html

OK

---- You bet, Greg!

 

[amirm]

Amir, I think you should now clarify the noisefloor issue as it is directly related to audibility. The grass that is seen on the FFT graphs you presented are NOT the noise floor, right? These figures need to be adjusted according to the FFT bin size used on each measurement. Can you adjust these results, lest people incorrectly interpret your graphs as below your -120dB audibility threshold?

The comments I made regarding masking relate to the individual distortion spikes. Those are not impacted by FFT bin size. They can be impacted by the window uses but the program compensates for that.

Edit: Just to be clear

This really needs to be cleared up before talk about masking threshold & audibility? Amir, can you state what the bin size used in each of your FFTs & what the above calc gives for the "actual" noise floor?

The calculation you show is overly simplistic. It assumes no window function and flat spectrum noise. The noise I saw with HDMI did not have flat spectrum so it does not get reduced in pretty increment as shown in that quote.

The focus my article for the most part was to compare HDMI to S/PDIF. In that regard, no compensation is necessary for FFT bin bandwidth. To compute the actual noise floor, the best thing would be to use the instrument's RMS meter but I no longer have the hardware to do that. There is an iterative process that can be performed on the measurements itself to get there but I am too lazy to write the code to do that . My AP has a macro that automates it but per my earlier note, I did not use it for this work. So this would need to wait for a different time.

 

[amirm]

And guys, let's not turn a thread about "kudos to amir" to one where I have to close the thread because we are arguing with each other . I am hoping the article will speak for itself once I publish it.

 

[NorthStar]

-- If I may, on the subject of 'jitter', an undesirable digital audio artifact (measurably and auditory, depending of the amount level); the various type of audio signal transmissions and connections, like how much data a single wire cable is transmitting, and at what speed, an HDMI cable is very complex indeed, and even an S/PDIF cable can transmit stereo hi-res audio (LPCM) or compressed multichannel (up to 6.1-channel) audio, so there too it has some restrictions as to transmitting 'properly' without interaction, or interference from adjacent channels. ...I sincerely believe.
{I read that the length of an HDMI cable affects the timing, speed, and sync between picture and sound, and that it is preferable to go longer than shorter; like 2 meters HDMI cable instead of a 1 meter one.}

Analog is free of those restrictions, and if digital would follow the analog path, we would have only one cable per channel. ...A digital coaxial cable for the Front Left channel, another one for the Front Right, one for the Center, and so on.

Yes, progress is a digital world, but not yet fully developing and performing.
=> With simplicity (one single cable for all) comes complexities, and anomalies.

* Amir, it is impossible to discuss your article without everyone having access to it.
So for now I will eclipse myself and only return when your article will be made available to us all. :b

 

[jkeny]

The comments I made regarding masking relate to the individual distortion spikes. Those are not impacted by FFT bin size. They can be impacted by the window uses but the program compensates for that.

Yes relative comparisons between spikes can be made but no evaluation about audibility (<-120dB) unless we know the actual value of . When you state this in your article "We see noise level at -135 dB, which is well below our target of -120 dB. There are two symmetrical spikes, which likely are jitter induced (I did not have time to isolate the cause as jitter or voltage modulation). Their levels are completely benign at -123 db." - we really cant say if the spikes you mention are ACTUALLY below -120dB (benign) if the noise floor of -135dB is in question.

The calculation you show is overly simplistic. It assumes no window function and flat spectrum noise. The noise I saw with HDMI did not have flat spectrum so it does not get reduced in pretty increment as shown in that quote.

Nonetheless, you don't deny that the accurate noise floor is not what you state in the article & therefore nothing can be said about audibility?

The focus my article for the most part was to compare HDMI to S/PDIF. In that regard, no compensation is necessary for FFT bin bandwidth. To compute the actual noise floor, the best thing would be to use the instrument's RMS meter but I no longer have the hardware to do that. There is an iterative process that can be performed on the measurements itself to get there but I am too lazy to write the code to do that . My AP has a macro that automates it but per my earlier note, I did not use it for this work. So this would need to wait for a different time.

Whether you can now make accurate noise floor measurements is probably immaterial - the statements in the article relating to audibility need to be clarified, I believe.

 

[amirm]

Yes relative comparisons between spikes can be made but no evaluation about audibility (<-120dB) unless we know the actual value of . When you state this in your article "We see noise level at -135 dB, which is well below our target of -120 dB. There are two symmetrical spikes, which likely are jitter induced (I did not have time to isolate the cause as jitter or voltage modulation). Their levels are completely benign at -123 db." - we really cant say if the spikes you mention are ACTUALLY below -120dB (benign) if the noise floor of -135dB is in question.

