Jitter problem

[One and a half]

USB 3.0 is fully compatible with USB 2.0. The only difference is the transfer rate (USB 3.0 is about ten times faster than 2.0). A USB 2.0 device will happily operate at USB 2.0 speed when connected to a USB 3.0 port, and USB 2.0 driver software functions perfectly well with 3.0 ports. And I have the El Capitan OS X. I've never heard of any USB issues. Anyway, I used the Toslink output of the Mac so any USB issue is irrelevant.

 

For what it's worth, the USB3.0 implementation in my MacBookPro Retina 2013 has achieved these notable milestones.

 

- FW upgrade to Grace Design m920 via USB - FAILED (OSX & WIN)

- USB Playback of Grace design M920 for DoP64 & 128 - FAILED (OSX & WIN)

- KORG DS-DAC-10 USB playback >96kHz - FAILED (WIN)

- USB FW upgrade for Samsung BD Writer SE-506 - FAILED (OSX & WIN)

 

All the above issues were fixed when using an 'older' genuine USB 2.0 port. That's why I'm suspicious of Apple/Intel USB3.0 ports properly changing gears for USB Audio Class 2 devices.

 

As for using the optical out on the Mac Mini, have a look at this post.

Yes the post is so 'old' it may not mean anything now in this modern age of advanced knowledge, however manufacturing makes things cheaper, not necessarily better.

[Abtr]

For what it's worth, the USB3.0 implementation in my MacBookPro Retina 2013 has achieved these notable milestones.

- FW upgrade to Grace Design m920 via USB - FAILED (OSX & WIN)

- USB Playback of Grace design M920 for DoP64 & 128 - FAILED (OSX & WIN)

- KORG DS-DAC-10 USB playback >96kHz - FAILED (WIN)

- USB FW upgrade for Samsung BD Writer SE-506 - FAILED (OSX & WIN)

All the above issues were fixed when using an 'older' genuine USB 2.0 port. That's why I'm suspicious of Apple/Intel USB3.0 ports properly changing gears for USB Audio Class 2 devices.

I have the same problematic results using USB 2.0 ports, albeit with Windows PCs. I even tried laptops with both USB 2.0 and USB 3.0 and there is no measurable or audible difference between Windows USB 2.0 or 3.0, or Mac USB 3.0.

 

As for using the optical out on the Mac Mini, have a look at this post.

Yes the post is so 'old' it may not mean anything now in this modern age of advanced knowledge, however manufacturing makes things cheaper, not necessarily better.

I suppose it's possible that the optical out of the Mac contains "a ton of jitter", but what strikes me is that its USB output apparently contains exactly the same amount of jitter. This leads me to believe that there's a common cause, but I might be wrong. I will have to do more precise measurements to be sure. Anyway, the (asynchronous) MX-DAC not only buffers and reclocks USB input but also optical and coaxial input. Thus, the DAC's optical input should be immune to just jitter as long as the (SPDIF) input signal is bit perfect, which it obviously is not..

[sandyk]

[Thus, the DAC's optical input should be immune to just jitter as long as the (SPDIF) input signal is bit perfect, which it obviously is not.. /QUOTE]

 

What makes you so sure it isn't ?

[Abtr]

What makes you so sure it isn't ?

As I said, the DAC buffers and reclocks its optical input. So any sound degradation cannot be caused by only jitter. Computer jitter must have become analogically encoded (through some mysterious opamping) into the otherwise bit perfect audio stream. I wouldn't call the result bit perfect as relative to the original bits in the audio file..

[Abtr]

Another update. I tried a Synology DS216se. This NAS has a Linux based OS (DSM 6.0) and it has the explicit option to play 'high quality audio' directly through its USB class 2 output to a DAC. Alas, the significant audible and measurable difference between computer - and CD transport remained unchanged. So, by now we have three very different computers (Windows, Mac and Linux) degrading audio to the same extent and most probably in the same way.

What can this be? My guess is that in some processing stage, in all three computer systems, for some reason, the audio signal is either compressed or DA-AD converted. At least that's how the result sounds and measures..

[MrSprout]

I've been switching back and forth between PC and CD for the last year or so. The PC always sounds better, but it takes up so much shelf space with all the converters and power supplies that I get annoyed and stick the CD player back in, only for me to miss the transparency and timing of the PC and then switch back again...

