Can DOS 2 Phantoms reproduce acoustic programme above 24kHz?

[streamy]

Indeed, that’s also what I start to question. Anyway, let me mention in what is the audiable part 20-20k I’m very happy with the performance of the Gold Phantoms.

[struts]

Indeed, that’s also what I start to question. Anyway, let me mention in what is the audiable part 20-20k I’m very happy with the performance of the Gold Phantoms.

Yes, same here, although mine are in our kitchen/dining area so 99% of the time they are playing background music while we are cooking or eating. That said this is the system in the house that gets used the most by far!

But I can’t help feeling that this is both sad and ironic. After all the anguish about support for hi res, 24/192, 24/96, speculation about which DAC chips, which inputs, which vintages of units, and attempts to divine the truth from contradictory spec sheets it turns out that Phantoms of all vintages may well have been ASRC-ing everything to 24/48 internally all along. I don’t know whether to laugh or cry!


Sent from my iPhone using Tapatalk

[ragwo]

Indeed, that’s also what I start to question. Anyway, let me mention in what is the audiable part 20-20k I’m very happy with the performance of the Gold Phantoms.

Yes, same here, although mine are in our kitchen/dining area so 99% of the time they are playing background music while we are cooking or eating.  That said this is the system in the house that gets used the most by far!

But I can’t help feeling that this is both sad and ironic.  After all the anguish about support for hi res, 24/192, 24/96, speculation about which DAC chips, which inputs, which vintages of units, and attempts to divine the truth from contradictory spec sheets it turns out that Phantoms of all vintages may well have been ASRC-ing everything to 24/48 internally all along.  I don’t know whether to laugh or cry!


Sent from my iPhone using Tapatalk

I find it quite hilarious. If it has been 24/48 all the way, placebo has fooled me time after time. It's just wonderful how our brain works! Reminds me of the time when I was tuning a PA with the parametric EQ in bypass mode. It took a while before I realized what was going on. I had tuned the PA in that venue hundred times before, and placebo hit me hard in the forehead. Still makes me laugh. 
@streamy Thanks for investigating into the matter, clever way to discover what's going on.

[streamy]

As follow-up I generated the files as a set of 24/48k flacs and played them. Surely, taking into account that the 25kHz sine signal was almost canceled out due to Nyquist violation. Surprisingly, or not, all signals played through and either I heard the 1kHz signal or the noise on the tweeters confirming that the drivers were not muted. This same as the tests before all tracks were played via the RPi hat through optical in.

In continuation I played now all 15 tracks through the Devialet Spark software. To my surprise the 25kHz sine signal was not muted for all three tracks (48k, 96k and 192k). Here only a capable microphone like the UMIK-2 will show if the Phantom really play hires. So there is still hope.
The same when playing the tracks through UPnP. All files were played, no muting!
Let’s wait and see what @struts can record once he receives the UMIK-2.

[struts]

As follow-up I generated the files as a set of 24/48k flacs and played them. Surely, taking into account that the 25kHz sine signal was almost canceled out due to Nyquist violation. Surprisingly, or not, all signals played through and either I heard the 1kHz signal or the noise on the tweeters confirming that the drivers were not muted. This same as the tests before all tracks were played via the RPi hat through optical in.

In continuation I played now all 15 tracks through the Devialet Spark software. To my surprise the 25kHz sine signal was not muted for all three tracks (48k, 96k and 192k). Here only a capable microphone like the UMIK-2 will show if the Phantom really play hires. So there is still hope.
The same when playing the tracks through UPnP. All files were played, no muting!
Let’s wait and see what @struts can record once he receives the UMIK-2.

More good stuff here [mention]streamy [/mention], thanks!

Just to be clear, when you say you then played the tracks “through Spark” can you describe the signal path? Were the tracks you played through the RPi not also controlled via Spark?

Hoping that proper measurements will show that an end-to-end 24/192 signal path has been supported at some point at least!


Sent from my iPhone using Tapatalk

[ogs]

We still have to remember that the speaker upsamples to 192kHz. That is what Devialet has told us. If that is correct or not is something else entirely... maybe right for 'old' Phantoms, but not the 'new'?