The statement that -123 db distortion is benign is independent of the noise level.

Nonetheless, you don't deny that the accurate noise floor is not what you state in the article & therefore nothing can be said about audibility?

Disagree. See above.

Whether you can now make accurate noise floor measurements is probably immaterial - the statements in the article relating to audibility need to be clarified, I believe.

As I said, at some point I will compute the noise floor for the purpose of computing dynamic range. Dynamic range was not the focus of the article even though there are references here and there to noise floor level.

 

[opus111]

As I said, at some point I will compute the noise floor for the purpose of computing dynamic range. Dynamic range was not the focus of the article even though there are references here and there to noise floor level.

Where in the article you state that -135dB noise floor level (taken from the FFT) exceeds your -120dB requirement that does look to me like an error - its not an apples to apples comparison.

 

[amirm]

Where in the article you state that -135dB noise floor level (taken from the FFT) exceeds your -120dB requirement that does look to me like an error - its not an apples to apples comparison.

Well, what you quoted from me was the answer to the point you are making so I won't repeat it again....

 

[opus111]

Well, what you quoted from me was the answer to the point you are making so I won't repeat it again....

I ddin't see a 'oh, sorry I screwed up' from you Amir - did I miss it?

 

[amirm]

I ddin't see a 'oh, sorry I screwed up' from you Amir - did I miss it?

I said earlier:

Yes, those are FFT noise floor. I had revised the article to get that statement in there but not sure if it made it into the article.

Alas, I just checked and the version that went to print didn't include the note regarding FFT impact on noise floor. I

As I said, it is on my Todo list to update the article for the online version. It simply is not easy task as I mentioned due to non-wideband noise. Since the point of the article does not depend on this, I will probably publish it as is and circle back to address the dynamic range.

Such kudos thread this turned into.....

 

[NorthStar]

-----

 

[Daveski]

-- If I may, on the subject of 'jitter', an undesirable digital audio artifact (measurably and auditory, depending of the amount level); <brevity snip> Analog is free of those restrictions, ...

Certainly, LP and tape isn’t free of timing errors. Also, a highly modulated groove can have influence on or bleed over to adjacent groves and the density of the modulation can change the drag and friction and have influence on the velocity and timing. This is not totally unlike when Amir notes how digital jitter is influenced by the data itself (simultaneous switching effects) or when what the video circuits are doing can affect the timing of the audio circuits.

Below is a link to a video from phono-cartridge designer Peter Ledermann that was previously posted here. In it, he describes how an LP and phono-cartridge is very much plagued by similar effects and argues that it really isn’t so much different than digital in many regards. In fact, he discusses a stylus on a cantilever being dragged through a rotating spiral of vinyl in terms of jitter and sample rate.

http://vimeo.com/62377436# 9:34 is the timestamp where he begins discussing LP reproduction, and analog in general, in digital terms of jitter, sample rate, and bit-depth. And why such concepts and terms apply to analog too.
an

 

[NorthStar]

-- Cool, and you're right. ...Actually, any and all spinning music/video mediums are prone to timing issues, sync issues, jittery issues, etc., etc., etc. ...All mechanical imperfections. ...Tapes, albums, CDs, tra-la-la ...

The day music and movies will perform without spinning, that'll be the day! ...MP3 players, PCs?

 

[Ethan Winer]

Nowhere in there is a phrase "not necessarily audible".

You are wrong, and my original Post #1 is still here above as proof:

Son of a gun, Amir came to the same conclusion in this article, and he even explained why it's not necessarily audible (masking of nearby frequencies).

--Ethan

 

[Ethan Winer]

is your position now that jitter MAY be audible as Amir stated?

My position has always been that any artifact can be audible if it's loud enough.

under what circumstances do you now claim it might be audible?

If it's loud enough to not be masked by the music itself. This is pretty basic stuff, no?

--Ethan

 

[Ethan Winer]

I saw this thread as quite a concession by Ethan. He has been much more moderate of late.

Not at all. My contention has always been that jitter is too soft to be an audible problem. When Amir and I first debated this a while back I wasn't aware of how much worse HDMI jitter is than S/PDIF. So I learned something valuable from that exchange, and I even included that in my Audio Expert book. Now Amir is agreeing that as bad as HDMI jitter looks on a graph, the amounts he measured are not likely to be audible. So it's a win-win for everyone here, except those who constantly aim to create friction.

--Ethan

 

[NorthStar]

We all know that various type of digital distortions are masked by other electrical/mechanical distortions, like the loudspeakers for example.

 

[Daveski]

We all know that various type of digital distortions are masked by other electrical/mechanical distortions, like the loudspeakers for example.

Again, it's not just a digital thing...