 

In all that time I've never noticed a clear difference in frequency response between the two, but the little Zotac NUC and Asus laptop I tried were perhaps slightly duller than my blinged out PC now I think about it. I just put it down to a lack of transparency at the time and stopped using them (the Asus was old and the Zotac was an experiment to see if I could shrink the PC and get the same sound quality - nope). Both the PC and now my retina macbook pro running Win7/Foobar/WASAPI sound the same as my CD transport in terms of frequency response I'm sure. I'm using the Macbook right now feeding a Caiman 2 with a sysconcept toslink cable and its great - a little lacking in clarity and composure compared to the PC (partly due to reading files over wi-fi), but its at least a match for the CD player.

 

I saw the comment earlier about de-emphasis. My knowledge on that front is limited but did the CD get burned with any settings like that? Ie is it the CD that's wrong because of the way it was written? It definitely won't be DA-AD in the computers if you're using USB out.

 

One other thought is that I got noticeable differences in presentation between CAT 6a and CAT 7 network cables (preferring CAT 7) when using a Yamaha streamer. I get better results still using local storage on the PC. Have you tried that, or tried a local USB stick?

 

And for the record I tried using my QNAP NAS as a player as well, but it wasn't any better than the old laptop or Zotac. Very uninspiring. It might have got better if I'd left it playing for longer (USB port burn-in?), but I cut my losses and went back to PC.

[MrSprout]

I've remembered something else: shortly after getting my Hiface EVO USB converter I experimented with USB cables, comparing Audioquest, Cardas and two different generic (well used) printer cables. One of the generic cables was so dull it was as though it was blunted in the treble. The other (with a graphite ring) was really exciting but a bit messy and bright in the treble. And that was just the freebies! The Audioquest and Cardas cables both went through very different burn in processes and both sounded crap to start with. Your computer set-up always has a USB cable in it I believe, so maybe try another one that looks physically different and see if that changes anything?

[Abtr]

I've been switching back and forth between PC and CD for the last year or so. The PC always sounds better, but it takes up so much shelf space with all the converters and power supplies that I get annoyed and stick the CD player back in, only for me to miss the transparency and timing of the PC and then switch back again...

 

In all that time I've never noticed a clear difference in frequency response between the two, but the little Zotac NUC and Asus laptop I tried were perhaps slightly duller than my blinged out PC now I think about it. I just put it down to a lack of transparency at the time and stopped using them (the Asus was old and the Zotac was an experiment to see if I could shrink the PC and get the same sound quality - nope). Both the PC and now my retina macbook pro running Win7/Foobar/WASAPI sound the same as my CD transport in terms of frequency response I'm sure. I'm using the Macbook right now feeding a Caiman 2 with a sysconcept toslink cable and its great - a little lacking in clarity and composure compared to the PC (partly due to reading files over wi-fi), but its at least a match for the CD player.

 

I saw the comment earlier about de-emphasis. My knowledge on that front is limited but did the CD get burned with any settings like that? Ie is it the CD that's wrong because of the way it was written? It definitely won't be DA-AD in the computers if you're using USB out.

 

One other thought is that I got noticeable differences in presentation between CAT 6a and CAT 7 network cables (preferring CAT 7) when using a Yamaha streamer. I get better results still using local storage on the PC. Have you tried that, or tried a local USB stick?

 

And for the record I tried using my QNAP NAS as a player as well, but it wasn't any better than the old laptop or Zotac. Very uninspiring. It might have got better if I'd left it playing for longer (USB port burn-in?), but I cut my losses and went back to PC.

Hi MrSprout - SQ may very much depend on the CD transport you are using, though I personally don't have any that sounds worse than a PC. I'm not sure my MX-DAC (or any DAC for that matter) reclocks only USB input and not toslink or coaxial input, so certain CD transports may sound worse than computers.. And I ripped the test tracks from an original CD and burned these back on a CD with consistent results: clean CD performance versus massive jitter with PC playback.

