Phantom RAAT after firmware 2.14.4 update plays not 96k/24 lossless, only shows it

[ragwo]

Thanks a lot @struts, even if it's kind of bad news.

[struts]

My pleasure @ragwo, although I personally see it more as no news than bad news.

Have to admit that given Devialet’s track record I am a little bit surprised that so many people took this at face value and got their hopes up.  If they had really broadened the signal pipe from 24/48 to 24/96 Devialet would have been shouting it from the rooftops and would probably even have given it a new TLA! 

[ogs]

My pleasure @ragwo, although I personally see it more as no news than bad news.

Have to admit that given Devialet’s track record I am a little bit surprised that so many people took it at face value and got their hopes up.  If they had really broadened the signal pipe from 24/48 to 24/96 Devialet would have been shouting it from the rooftops and would probably even have given it a new TLA! 

I believe @streamy  wrote "..192k/24 is converted to 96k/24 and 176.4k/24 to 88.2k/24." If this is correct the resampling routine in Phantoms is changed. I seem to remember that Phantoms could deal with 'up to 96kHz' in the previous version?

[struts]

I believe @streamy  wrote "..192k/24 is converted to 96k/24 and 176.4k/24 to 88.2k/24." If this is correct the resampling routine in Phantoms is changed. I seem to remember that Phantoms could deal with 'up to 96kHz' in the previous version?

Hi @ogs!

He did indeed and I’ll leave it to him to explain what he meant by that, but other than the slightly cryptic information on Devialet’s website as far as I am concerned Phantom is a black box. You can measure what is going in (i.e. which protocols/resolutions are accepted at the inputs) and what comes out (i.e. acoustically), but as to what is going on in between that is pure speculation.

We thought that Roon’s signal path was giving us a little peek inside the black box but even that turned out to be a bit of an “unreliable witness”. But the measurements don’t lie, regardless of what goes in (as far as the DOS 2 inputs are concerned, i.e. except for Direct Optical) what comes out is never in excess of 24/48.


Sent from my iPhone using Tapatalk

[ogs]

Maybe it is Roon that does the "...192k/24 is converted to 96k/24 and 176.4k/24 to 88.2k/24."? Roon does power-of-two resampling for compatibility reasons if one tells it to. It's important that Phantoms report at least 96kHz max to Roon for those who use MQA. If 48kHz is max on a device the MQA processing in Roon is disabled.
Would you happen to know if a pair of Phantoms plays 44.1 natively or is this up-sampled to 48kHz?

[struts]

Would you happen to know if a pair of Phantoms plays 44.1 natively or is this up-sampled to 48kHz?

I would guess that it plays everything up to 24/48 natively. Most of Devialet’s stability problems seem to be related to network bandwidth/latency in wireless scenarios in conjunction with higher bitrate signals, so why increase the bitrate artificially?

But that is pure speculation. I am not aware of any way I can test for either zero-padding at the LSB or upsampling by measuring the acoustic output.


Sent from my iPhone using Tapatalk

[struts]

Thanks @struts for doing the measurements and showing the plots here. Indeed, what I can see from your attached pics indicated clearly that nothing has changed when it comes to what can be heard acoustically.

I logged tickets with both Devialet (7th Feb) and Roon (31st Jan) on this issue.  While Roon support has yet to even acknowledege my request Devialet picked it up within 2 days and today I received this email from Paul at Devialet customer service:

Paul (Devialet)
Feb 22, 2022, 5:01 PM GMT+1

Hello Struts,

I hope this mail finds you well.

Our team has found the origin of this issue, and a fix will be provided with a future update 🙂️

We do not have any release date yet, so it will most certainly take several weeks.

I'm keeping your file open until then, to inform you when the update will be available, and to check with you the downsampling step is correctly shown afterward.

Kindest regards,

Paul | Customer service & support Devialet

So:
1. Full marks to Devialet for their handling of my ticket.  Prompt, courteous and completely satisfactory for me at least.  Thank you Paul!  I’ll be keeping an eye out for the fix in the next maintenance release.
2. Downsampling to 24/48 is definitely still taking place as shown by my measurements.  It was, as suspected, simply a regression bug in the Signal Path reporting in RAAT introduced in firmware 2.14.4.

@streamy you might like to edit the subject of this thread to show that this turned out not to be true.  Would save folks browsing the forum getting their hopes up!


Sent from my iPhone using Tapatalk

Just tested this in the latest 2.16.1 firmware and can confirm that the bug is fixed. I played 24/192 and 24/96 source material from my library via RAAT to my Phantoms and the SRC to 24/48 is now correctly being reported by Signal Path which classifies the signal as “Enhanced” rather than “Lossless”.

   

[RJV]

I noticed the same in Roon. The sample rate conversion takes place in the Phantoms. 96 KHz to 48 KHz is NOT what Devialet promises in the Phantom I specs. I hope that they will be solving this ASAP with a firmware update. I have contacted Devialet about this issue and they admitted their shortcomings and apologized for it. Apologies are easily made - Solving this issue is another thing. I wonder why this could even be possible in the first place.

[supaDean]

Ask yourself... are you honestly hearing a difference in the sound?