Understanding Jitter in Digital Audio: Measurements and Listening Tests

[PeteL]

I can have all the bits written on a piece of paper or statically stored on a disc. There is no jitter in data. The jitter comes from clock perturbations upon playback as it is a real time dynamic effect. So as jitter basically is a sample reconstructing too early or too late such time has no meaning for the stored sample data. It can only be early or late as it is played back and clocked out.

Now even a free oscillating crystal isn't purely jitter free. Plus various other noises can get in to alter exact timing upon playback depending upon design. Yet these can only occur during actual playback and are a fluctuation in the clock and not anything in the data itself.

Oh, I forgot that Pkane's Distort also does jitter. You should give it a try.

OK but I don’t see where I was suggesting that stored data may be be affected by jitter. What does it have to do with my original question?, I was obviously not talking about bit flip on the stored data, I was asking if there was cases where the transfer is bit perfect but reconstructed with a heavily jittery clock could lead to a distorted analog waveform, or if we should only worry about transmission jitter where clocking error could cause actual bit flips. Distort will just enlight me about audibility of jitter, but that’s not what I was asking.

 

[wyup]

With S/PDIF, things get a little more involved. The receiver has to synthesize a high-rate clock, typically 256x fs or more. This involves a PLL, and even if the incoming data stream is completely free of jitter, some will be created here. How much depends on the PLL design, which also determines how well jitter from the input is suppressed.

How do you explain the superior spidif jitter removal of ESS and CirrusLogic dac implementations compared to AKM in regard to upsampling? I guess they all use similar high-rate clocks and PLL design.

If I feed a good-enough stable spdif signal, would the upsampling chip introduce more jitter than without it?

 

[thats_it]

That latest AMIR video on jitter is OUTSTANDING. Years ago, actually bought the big thick Pohlmann and Watkinson books on digital audio. And I've got all my EE university textbooks and many other PAPER books resting on the traditional WOOD shelf ... and none of them were really helpful in understanding jitter as it relates to dig. audio.
John Atkinson, in the Measurements section of his reviews, DID go into the J-Test and Dunn papers ...and this from about a year ago:
https://www.stereophile.com/content/2020-jitter-measurements

 

[mansr]

How do you explain the superior spidif jitter removal of ESS and CirrusLogic dac implementations compared to AKM in regard to upsampling? I guess they all use similar high-rate clocks and PLL design.

If I feed a good-enough stable spdif signal, would the upsampling chip introduce more jitter than without it?

Which Cirrus and AKM chips are you referring to? AFAIK, none of their DAC chips incorporate any de-jittering features. ESS DACs use a high-rate ASRC with a DPLL that seems to work quite well.

 

[wyup]

Which Cirrus and AKM chips are you referring to? AFAIK, none of their DAC chips incorporate any de-jittering features.

Here you can see Amirm's jitter test on his Topping D30Pro review, which uses cirrus logic 4x cs43198 dac chip. AKM doesn't seem to have de-jittering features on spdif.

When I measure jitter on DACs these days, usually USB is great but Toslink and Coax, not so much. D30Pro goes against the trend, producing superb and identical results with all three inputs:

ESS DACs use a high-rate ASRC with a DPLL that seems to work quite well.

Do you know what is the sampling frequency at the output of this process? If I understand well it upsamples to the high-rate clock of the oscillator (256x), does it downsample to a lower frequency before D/A conversion?
And last, chips that do not do ASRC like AKM, what do they do to lock to the spdif signal, do they use PLL the same way at source rate?

 

[Wes]

Where are you reading the requirement for a degree for IEEE? As far as I know, they accept equivalent work experience in lieu of a degree.

As to AES membership, you have to have three sponsors to get accepted as a member. It is not as simple as just paying the money.

I personally don't like IEEE because they take free papers from authors and then charge an arm and a leg for access to them. They think everyone is a major corporation and can pay their fees.

slight OT diversion - this is a BIG DEAL to me - i think if the research is funded with tax dollars (NIH, ONR, NSF, etc.) then it should be available free to the public (with secrets being an exception)

 

[scott wurcer]

slight OT diversion - this is a BIG DEAL to me - i think if the research is funded with tax dollars (NIH, ONR, NSF, etc.) then it should be available free to the public (with secrets being an exception)

Totally agree, I worked with some folks at NIH years ago to extend some of their image processing algorithms. There was never a question that all of this IP was free to the public. If you delve into it things like Octave and Python have an enormous amount of very sophisticated code available for free.

 

[B4ICU]

Upsampling a CD (normal) is done before conversion to digital, as SPDIF can carry by STD up to 48k Sample top, of two channels. On CD it's 44.1.
For some DAT or Mini Disc, it was less.