 

 

I've remembered something else: shortly after getting my Hiface EVO USB converter I experimented with USB cables, comparing Audioquest, Cardas and two different generic (well used) printer cables. One of the generic cables was so dull it was as though it was blunted in the treble. The other (with a graphite ring) was really exciting but a bit messy and bright in the treble. And that was just the freebies! The Audioquest and Cardas cables both went through very different burn in processes and both sounded crap to start with. Your computer set-up always has a USB cable in it I believe, so maybe try another one that looks physically different and see if that changes anything?

I used a number of different USB cables (including AQ Cinnamon) and toslink between Mac Mini and DAC, all with no audibly or measurably different results. (Ethernet cables are irrelevant here.)

[MrSprout]

Hi MrSprout - SQ may very much depend on the CD transport you are using, though I personally don't have any that sounds worse than a PC. I'm not sure my MX-DAC (or any DAC for that matter) reclocks only USB input and not toslink or coaxial input, so certain CD transports may sound worse than computers..

 

I have the Micromega T-Drive and Cambridge CXC transports. I've also used a Pioneer PD91, Sony CDP-337ESD (KSS-190A), Pioneer PDS-703 and various other players. Plus a Tascam DA-3000 for a while, although the interface on that was frustrating with large capacity CF cards. The blinged out PC beats them all, although most players improved with a Theta TLC or W4S Remedy added on to reclock the spdif - things got a lot closer then.

[Abtr]

I have the Micromega T-Drive and Cambridge CXC transports. I've also used a Pioneer PD91, Sony CDP-337ESD (KSS-190A), Pioneer PDS-703 and various other players. Plus a Tascam DA-3000 for a while, although the interface on that was frustrating with large capacity CF cards. The blinged out PC beats them all, although most players improved with a Theta TLC or W4S Remedy added on to reclock the spdif - things got a lot closer then.

Okay, without venturing into the realm of subjective listening and personal taste, I can hear details in the sound from a 16/44.1 CD transport (Arcam FMJ CD23T) connected to my MX-DAC that are inaudible when playing the same tracks from a PC (Windows, Mac OSX, or Linux), even when playing in so called exclusive 'bit perfect' (ASIO/WASAPI) mode. And the objective test I presented earlier clearly demonstrates the presence of excessive jitter in the digital audio signal of all my personal computer systems, measured as a selective degradation or cancelation of higher frequencies in the analogue domain, which is totally absent with the CD transport.

 

I don't know how it's possible that your (subjective) experience indicates the opposite is true. Until I see some objective measurements, I must assume that you prefer jitter degraded sound over a pure untouched audio signal..

[MrSprout]

I don't know how it's possible that your (subjective) experience indicates the opposite is true. Until I see some objective measurements, I must assume that you prefer jitter degraded sound over a pure untouched audio signal..

 

Now you're just being silly. :-) The PC offers cleaner, clearer, more intelligible results than the CD players. That wasn't true until I cleaned up the electrical noise (galvanic isolation, better PSUs, mains filtering, feeding the SSDs from separate supplies, feeding the USB card and USB converter from seperate supplies) and using WASAPI. The OS choice can also result in cleaner sound, for example AP Linux was much cleaner than Win7 but more fiddly. The latest version of AP Linux doesn't work with my USB converter unfortunately, and that's not an issue you get with CD players. When I've replaced my USB card in the main system I'll see if I can send you some files (I broke the old one).

 

My Macbook/Win7 is still sounding just as entertaining and lively as my CD player - it's very musical. It's not perfect, but whether or not that translates into your graphs I don't know. I doubt I could tell them apart in a blind test.

[Abtr]

Now you're just being silly. :-) The PC offers cleaner, clearer, more intelligible results than the CD players. That wasn't true until I cleaned up the electrical noise (galvanic isolation, better PSUs, mains filtering, feeding the SSDs from separate supplies, feeding the USB card and USB converter from seperate supplies) and using WASAPI. The OS choice can also result in cleaner sound, for example AP Linux was much cleaner than Win7 but more fiddly. The latest version of AP Linux doesn't work with my USB converter unfortunately, and that's not an issue you get with CD players. When I've replaced my USB card in the main system I'll see if I can send you some files (I broke the old one).

