Bluesound releases 2024 ICON streamer

[GXAlan]

It is only a matter of time before other devices receive the new MQA technologies.

Not necessarily. It may require the ESS DAC for example.

Let me be my own judge on that, please.

By the way: what is it doing ?
And how to know it's active or not ?

It’s a unique digital filter which is always on. They claim that strict brick wall filtering adds pre ringing (think Denon), and the slow roll off allows imaging artifacts to come in (think Marantz) or even a NOS DAC has zero ringing, but a lot of imaging artifacts.

QRONO is its own filter that is leakier than a brick wall, but not so leaky as a NOS, and is optimized differently depending on the sample rate since 44.1 kHz sample rate has a much higher risk of imaging artifacts while high res content does not.

The white paper doesn’t really show sufficient detail, despite what @pogo likes to suggest (a friendly jest there) because all of these filters matter the most, if at all, with 44.1 kHz content and they didn’t show that.

Just like Meridian popularized the apodizing filter, MQA Labs is trying to popularize the QRONO filter.

Importantly, they suggest that the full scale high frequency tests that show imaging artifacts is so rare but that the ringing is common, and thus, they are saying that they will accept worse measurements for THD with wide bandwidth testing because for real music, the actual artifacts are low.

In other words, perfect brick wall = lots of ringing, but no imaging artifacts above nyquist. NOS dac has zero ringing but the maximum number of imaging artifacts. Everything is somewhere in the middle, and QRONO believes their middle is better than everyone else’s middle so confidently, it’s not user selectable.

 

[pjug]

Let's hope we can disable it.
I don't want to be forced to get rid of all my Nodes and PowerNodes.

Apparently you can't disable it. But you can get around it if you upsample externally. With Roon, for example. I don't like the idea of MQA and Qrono but honestly I think it will sound fine. Just as upsampling with a sharp linear phase filter will also sound fine.

 

[Galliardist]

Not necessarily. It may require the ESS DAC for example.


It’s a unique digital filter which is always on. They claim that strict brick wall filtering adds pre ringing (think Denon), and the slow roll off allows imaging artifacts to come in (think Marantz) or even a NOS DAC has zero ringing, but a lot of imaging artifacts.

QRONO is its own filter that is leakier than a brick wall, but not so leaky as a NOS, and is optimized differently depending on the sample rate since 44.1 kHz sample rate has a much higher risk of imaging artifacts while high res content does not.

The white paper doesn’t really show sufficient detail, despite what @pogo likes to suggest (a friendly jest there) because all of these filters matter the most, if at all, with 44.1 kHz content and they didn’t show that.

Just like Meridian popularized the apodizing filter, MQA Labs is trying to popularize the QRONO filter.

Importantly, they suggest that the full scale high frequency tests that show imaging artifacts is so rare but that the ringing is common, and thus, they are saying that they will accept worse measurements for THD with wide bandwidth testing because for real music, the actual artifacts are low.

In other words, perfect brick wall = lots of ringing, but no imaging artifacts above nyquist. NOS dac has zero ringing but the maximum number of imaging artifacts. Everything is somewhere in the middle, and QRONO believes their middle is better than everyone else’s middle so confidently, it’s not user selectable.

The white paper, which I regard as a bit of a joke to be honest - I commented somewhere on it - states that QRONO d2a is a filter set. The filters described in the paper are not guaranteed to be representative of what you get. Remember that MQA also had/has a filter set: it was just that 99.9% of MQA files used just the one. @mansr posted on that, IIRC. You may be right: the white paper looked like a poorly edited version of a description of old MQA in parts. And remember, their "new information" came from 2010.

I note that the update refers to QRONO as a codec. This may be the QRONO dsd bit, but also may be an indication that other processing is going on than just a filter. The white paper refers quite specifically to the MQA triangle where high frequency content was cut in MQA - remember a couple of GoldenSound's tests that didn't work because he hadn't realised that? - so, is QRONO also throwing away that information? Without proper reverse engineering of the codec, we don't actually know what is going on.

That's why I'm recommending caution here with buying a QRONO enabled device - it's possible that there may be a surprise when we do know what it's doing. Amd it's one thing to sell a device and advertise it with QRONO, and another to update older devices with a supposedly very different playback method, without it being defeatable on those older devices - or at least optional. It may be impossible to avoid that firmware update - I don't know as I don't own any of the devices concerned.

I know I was annoyed when our Samsung TV got a new half-working interface dumped on it, and that didn't change its actual performance as a display.

 

[Rja4000]

The white paper, which I regard as a bit of a joke to be honest - I commented somewhere on it - states that QRONO d2a is a filter set. The filters described in the paper are not guaranteed to be representative of what you get. Remember that MQA also had/has a filter set: it was just that 99.9% of MQA files used just the one. @mansr posted on that, IIRC. You may be right: the white paper looked like a poorly edited version of a description of old MQA in parts. And remember, their "new information" came from 2010.

I note that the update refers to QRONO as a codec. This may be the QRONO dsd bit, but also may be an indication that other processing is going on than just a filter. The white paper refers quite specifically to the MQA triangle where high frequency content was cut in MQA - remember a couple of GoldenSound's tests that didn't work because he hadn't realised that? - so, is QRONO also throwing away that information? Without proper reverse engineering of the codec, we don't actually know what is going on.