Upsampling was designed and practiced at the early days of DAC (CDs) when the LPF at 20kHz was impacting phase and noise artifact into sound.
They were very steep, to get the sampling Fr. down by 80dB plus, to avoid it's presence at output.
The Upscaling (by X2, X4, X8 etc'.) was to move the clock rate upwords, to ease the filtering knee and reduce the artifacts of those issues.
Later, 1 bit DACs were introduced by Philips and most of the oversampling became history. Some designs, didn't use a filter at all and allowed the analog limited bandwidth circuitry to deal with that super sonic Fr.
Jitter and clock: as long as there is a clock and data, the sampling timing is critical. If some data words, are sampled with a slight delay or early slightly, if the audio wave would be a pure sine, it would show distortion. (Mr. Amir's graph shows it. If that would be over the slop, and not at the peak, it would show much better. At peak dV vs. dt, is minor (Cos Q is 0). Two next samples at peak might be identical or very similar, while on the slop, its maximized (most at zero crossing, whe Cos Q is 1).
On this thread, for some 50p Sec jitter is too good, for others it's not enough. Sought it out please. I don't want to be part of that debate.

 

[AndreaT]

Another superb introduction to the basic of jitter. As I hope I understood it, the necessary high rate frequency of sampling (44.1 KHz) to allow a PCM recording of analog signals up to 20 KHz plus 2.1 KHz space for the filter to prevent digital aliasing, requires very precise pulses every 1/44.1 K times per sec. Even a very small fluctuating error in the rhythm of sampling adds up in the D to A conversion and has the potential to create those - 70 dB side bands to the 12 KHz signal that might interfere with audibility.

I wonder if the poor sound quality of the first CD players in the 1980’s was due to high jitter maybe as high as in the microsec range! As I painfully invested in the CD edition of West Side Story from a DG recording (Karajan, Carrera, Ti Kanawa) in 1985, I remember my huge disappointment in comparing it to the LP version played on an Empire 208/Shure V15 IV turntable. The CD player was one of the first marketed under Marantz brand. Perfect sound forever certainly it was not!

Amirm, have you considered testing a CD player from the mid 1980’s just to comment on the progress of precision timing and other possibly obvious to the ear improvements of 44.1/16 digital sources of the past 35 years?

 

[amirm]

Amirm, have you considered testing a CD player from the mid 1980’s just to comment on the progress of precision timing and other possibly obvious to the ear improvements of 44.1/16 digital sources of the past 35 years?

I tried to buy the original Sony CD player but the ebay seller wanted too much for it. I no longer have my own from that era. Oldest one I have is circa 1999.

 

[B4ICU]

Early CD players: some sounded good: https://www.hifinews.com/content/philips-cd100-vintage
However, at the early 80's the HW available was somewhat limited: DAC's of 16 bit resolution at conversion rates of 44.1K sps and implementation of very sharp filters, to roll down 80 dB plus over 1 octave (20kHz, vs 44.1kHz) had an impact on sound. Not to say, that in the Analog world a 16 bit DAC (or ADC) is as good as a 14 bit only, due to conversion accuracy error and zero bias. Most of the above were in a single unit (not transport and DAC, so clocking the data was good. I do not recall talking about Jitter till a bit late on.

Most of that debate around CD problems rise due to the stupid assessment that digital is perfect, so it should sound perfect...Many of the poor sounding, came from bad recordings, especially converting stuff from LPs and tapes to CD's. This justified some audiophile firms like Mobile fidelity, Reference Recordings (RR) and more, to do something about it. Even TELARC who declared of digital recordings with no compression came out with some poor sounding CD's. Even Chesky, who is good, missed a few. Not all sound as good as The Raven by Rebeka Pidgen (Spanish Harlem): https://www.discogs.com/es/Rebecca-Pidgeon-The-Raven/release/4500328 later, that album was also become avaliable on SACD.

With that said, some had recognized that CD could be improved. HCDC was first to try. Test showed it didn't deliver. SACD is tackling resolution both on sampling rate (time scale) and Amplitude (bit count). Many times playing a Good CD, after hearing a SACD got me the impression that the CD sounded as goog. In some other occasions, as TBM Midnight Sugar:

The SACD exceeds the quality of the 24K gold CD version by far.
Hi RES was also introduced, as well as 24 bit and 192k smps. Didn't worked for me.
At the same time, analog vinyl(33.33 RPM!) came back big. I really do not get the obsession to have all the clicks and pops with a highly distorted and BF and dynamics limited. It's mostly like play a CD with having the frier cooking some fries and the dishwasher working.
Most audio, for a while is to go on MP3. The truth is that most don't care for high quality sound. We are a minority with extinction danger. Look on how many audio magazines remained at airport book shops, vs the 90s and 2000. Practically none.

 

[LTig]

Early CD players: some sounded good: https://www.hifinews.com/content/philips-cd100-vintage
However, at the early 80's the HW available was somewhat limited: DAC's of 16 bit resolution at conversion rates of 44.1K sps and implementation of very sharp filters, to roll down 80 dB plus over 1 octave (20kHz, vs 44.1kHz) had an impact on sound. Not to say, that in the Analog world a 16 bit DAC (or ADC) is as good as a 14 bit only, due to conversion accuracy error and zero bias. Most of the above were in a single unit (not transport and DAC, so clocking the data was good. I do not recall talking about Jitter till a bit late on.

