Jitter vs no-Jitter Part III

[esldude]

I do not think that this test is appropriate for perceiving "the sound of jitter". When most people talk about jitter in playback they are referring to the jitter in the DAC actually producing the analog signal they are listening to. This is NOT what is being tested here.

 

In these tests we have two identical DACs, with one with supposedly higher jitter. But "WE" the test takers, are not listening to the output of those two DACs. For us the analog out of those DACs are sent to an ADC, which converts back to digital, that digital gets sent (somehow) to OUR DACs which generate another set of analog signals, which are what we are actually listening to.

 

The jitter in OUR DACs does not change when listening to the different tracks, so we are not actually listening to jitter. Think about what is happening here, the analog outs from the two original DACs are converted to digital by an ADC with its own jitter and its own digital filter. That goes to a DAC with ITS own jitter and digital filter. What is the probability of this chain faithfully reproducing the small differences in analog out from the two original DACs due to a jitter difference? I would say pretty small.

 

So what I think we are actually hearing is how jitter affects the interactions between multiple digital <-> analog conversions, NOT the effects of jitter itself.

 

So I don't necessarily believe that people prefer the jitter, but that the jitter is acting as a kind of dither in the multiple conversions.

 

This test really needs to be done on the actual analog outs of the two original DACs in order for it to really be about the "sound of jitter" rather than the effects of jitter on multiple conversions.

 

BTW I chose B and D. I have built many different DACs for which I do know which has higher and lower jitter, the differences I hear in B and D are similar to what I hear with lower jitter DACs. It is interesting that D sounded the best (by a fairly high margin). I need to try an experiment where I add a little bit of digital dither before the upsampling filter in one of my DACs.

 

John S.

 

Say what????

 

If one DAC has jitter of rather high levels and one doesn't, you record that then you get one signal with a little jitter, plus a little more from the ADC plus a little more from a good DAC upon playback. The other signal has the results of high jitter, plus a little jitter from the ADC, plus a little more from a good DAC upon playback. Giving one signal with the sound of 3 small amounts of jitter. Another signal with 2 small amounts of jitter and one large amount of jitter.

 

I agree best comparison is directly between low and high jitter without the in between effects. I don't agree that since jitter manifests itself as low level noise then you are merely hearing something like dithering. If both signals are that low in jitter level then you may not hear it either directly or recorded. Without knowing the levels of jitter in these tests one couldn't say if it is surely audible.

 

I am not saying these files had jitter that was audible. Test results don't indicate it was reliably audible (skipping over issues with this type of testing). I don't think the fact we weren't listening to recordings is a reasonable explanation. I do get that some very low level noise will pass thru upsampling with variable results. I don't see how jitter causing such low levels would also be readily audible directly, but not when going through an extra ADC/DAC conversion.

[sandyk]

In these tests we have two identical DACs, with one with supposedly higher jitter. But "WE" the test takers, are not listening to the output of those two DACs. For us the analog out of those DACs are sent to an ADC, which converts back to digital, that digital gets sent (somehow) to OUR DACs which generate another set of analog signals, which are what we are actually listening to.........

This test really needs to be done on the actual analog outs of the two original DACs in order for it to really be about the "sound of jitter" rather than the effects of jitter on multiple conversions.

 

That has always been my contention too, yet people like Archimago, Mitchco, etc. keep insisting about the validity of the results of their measurements taken after yet another A to D conversion.

[bibo01]

I do not think that this test is appropriate for perceiving "the sound of jitter". When most people talk about jitter in playback they are referring to the jitter in the DAC actually producing the analog signal they are listening to. This is NOT what is being tested here.

 

In these tests we have two identical DACs, with one with supposedly higher jitter. But "WE" the test takers, are not listening to the output of those two DACs. For us the analog out of those DACs are sent to an ADC, which converts back to digital, that digital gets sent (somehow) to OUR DACs which generate another set of analog signals, which are what we are actually listening to.

