03-16-2019, 12:28 PM
(Modification du message : 03-16-2019, 12:51 PM par a supprimer merci.)
Pour dire les choses différemment:
- le test demontre que le lecteur ne change pas la representation digitale du fichier audio (les bits)
- ce que le test n'aborde pas c'est la qualité "electrique" du signal en sortie du lecteur (le bruit)
Quel que soit le mode de transmission du signal digital au DAC, coaxial, USB, Toslink, I2S, des perturbations électriques sont transmises au DAC.
Voici ce qu'explique, par exemple, John Brown, l'ingénieur d'ECDesigns:
"Source noise spectrum can be measured when streaming digital silence. However, it is likely that this spectrum changes radically when streaming music. When streaming music, source noise is masked by this much stronger data signal and the CPU load in the source is likely to increase considerably.
Based on my personal research the source dependency problem has to be (partially) caused by ripple voltage on the digital interface signal(s). With USB this translates to a ripple voltage on the differential signals, differential interfaces cancel common mode noise but fail to cancel source related (unequal) noise on both interface signals. With S/PDIF coax it translates to ripple voltage on the electrical signal. With Toslink the ripple voltage is translated to fluctuations in the light output and translated to ripple voltage in the optical receiver.
The problem is that we can't simply use a "voltage stabiliser) to get rid of this ripple as the bandwidth of such circuit would be far too low. Even zener diodes are not fast enough and introduce unwanted non-linear effects that make matters even worse.
The only practical solution is using cleanest possible digital audio source and this also applies to many other DACs, even if these are based on built-in low phase noise clocks and FIFO buffers.
- le test demontre que le lecteur ne change pas la representation digitale du fichier audio (les bits)
- ce que le test n'aborde pas c'est la qualité "electrique" du signal en sortie du lecteur (le bruit)
Quel que soit le mode de transmission du signal digital au DAC, coaxial, USB, Toslink, I2S, des perturbations électriques sont transmises au DAC.
Voici ce qu'explique, par exemple, John Brown, l'ingénieur d'ECDesigns:
"Source noise spectrum can be measured when streaming digital silence. However, it is likely that this spectrum changes radically when streaming music. When streaming music, source noise is masked by this much stronger data signal and the CPU load in the source is likely to increase considerably.
Based on my personal research the source dependency problem has to be (partially) caused by ripple voltage on the digital interface signal(s). With USB this translates to a ripple voltage on the differential signals, differential interfaces cancel common mode noise but fail to cancel source related (unequal) noise on both interface signals. With S/PDIF coax it translates to ripple voltage on the electrical signal. With Toslink the ripple voltage is translated to fluctuations in the light output and translated to ripple voltage in the optical receiver.
The problem is that we can't simply use a "voltage stabiliser) to get rid of this ripple as the bandwidth of such circuit would be far too low. Even zener diodes are not fast enough and introduce unwanted non-linear effects that make matters even worse.
The only practical solution is using cleanest possible digital audio source and this also applies to many other DACs, even if these are based on built-in low phase noise clocks and FIFO buffers.