The first Philips CD players used a 14 bit DAC with 4x oversampling to circumvent the steep reconstruction filter.

Most of that debate around CD problems rise due to the stupid assessment that digital is perfect, so it should sound perfect...Many of the poor sounding, came from bad recordings, especially converting stuff from LPs and tapes to CD's.

The first 3 CDs I bought (in 1985) together with a Philips CD101 (on sale for € 350 in todays currency) were:

• Joe Jackson Body and Soul (1984)
• Barbara Thompson Pure Fantasy (1984)
• Beethoven Symphony Nr. 9 by Otmar Suitner and the Staatskapelle Berlin (1982)

There was a time where Pure Fantasy sounded a bit too hard to me. Not any more (probably due to hearing loss), but the other albums are very good, Body and Soul is excellent and regarding dynamics probably the best pop recording I own.

 

[Mnyb]

The first Philips CD players used a 14 bit DAC with 4x oversampling to circumvent the steep reconstruction filter.

The first 3 CDs I bought (in 1985) together with a Philips CD101 (on sale for € 350 in todays currency) were:

• Joe Jackson Body and Soul (1984)
• Barbara Thompson Pure Fantasy (1984)
• Beethoven Symphony Nr. 9 by Otmar Suitner and the Staatskapelle Berlin (1982)

There was a time where Pure Fantasy sounded a bit too hard to me. Not any more (probably due to hearing loss), but the other albums are very good, Body and Soul is excellent and regarding dynamics probably the best pop recording I own.

Body and soul is excellent imho the actual recording trumphs all tech involved .

Wonder if even the origianl philips player would be possible to pick out in a blind test using normal music at typical levels .

You may hear the noise of it if sinad is not up to current standards ?

 

[mansr]

Amirm, have you considered testing a CD player from the mid 1980’s just to comment on the progress of precision timing and other possibly obvious to the ear improvements of 44.1/16 digital sources of the past 35 years?

There are some measurements of an early Philips CD player: https://www.audiosciencereview.com/forum/index.php?threads/philips-cd150-measurements.11296/

I tried to buy the original Sony CD player but the ebay seller wanted too much for it.

I have one of those. One of these days I should give it the tune-up it needs and run some measurements. One day. Not today. Sorry.

The CD player was one of the first marketed under Marantz brand.

Weren't those rebadged Philips players?

 

[mansr]

Not all sound as good as The Raven by Rebeka Pidgen (Spanish Harlem):

They must have been aiming for a world record in reverb with that one.

 

[LTig]

Body and soul is excellent imho the actual recording trumphs all tech involved .

Yep. A good recording trumps all.

Wonder if even the origianl philips player would be possible to pick out in a blind test using normal music at typical levels .

Well, when I replaced the CD101 in 1991 due to playback problems by a cheap Luxman D-113D CD-drive and an Arcam Black Box 3 DAC the new combi sounded much better to me, and when I replaced the Luxman drive by a Kenwood DP-X 9010 drive (which I got so cheap I couldn't resist) the sound improved even more - to my ears back then. However, knowing what I know today about subjective listening results I'm not so sure whether these impressions hold any truth at all.

You may hear the noise of it if sinad is not up to current standards ?

Don't think so. 14 bit with 4x oversampling is AFAIK comparable to 16 bit linear. The former has 4 times quantization noise but is s also spread over 4 times the bandwidth, so within the audible range quantization noise levels should be the same.

 

[Jerry Sobel]

Great video! So what is the purpose of these very expensive clocking devices? Is it to really make a change (your video says otherwise) or is it to make money by selling snake oil?

 

[LTig]

Great video! So what is the purpose of these very expensive clocking devices? Is it to really make a change (your video says otherwise) or is it to make money by selling snake oil?

If you talk about the devices sold to audiophiles it's certainly the latter. The only reason to use clocking devices at all is in a recording studio to synchronize all DACs and ADCs to one single master clock, not to improve sound (which they can't).

 

[PeteL]

Great video! So what is the purpose of these very expensive clocking devices? Is it to really make a change (your video says otherwise) or is it to make money by selling snake oil?

Not snake oil, master clocks outputing word clock are for professionnal environment where many equipment have to be synced, not judt a transmitter and a reciever, like digital consoles, recorders, etc.

 

[AndreaT]

There are some measurements of an early Philips CD player: https://www.audiosciencereview.com/forum/index.php?threads/philips-cd150-measurements.11296/


I have one of those. One of these days I should give it the tune-up it needs and run some measurements. One day. Not today. Sorry.


Weren't those rebadged Philips players?

There are some measurements of an early Philips CD player: https://www.audiosciencereview.com/forum/index.php?threads/philips-cd150-measurements.11296/


I have one of those. One of these days I should give it the tune-up it needs and run some measurements. One day. Not today. Sorry.


Weren't those rebadged Philips players?

Thanks!! Interesting!