 

The jitter in OUR DACs does not change when listening to the different tracks, so we are not actually listening to jitter. Think about what is happening here, the analog outs from the two original DACs are converted to digital by an ADC with its own jitter and its own digital filter. That goes to a DAC with ITS own jitter and digital filter. What is the probability of this chain faithfully reproducing the small differences in analog out from the two original DACs due to a jitter difference? I would say pretty small.

 

So what I think we are actually hearing is how jitter affects the interactions between multiple digital <-> analog conversions, NOT the effects of jitter itself.

 

So I don't necessarily believe that people prefer the jitter, but that the jitter is acting as a kind of dither in the multiple conversions.

 

This test really needs to be done on the actual analog outs of the two original DACs in order for it to really be about the "sound of jitter" rather than the effects of jitter on multiple conversions.

 

BTW I chose B and D. I have built many different DACs for which I do know which has higher and lower jitter, the differences I hear in B and D are similar to what I hear with lower jitter DACs. It is interesting that D sounded the best (by a fairly high margin). I need to try an experiment where I add a little bit of digital dither before the upsampling filter in one of my DACs.

 

John S.

 

John, you are wrong because the ADC used in this test is able (it actually did) to detect phase jitter around a few femto-seconds. So it does not introduce a masking alteration. It just shows digitally the exact result of the analog signal converted by the DAC and its alteration.

 

This is confirmed by the fact that this ADC never had its "presence" heard by making double-conversion with several DACs using different architectures (sigma-delta, ladder, multibit).

 

To say that the next time you playback jitter does not change it is wrong because the file already contains a nonlinear alteration, which goes beyond the jitter produced by a DAC with modern oscillators. This alteration also reoccurs regularly by re-recording signal from analog output stage, providing one does not use a DAC that produces more than 200ps of jitter.

At best, a small quantity will be added by the conversion of the playing DAC. However, the alteration remains present in a fairly obvious shape.

 

This is confirmed by Tom Capraro's NTD*. Jittered files are always less consistent than those with less-jitter when both are compared to the original master signal.

 

Dithering is non important here because it enters at very low and subtle levels. Here we are far beyond quantization noise - we are talking about waveform presenting a different geometry.

 

Finally, listening directly from DACs' analog outputs (I tried) does not change a thing compared to what emerged with listening to audio files.

 

EDIT: later on I will post graphics about recording from jittered analog output

 

 

* This is the only measuring system that has demonstrated an improvement in some systems with your designed Regen using a dynamic signal.

[bibo01]

Follow up to my above post.

 

The following pictures are acquisitions from analog output of DAC with broken clock (and high jitter). One can see measurements of jitter detected from analog output, and jitter detected with jittered acquisition instead of FC/4 tone.

 

One can CLEARLY notice that all sidebands and noise spikes correspond so perfectly by superimposition. The only change, as seen in the second picture, is that some sidebands produce a 9dB higher amplitude after the second conversion.

 

Indeed, we can say that +9dB corresponds to about +300 picoseconds of jitter. So potentially a user would find himself listening to a file with jitter below hearing threshold, and another one with about 500 picoseconds of jitter.

 

Therefore, there is no risk that the jitter may level off in our double conversion passage. As a matter of fact, all of the above further supports our test - in reality, listening with various DACs, one obtains the same sideband and noise with a further increased level of jitter.

 

 

[bibo01]

Just to complete what I was saying above...

 

I made a direct acquisition of jitter (from broken DAC) and D/A conversion, but this time converting jittered file with the good DAC.

Amazingly, the graphics overlap perfectly and side bands show no disparity in amplitude.

We can then say that in the "broken" DAC a subsequent conversion increases side bands of extra 9dB, whilst with the good DAC jitter (going through the process A/D + D/A + A/D) undergoes no variation, keeping intact the estimated value of 200ps.

This is an important factor for the conversion chain as it proves to be highly transparent (it adds nothing and takes away nothing).