Decoder or render.setting phantom in audirvana and roon

[struts]

I have no experience with Audivarna but I use Roon with an Expert Pro. As far as MQA goes, the situation with the Expert Pro range and the Phantoms is the same: Devialet provide no support at all for MQA. No Devialet product can decode or render an MQA stream.

In the device settings in Roon for your Phantom you should find 2 settings. The first, in the "Audio Device" setting is "MQA capabilities" which provides the options of No MQA support, Decoder and Renderer, Decode Only, Renderer Only. The second setting is in the Advanced Settings which follow the Audio Device settings in the menu. In the Advanced Settings you will find an option for "Enable MQA Core Decoder" with a simple Yes/No option.

The way MQA works with Devialet products is like this:

If a device, in this case your Phantoms or my 140 Pro, is sent an unprocessed MQA stream, it will play a core part of that stream which it sees pretty much as a standard CD quality audio stream. It neither decodes or renders the MQA data in the stream, it simply ignores it.

Some software apps including Roon can decode the MQA stream Before sending it to our respective devices. Struts has said Audivarna can also decode the stream and I'll accept that advice. The second setting in Roon which I mentioned above, "Enable MQA Core Decoder" controls whether or not Roon decodes the stream and there will be a similar setting in Audivarna. If you set that setting to "Yes" the MQA stream will be decoded and instead of a 44.1 kHz or 48 kHz stream, your Phantoms will be sent an 88.2 kHz or 96 kHz stream. That's the "first unfold" in MQA terms but the stream will also include some MQA data for the second and third unfolds which have to be done by a hardware device and can't be done by Roon or Audivarna. Your Phantoms and my 140 Pro ignore that extra MQA data because they don't have the requisite chip to process it but they will play the 88.2/96 kHz stream.

It really doesn't matter what you set the "MQA capabilities" setting to. Devialet products have no MQA capabilities so you may as well set it to none but regardless of what you set it to, Devialet Products can't use the MQA component of the stream and will ignore it. You should hear no difference whatever you set "MQA Capabilities" to because Devialet products do nothing with the MQA data part of the stream, they simply ignore it. All that setting does is to control whether or not Roon strips the MQA code from the stream. I don't know if Audivarna has such a setting.

The only setting which will make a difference for you is the "Enable MQA Core Decoder" setting in Roon or whatever that setting's equivalent is in Audivarna. You should hear a difference between having Roon/Audivarna perform the core decoding step or not. That's the only thing which will make a difference for you. Whether it will sound better or worse to you is something I can't answer because some people like MQA and some don't and the only way for you to find out whether you'll like it or not is to try it.

My recommendation, and what I do, is to set "MQA capabilities" to "No MQA support" and the "Enable MQA Core Decoder" setting to "Yes" in Roon and to use whatever the equivalent settings in Audivarna are. That way you'll get the benefit of the first unfold being done in Roon/Audivarna.

If you decide you don't like MQA, my advice is to select a CD quality or higher non-MQA encoded version of the song/album if that is available from your streaming service. There is an element of lossy encoding involved with MQA which to my ears makes an undecided MQA stream sound slightly worse than a CD quality non MQA encoded stream.

Thanks @David A, clear, correct and comprehensive as usual (I suspect you were a tech pub author or teacher in another life  ).

The relevant settings in Audirvana are found under "Audio Settings".  There are two:
1. "Auto-detect MQA devices" (toggle on/off).  While the purpose here is clear, the actual effect is not, and I can't find any documentation.  Presumably, if set to "off" the actual behaviour of Audirvana is determined by setting 2. and if set to "on" presumably the actual behaviour is determined by whatever capabilities the DAC presents and the optimum setting chosen? 
2.  "DAC not detected as MQA, use as" (not MQA/MQA renderer, MQA decoder).  According to my understanding there are only two alternatives here, either Audirvana performs software decoding or it doesn't.  So why the three options?  Using the DAC as a decoder means that Audirvana should not perform software decoding, and using it as a renderer means it should, but what about "not MQA"?  In this case does Audirvana perform software decoding and send 24/88.2 or 24/96 or not and send the 24/44.1 or 24/48 original file?  Curious minds want to know!

Unless someone here knows I'll post the question on the Audirvana support community.

Great questions @eladeb!

Edit: Actually it occurs to me I can test this pretty easily myself with my Dragonfly, when I have a moment.  Stay tuned.

OK, here it comes.  I tested playing the MQA (24/192) version of Norah Jones Come Away with Me from Tidal on both a Dragonfly Red (MQA-capable) and a Meridian Explorer (not MQA-capable).  I tried all three settings for "DAC not detected as MQA, use as" with the "Auto-detect MQA devices" switch both on and off (the results in both cases were the same for both DACs so what this switch does is a bit of a mystery.  Whatever it does it does not seem to influence whether the decoder in Audirvana is engaged or not).  
The results for the Explorer were:

• not MQA: Audirvana reports sending 24/96, confirmed by Explorer
  
• MQA renderer: Audirvana sends 24/96, confirmed by Explorer
  
• MQA decoder: Audirvana sends 24/48, confirmed by Explorer
  

The results with the Dragonfly were:

• not MQA: Audirvana reports sending 24/96, Dragonfly lights Magenta indicating 96KHz
  
• MQA renderer: Audirvana sends 24/96, Dragonfly lights purple indicating "MQA"
  
• MQA decoder: Audirvana sends 24/48, Dragonfly lights blue indicating 48KHz
  

So, it seems that my suspicions were correct inasmuch as Audirvana performs core decoding (first unfold) regardless of whether the "use as" switch is set to "no MQA" or "MQA renderer" (but not if it is set to "MQA decoder").  Presumably then "MQA renderer" signals to an MQA renderer to perform the second unfold, whereas "not MQA" doesn't (and this presumably only in manual mode whereas the decision is taken automatically in "Auto-detect" mode?).  Unfortunately since the Dragonfly is limited to 24/96 and the result is the same regardless this is impossible for me to prove.

What I can conclude however is that Audirvana performs the first unfold if the receiving DAC is capable of processing the higher bitrate, regardless of whether it is MQA-capable or not, and that the only way of stopping it is to set "use as" to decoder.  So the purpose of two switches and three settings is still a bit of a mystery to me.