 

My Macbook/Win7 is still sounding just as entertaining and lively as my CD player - it's very musical. It's not perfect, but whether or not that translates into your graphs I don't know. I doubt I could tell them apart in a blind test.

Well, if it's possible to upgrade a PC or any of its components by using (separate) linear PSUs, that would demonstrate that DA-AD conversion (opamping) is going on somewhere in the PC's audio chain even in exclusive (WASAPI) modus. I use the Intona galvanic isolator/reclocker and an Audio-gd asynchronous USB converter/reclocker and I can't get decent sound from a PC. This basically means that (at least without exotic mods) a PC is simply unfit for audiophile sound reproduction..

[davide256]

Well, if it's possible to upgrade a PC or any of its components by using (separate) linear PSUs, that would demonstrate that DA-AD conversion (opamping) is going on somewhere in the PC's audio chain even in exclusive (WASAPI) modus. I use the Intona galvanic isolator/reclocker and an Audio-gd asynchronous USB converter/reclocker and I can't get decent sound from a PC. This basically means that (at least without exotic mods) a PC is simply unfit for audiophile sound reproduction..

 

Which is why Aurender and Lumin have a market...

[esldude]

Well, if it's possible to upgrade a PC or any of its components by using (separate) linear PSUs, that would demonstrate that DA-AD conversion (opamping) is going on somewhere in the PC's audio chain even in exclusive (WASAPI) modus. I use the Intona galvanic isolator/reclocker and an Audio-gd asynchronous USB converter/reclocker and I can't get decent sound from a PC. This basically means that (at least without exotic mods) a PC is simply unfit for audiophile sound reproduction..

Certainly entertaining posts. Some whimsical conflation of ideas.

[Abtr]

To summarize: 1) All normally used PC-type transports for digital audio-files produce about the same excessive amount of jitter. I tried the USB and Toslink output of a number of Windows PCs, one Mac Mini running OSX and a Linux NAS, all connected to the same DAC, and all show a comparable level of sound degradation in the analogue domain caused by digital timing errors. 2) The Toslink or coax output of a decent CD-transport, playing through the same DAC, contains no jitter and doesn't show any (measurable) degradation of analogue sound.

 

The conclusion must then be: either some (highly unlikely) analogue processing and subsequent resampling, or some digital processing and resampling (e.g. floating point sound level control) is going on inside the PC, i.e., at some processing stage in between the audio-file and the USB data-stream.

 

As far as a linear DC power supply is concerned, this is important for USB to S/PDIF conversion and the final digital to analogue conversion where opamping takes place. These are the most likely sources of audible timing errors, not the PC. But a proper modern DAC with a proper asynchronous USB implementation and a proper linear PSU produces only a minute amount of jitter which is practically unmeasurable and inaudible.

 

Now I can't be the only one that encountered this problem. Possibly many members of the computer audio community have become used to the level of jitter produced by PCs. And jitter has the ability to mask other (analogue) problems in a sound system. When jitter is reduced, the other problem may become more obvious and more disagreeable.. Another possibility remains that I'm overlooking some error in my own system. So any reasonable suggestions are welcome..

[esldude]

You do realize you haven't measured jitter? You do realize some of your ideas simply aren't how things work? My suggestion would be to listen to Mansr's suggestions earlier in the thread and approach things that way if you want to learn what is going on. You don't have a good conception of how digital to analog conversion works, op-amping is not a thing, and you really need to back up and listen to someone about what you think is jitter, in fact is not jitter.

 

These are reasonable suggestions.

[Abtr]

You do realize you haven't measured jitter?

No I don't, I measured the analogue consequence/result of digital timing errors. This is called 'jitter'.

 

You do realize some of your ideas simply aren't how things work?

I know one of my ideas is unlikely to be true, but I nevertheless presented it if only to illustrate the type of possible cause we may be looking for here.

 

My suggestion would be to listen to Mansr's suggestions earlier in the thread and approach things that way if you want to learn what is going on.

Are you suggesting I do a so called J-test? That will only reveal the quality of the USB signal, not the actual timing errors that are already contained therein.

 

You don't have a good conception of how digital to analog conversion works, op-amping is not a thing, and you really need to back up and listen to someone about what you think is jitter, in fact is not jitter.

 

These are reasonable suggestions.

