Bluesound NODE2 to connect to Phantom

[iliapas]

Hi, Gremlin.

Just so we're on the same page, the "resolution" in audio is defined by the bit-depth of the audio stream.
And yes, you are right, that is the same as dynamic range. Some people refer to the pair of sampling rate and bit-depth as "resolution", which is why try to avoid the term altogether.

So, regarding the dynamic range: regardless of the volume you are listening at, dynamic range is defined not in absolute terms, but in relative. Moreover, before being amplified, the signal being output from your DAC is hardly audible, but still retains all the dynamic range that has been reconstructed from the stream by the DAC.
And the 24bit signal at peak level will have exactly 8bit more dynamic range than 16bit signal at same level.

So when you reduce the level on the source equipment, you are in fact applying negative gain to the signal, and zeroing out the extra information.
So when you have source signal of 24bit, at -48dB volume, applied at source, what gets input to the Phantom DAC is in fact only 16bit of actual data left, and 8bits of zeroes at every sample. So the DAC reconstructs the signal with lower dynamic range, which is then amplified.

If you cannot discern the difference between 24bit and 16bit, that's fine, and mny people would argue that 16bit is more than enough for anything.
But the fact is that bit-depth=dynamic range is reduced by applying volume control at source. You can probably hear the effect by listening at very low volume to a 16bit source - you will probably hear hiss and noise because the effective dynamic range left intact is very little.

[iliapas]

You can play down to -48 dB and loose no information from a 16-bit CD.
Right?

Hi, Pleyel.
This is wrong, with improbable exceptions.
There is no way to tell the DAC to reduce volume (unless you . DAC doesn't do volume control, the amplifier does.

If you are not changing the amplification in the amplifier, then you want the output from DAC to be lower level.
Signal from DAC is always output at same absolute levels. So if you want reduced levels you need to start zeroing some of the bits.

If your source upconverts your 16bit stream into 24bit and then applies the gain, you may really be ok until -48dB, but most consumer equipment does not upconvert audio signals, and if it does, upconversion and gain is not done with high precision, thus adding noise to the signal. Most equipment when given 16bit signal will just forward that as-is. A 24-bit DAC will accept a 16bit stream fine, and reconstruct it with appropriate dynamic range for 16bit. If only 10bits of those 16bits are used, the dynamic range resulting will be that of 10bit signal. The amplifier will then amplify whatever signal the DAC outputs, and, as a side note - most amplifiers respond much worse (higher distortion) close to their maximum than at lower levels. So you lose fidelity by reducing bit-depth or gaining noise during upconversion/neg-gain in the source and possibly lose fidelity again by driving the amplifier at it's maximum, amplifying maximally the reduced noise floor.

Just as a reference, read this paper by ESS regarding digital volume control, but note that the solutions mentioned are not applicable if you reduce bit-depth at source, as the data path between the source and the DAC is limited by the optical interface, through which you HAVE to send the zeroes, to attain the volume-control effect. So the DAC never sees the original data, just zeroes, so there's nothing that can be done to repair the signal.

http://www.esstech.com/files/3014/4095/4...ontrol.pdf
TLDR: First couple of slides (1-9) is what happens when you apply volume control at source.

[Pleyel]

You can play down to -48 dB and loose no information from a 16-bit CD.
Right?

Hi, Pleyel.
This is wrong, with improbable exceptions.
There is no way to tell the DAC to reduce volume (unless you . DAC doesn't do volume control, the amplifier does.

If you are not changing the amplification in the amplifier, then you want the output from DAC to be lower level.
Signal from DAC is always output at same absolute levels. So if you want reduced levels you need to start zeroing some of the bits.

If your source upconverts your 16bit stream into 24bit and then applies the gain, you may really be ok until -48dB, but most consumer equipment does not upconvert audio signals, and if it does, upconversion and gain is not done with high precision, thus adding noise to the signal. Most equipment when given 16bit signal will just forward that as-is. A 24-bit DAC will accept a 16bit stream fine, and reconstruct it with appropriate dynamic range for 16bit. If only 10bits of those 16bits are used, the dynamic range resulting will be that of 10bit signal. The amplifier will then amplify whatever signal the DAC outputs, and, as a side note - most amplifiers respond much worse (higher distortion) close to their maximum than at lower levels. So you lose fidelity by reducing bit-depth or gaining noise during upconversion/neg-gain in the source and possibly lose fidelity again by driving the amplifier at it's maximum, amplifying maximally the reduced noise floor.