[alandbush]

Hoping that proper measurements will show that an end-to-end 24/192 signal path has been supported at some point at least! 

I can no longer find the post to support my recollection but I recall a report from a member who attended an Oxford Audio show where Matthieu Pernot (Devialet engineer) provided answers to forum members' questions.  It was reported that Matthieu informally opined that the Phantom, although accepting 192kHz, was unlikely to reproduce that signal because of the tweeter limitations.

If nothing else, that would suggest that, at the time, no resampling to less then 192kHz was applied.

Still, looking forward to the test measurements.

[struts]

I can no longer find the post to support my recollection but I recall a report from a member who attended an Oxford Audio show where Matthieu Pernot (Devialet engineer) provided answers to forum members' questions.  It was reported that Matthieu informally opined that the Phantom, although accepting 192kHz, was unlikely to reproduce that signal because of the tweeter limitations.

If nothing else, that would suggest that, at the time, no resampling to less then 192kHz was applied.

Still, looking forward to the test measurements.

Yes, I think the tweeter in the Gold is rated up to 27kHz, equivalent to a minimum sampling frequency of 54kS/s. So in theory a sample rate of 192kS/s is overkill for this tweeter, but in practice it may still sound better due to the effects of low pass filters within the audible band.


Sent from my iPhone using Tapatalk

[streamy]

As follow-up I generated the files as a set of 24/48k flacs and played them. Surely, taking into account that the 25kHz sine signal was almost canceled out due to Nyquist violation. Surprisingly, or not, all signals played through and either I heard the 1kHz signal or the noise on the tweeters confirming that the drivers were not muted. This same as the tests before all tracks were played via the RPi hat through optical in.

In continuation I played now all 15 tracks through the Devialet Spark software. To my surprise the 25kHz sine signal was not muted for all three tracks (48k, 96k and 192k). Here only a capable microphone like the UMIK-2 will show if the Phantom really play hires. So there is still hope.
The same when playing the tracks through UPnP. All files were played, no muting!
Let’s wait and see what @struts can record once he receives the UMIK-2.

More good stuff here [mention]streamy [/mention], thanks! 

Just to be clear, when you say you then played the tracks “through Spark” can you describe the signal path?  Were the tracks you played through the RPi not also controlled via Spark?

Hoping that proper measurements will show that an end-to-end 24/192 signal path has been supported at some point at least! 


Sent from my iPhone using Tapatalk

To clarify I’m still on DOS1 with my Phantoms. Devialet Spark is the SW that came with the Phantoms and has the capability to play from local folders, streaming services as Qobuz,Tidal, ... internet radio etc. I copied the tracks to a local folder on a MacMini that runs Spark. Then I can stream the files from the local MacMini to Dialog by playing them in Spark and Dialog distributes them to the Phantoms the way they are configured in ‘rooms’. The concept is similar to Roon.
Spark was always specified to be able to stream hires up to 24/192k. It actually does the job very well by now, fully gapless, showing transparently how much of the track is already buffered at the end point. But many Phantom users didn’t like Spark. For personally it is much better than the current DOS2 solution where you have to find inter radio software, streaming software.... of course Spotify connect was also always available. Better than Spark is Roon, even if I preferred the Qobuz integration in Spark as it included my purchases to be played in hires, better than my subscription. Ok, I’m deviating from what is relevant to the sound reproduction of the Phantoms.
When I played the files through Roon in Spark the optical outputs under “live sources” are chosen. This is my standard setup. The optical input per Room is selected in Spark. Then you don’t have to open Spark and run everything in Roon. On one of the RPi’s I’ve installed the “Devialet Phantom Volume Control” extension and therefore can control the Phantom volume directly in Roon. No need to start Spark.

[ogs]

I should also say that wide bandwidth is not only about amplitude, it's also about timing. The following was written by Morten Lindberg of 2L records. I had a brief email conversation with him on the merits of Redbook (16/44) vs high(er) rez. Translated from Norwegian: "Sample frequency: 44.1kHz is OK if you only evaluate frequency bandwidth. If we also look at the time domain we need higher bandwidth to give us an impulse response that can reproduce transients without pre and post ringing."