What I think is jitter, in fact is not jitter? How is that a reasonable suggestion? And op-amping is the big thing in DA-conversion, there the audible difference between DACs largely originates..

[jhwalker]

No I don't, I measured the analogue consequence/result of digital timing errors. This is called 'jitter'.

 

 

I know one of my ideas is unlikely to be true, but I nevertheless presented it if only to illustrate the type of possible cause we may be looking for here.

 

 

Are you suggesting I do a so called J-test? That will only reveal the quality of the USB signal, not the actual timing errors that are already contained therein.

 

What I think is jitter, in fact is not jitter? How is that a reasonable suggestion? And op-amping is the big thing in DA-conversion, there the audible difference between DACs largely originates..

 

You have measured a treble rolloff and *assumed* it's jitter.

[Abtr]

You have measured a treble rolloff and *assumed* it's jitter.

I have measured a treble rolloff in the analogue reproduction by multiple PCs of an original digital (WAV encoded) noise signal, which is absent in a reproduced pure sine sweep as well as in the same reproduced noise signal streamed from a CD-transport. This can only be explained by digital timing errors ('jitter') and subsequent resampling somewhere in the PC, i.e., somewhere in between the original input file and the (USB) output stream..

[jhwalker]

I have measured a treble rolloff in the analogue reproduction by multiple PCs of an original digital (WAV encoded) noise signal, which is absent in a reproduced pure sine sweep as well as in the same reproduced noise signal streamed from a CD-transport. This can only be explained by digital timing errors ('jitter') and subsequent resampling somewhere in the PC, i.e., somewhere in between the original input file and the (USB) output stream..

 

". . . can only be explained by digital timing errors ('jitter') . . . " sounds like a big assumption to me, and one that is not proven; i.e., I think you're leaving out any number of other possible causes for this observation.

 

It's like saying, "One of my two cars is getting poor gas mileage; in fact, I have continually observed poor gas mileage in one of my cars, but not the other. Someone told me once that low air pressure in my tires could cause poor gas mileage. I don't know how to measure the air pressure in my tires, but it *must* be low air pressure causing the poor gas mileage - nothing else could be causing that." Seems like a bit of a leap.

 

But I'll bow out here, as I'm no expert. Perhaps others can chime in on other possible causes besides "jitter".

[esldude]

No I don't, I measured the analogue consequence/result of digital timing errors. This is called 'jitter'.

 

No you haven't. The way to measure the analog output for jitter is with a quarter sample rate tone. 11,025 khz for 44.1 material, or 12 khz for 48 khz sources. The Jtest is that tone plus a low level tone to run through all the various bits and stimulate jitter prone to the SPIDF interface (not USB). The sidebands around just the quarter sample rate tone will give some good insight into jitter levels.

 

Jitter is the result of samples occurring too fast or too slow and creating noise sidebands. Jitter is not going to show up as treble roll off. Not sure where you got that idea, but you are mistaken about it. Trying to solve a diagnosed problem from mistaken premises and faulty measurement methods will get you no where. Jitter is too small an effect to show up using a microphone as its effects will be buried in noise.

 

Are you suggesting I do a so called J-test? That will only reveal the quality of the USB signal, not the actual timing errors that are already contained therein.

 

See my above response

 

What I think is jitter, in fact is not jitter? How is that a reasonable suggestion? And op-amping is the big thing in DA-conversion, there the audible difference between DACs largely originates..

 

I don't know what "op-amping" is. It isn't a big thing in DA conversion as it isn't a thing. Many DACs feed the output to an op-amp to amplify the signal. You can also do that amplifying with tubes or discrete transistors. It is unreasonable to use a term for something it is not, and then proceed to test it in a way that doesn't test for it, and then proceed based upon a flawed and mistaken premise with flawed and mistaken test results. You could only end up with a flawed and mistaken fix.

 

 

Try reading this to learn what jitter is. With examples of good and poor jitter measurements.

NwAvGuy: Jitter Does it Matter?

 

It is okay to be mistaken. I have been myself at least once possibly even twice (yes that is supposed to be a joke knowing the mistakes I have made). So don't take this personally, it isn't meant that way. It is advice about what jitter is and to tell you the results you are seeing are from something else altogether. If you will listen, and take some advice (like Mansr tried to give) there are people here who will help you figure out what is actually going on in your acoustic measured results.