Just as a reference, read this paper by ESS regarding digital volume control, but note that the solutions mentioned are not applicable if you reduce bit-depth at source, as the data path between the source and the DAC is limited by the optical interface, through which you HAVE to send the zeroes, to attain the volume-control effect. So the DAC never sees the original data, just zeroes, so there's nothing that can be done to repair the signal.

http://www.esstech.com/files/3014/4095/4...ontrol.pdf
TLDR: First couple of slides (1-9) is what happens when you apply volume control at source.

Hi iliapa,
I respectfully, yet fully disagree, but will need more time to explain a little later.
Difference between resolution and quantization noise will need to be addressed.
I really confirm that lower resolution streams are LSB-padded to operate higher resolution DAC at full scale (wherever in between). This extremely simple operation is even not called upsampling, but part of the interfacing. Non power-of-2 gains are another story, as pointed, but do not impact resolution before -48 dB: they add in-band noise if no dithering is applied. The paper does not address the case of 16-bit music into 24-bit DAC, which specifically provisions 8 bits for attenuation.
More to come when I can.

PS: On another subject, h_u_g_e kudos for your Phantom protocol deciphering.

[iliapas]

Hi, Pleyel.
Indeed, maybe the explanation glossed over important parts and oversimplified some others.
So, to add some more detail, some base facts:

Most DACs accept 32bit input easily. So if you upconvert your 16bit or 24bit stream into 32bit, you have at least 8-16bits for volume control.
S/PDIF is limited to 24bit samples, so that's the max your DAC will get.
Most power-amps add a flat 36dB to the signal fed into them (set by the internal feedback ratio). On some, you can change the resistor value in the feedback loop to attentuate this.
Most of the time you attenuate in the preamp and then amplify in the power-amp.

Phantom has a different structure, applying complex DSP to the signal before playback, and controlling the internal power-amp directly. This means that the more accurate signal you provide it, the more accurate the results of it's internal math calculations. The maximal bit-depth you can provide it over S/PDIF is 24bit. If you attenuate beforehand, even with proper dither, you are still reducing the bit-depth, which may affect the accuracy of the resulting math calculations and audio fidelity. Playing a 24bit source stream, there is no way around losing fidelity, as the Phantom DSP never sees the original bits. Playing 16bit source stream on a proper upconverting and dithering/noise-shaping playback device , you indeed have the 8bits for volume control and there is no harm done, as no information is lost.

Regarding protocols, still seeking participants with a couple Phantoms + Dialog that are willing to run Wireshark and capture some of the communications between them. Please PM if you can help.

[Johnnydev]

Ok, so it doenst rely on bluetooth. Did not look at it because it is a little to basic with only volume control.

It is only volume yes, but if you have him, you don't won't to miss him anymore, you don't need then a NODE 2 or a Sonos etc. to satisfy your wife.

And you have the best sound quality with only using the Spark app. And that is a lot more joy then using several apps.

[MrSinister]

Ok, so it doenst rely on bluetooth. Did not look at it because it is a little to basic with only volume control.

It is only volume yes, but if you have him, you don't won't to miss him anymore, you don't need then a NODE 2 or a Sonos etc. to satisfy your wife.

And you have the best sound quality with only using the Spark app. And that is a lot more joy then using several apps.

Well I already have the node and thats ok for my wife. You can use an ir remote with the node and also use presets for the radio. Sound quality is not an issue for her.
Perhaps I will buy the devialet remote for my own use.

[martinskeem]

Assuming same source material at <= 24 bit, why would sound quality be better in Spark? Personally, I feel the UX of Spark is inferior to other (older) competitors, so I use an alternative.


Sent from my iPhone using Tapatalk

[MountainGuy]

So for those of you with the Node 2 (i.e. Johnnydev, MrSinister), are you saying that the Spark results in BETTER audio quality? How large is the gap?

Thanks!

[Johnnydev]

Ok, so it doenst rely on bluetooth. Did not look at it because it is a little to basic with only volume control.

So for those of you with the Node 2 (i.e. Johnnydev, MrSinister), are you saying that the Spark results in BETTER audio quality?  How large is the gap?  

Thanks!

No i don't have a node2 , i think the best sound is with spark and not with all kind of other things!!

[jvi]

I agree with Johnnydev's conclusion. On the other hand, keep on trying, you'll never know...