 

If you continue as you have, then jhwalker's analogy was just about right for what you are doing.

[Superdad]

Dennis, you have the patience of Job!

This thread really belongs in the dustbin. I've been speechless from the start. No disrespect meant Abtr.

[mansr]

The test files contain pink noise and logarithmic sine sweeps. This means the average power spectrum of both signals has the same 10 dB per decade roll-off. Playing the files on an iFi nano DAC (volume control set to output 1 Vrms at 1 kHz) and capturing its line output, I get these results:

 

 

As we can see, the output of the Nano tracks the test signals perfectly aside from a drop-off at very low frequencies (the capture device also contributes to this).

 

Make whatever you want of this.

[Abtr]

No you haven't. The way to measure the analog output for jitter is with a quarter sample rate tone. 11,025 khz for 44.1 material, or 12 khz for 48 khz sources. The Jtest is that tone plus a low level tone to run through all the various bits and stimulate jitter prone to the SPIDF interface (not USB). The sidebands around just the quarter sample rate tone will give some good insight into jitter levels.

 

Jitter is the result of samples occurring too fast or too slow and creating noise sidebands. Jitter is not going to show up as treble roll off. Not sure where you got that idea, but you are mistaken about it. Trying to solve a diagnosed problem from mistaken premises and faulty measurement methods will get you no where. Jitter is too small an effect to show up using a microphone as its effects will be buried in noise.

First of all, J-tests have been performed with various PC/laptops e.g. here: Archimago's Musings: MEASUREMENTS: Part I: Bit-Perfect Audiophile Music Players (Windows). and here: Archimago's Musings: MEASUREMENTS: Bit-Perfect Audiophile Music Players (Mac OS X). These measurements indicate that although any laptop or notebook produces a fair amount of jitter, any PC can output a bit perfect digital audio signal without problems. And Mansr and I already agreed that the digital audio output of a PC doesn't need to be jitter free; as long as it is bit perfect and (asynchronously) re-clocked in the DAC all should be well in the analogue sound domain.

 

What I meant by 'opamping' is the actual digital to analogue conversion in a DAC where opamps commonly perform a (hopefully) very precisely timed summation of binary weighted currents which is then converted to a proportional output voltage. I may have used the term incorrectly, but my point is that the DAC should be the only stage in the entire computer audio chain where jitter (incorrect timing) may audibly degrade the final analogue signal.

 

Now consider a situation where the digital sound signal is somehow processed and then resampled in the PC. That would encode the PC's jitter into the audio signal (just like recording jitter). The signal is then no longer bit perfect and a J-test, as I understand it, might not detect this if said resampling isn't applied to the played back J-test test file. Nevertheless, it would be audible in the resulting analogue sound. I'm not saying that this is actually happening, but I think I have eliminated all other possible causes of the obvious audible difference between PC playback and CD-transport that I experience.

 

Finally, regarding the specific jitter-test that I use, let's agree that (excessive) digital jitter translates in a clearly audible analogue effect (which is commonly also called jitter). So why is it so strange that jitter can simply be measured in the analogue domain? Using a noise signal that contains all frequencies in known quantities or magnitudes, clock jitter (or clock phase noise) will result in small phase-shifts that will predominantly shift and (partially) cancel higher frequencies. An FFT algorithm then measures what's left of each frequency component in the analogue sound. And although such a noise signal is extremely susceptible to jitter, in my experience this provides a very useful indication of the amount of sound degradation by jitter in the audio chain.

[Abtr]

The test files contain pink noise and logarithmic sine sweeps. This means the average power spectrum of both signals has the same 10 dB per decade roll-off. Playing the files on an iFi nano DAC (volume control set to output 1 Vrms at 1 kHz) and capturing its line output, I get these results:

 

[ATTACH=CONFIG]25262[/ATTACH][ATTACH=CONFIG]25263[/ATTACH]

 

As we can see, the output of the Nano tracks the test signals perfectly aside from a drop-off at very low frequencies (the capture device also contributes to this).

 

Make whatever you want of this.

Hi Mansr, what transport did you use?