That's why I'm recommending caution here with buying a QRONO enabled device - it's possible that there may be a surprise when we do know what it's doing. Amd it's one thing to sell a device and advertise it with QRONO, and another to update older devices with a supposedly very different playback method, without it being defeatable on those older devices - or at least optional. It may be impossible to avoid that firmware update - I don't know as I don't own any of the devices concerned.

I know I was annoyed when our Samsung TV got a new half-working interface dumped on it, and that didn't change its actual performance as a display.

Any idea which measurement would help understanding what it's doing?

Frequency response seems obvious.
The normal filter evaluation at different sampling rates.

Any other ?

 

[pogo]

See Technical White Paper.

 

[GXAlan]

Any idea which measurement would help understanding what it's doing?

Frequency response seems obvious.
The normal filter evaluation at different sampling rates.

Any other ?

See Technical White Paper.

The test is easy. You take a regular DAC and look at impulse response for different filters at different sample rates and compare them to QRONO.

If you had an old and new firmware version of the older nodes, that would even let you do some recordings of music with high frequency content and throw them into deltawave to compare.

We know QRONO switches based upon sample rate. It may also switch like Denon does with Alpha processing to go NOS when it detects the impulse test or traditional.

 

[Rja4000]

View attachment 425017
View attachment 425018

The test is easy. You take a regular DAC and look at impulse response for different filters at different sample rates and compare them to QRONO.

View attachment 425019

If you had an old and new firmware version of the older nodes, that would even let you do some recordings of music with high frequency content and throw them into deltawave to compare.

We know QRONO switches based upon sample rate. It may also switch like Denon does with Alpha processing to go NOS when it detects the impulse test or traditional.

OK
I usually measure those, but only at 44.1 kHz.
I didn't do it yet here. I'll do it at different sampling frequencies (which makes sense anyway. They are not the only ones to claim to use a less aggressive filter at higher sampling rates. Which makes total sense anyway.)

And I'll add a timing comparison.

 

[delamelon]

it would have been wise to align the end of the promo code on Dirac with Amir's test... I'm saying that, I'm saying nothing

 

[Rja4000]

it would have been wise to align the end of the promo code on Dirac with Amir's test... I'm saying that, I'm saying nothing

What do you mean?

(Node N132... Expect a few dBs more from the Icon)

 

[Kolllak]

(Node N132... Expect a few dBs more from the Icon

108 dB via the official data-sheet.

My measurement for Icon is below. It's done with Focusrite ADC, so not precise.
Anyway is rather close to the specs.

 

[pogo]

XLR or RCA?
I don't know which Focusrite you have used exactly, but it's probably a distortion mix of Icon and Focusrite.
Do you have Dirac Live?

 

[Galliardist]

Any idea which measurement would help understanding what it's doing?

Frequency response seems obvious.
The normal filter evaluation at different sampling rates.

Any other ?

Because QRONO is claimed so be a filter set, rather than a single filter, it could be detecting some test signals and using a sharper filter than in other circumstances. I’m not sure how to test for that.


Because of the mention of the triangle, a test with high frequency content is in order, as is a multitone test. If an 0dB signal at 19kHz is properly output, we can safely dump the white paper on that alone because it would show that the triangle is not implemented.

The presence of QRONO as a codec strongly suggests additional signal processing though, so it may need more than standard tests to understand what it is actually doing. The codec may only be for DSD processing but that needs to be examined separately anyway.

 

[Rja4000]

108 dB via the official data-sheet.

View attachment 425451

Correct.
The Icon uses twice as many DACs and on XLR outputs, we can expect twice the voltage.
So a NOISE improvement between 2dB and <6dB is to be expected.
3-4dB noise improvement sounds reasonable.
But that will translate in SINAD only IF (and that's a big "IF") they get rid of the Harmonic distortion - which will not improve with that.

My measurement for Icon is below. It's done with Focusrite ADC, so not precise.
Anyway is rather close to the specs.

View attachment 425450

Well, TD+N at -97.9dB, that's not really close.
Is that the XLR output ?

The guilty is probably the ADC.
The easiest way to improve this is to use Cross-correlation averaging, which will help reducing the ADC influence.

For that, you would need to cable one of the Icon outputs to several (2) ADC inputs.

I don't know if @pkane 's excellent Multitone now includes Cross-correlation...
If not, you may want to download REW and use its Cross-correlation setting.
With some care and patience (300 averages minimum) you may retrieve those 10dB missing here.

One good news is that THD is low. Much better than with the Node.

But still, you may easily test first the issue with Tone control I explained above:
Just set the tone control on and dial a -0.5dB bass correction and re-measure.
Even with the ADC impact, we should see it.

 

[Kolllak]

XLR or RCA?
I don't know which Focusrite you have used exactly, but it's probably a distortion mix of Icon and Focusrite.
Do you have Dirac Live?

XLR
Scarlett 2i2 3th gen, THD via Amir is below.
I agree with you that my ADC can be worse than the Icon.
But I believe at least Icon's SINAD is not lower than I got.

Scarlett 2i2 3th

Yes, I use Dirac. It was the main reason to buy the new Bluesound.
My measurements above are with Dirac off.

 

[pogo]

Please perform this quick test:

I would be interested to know whether post-processing is carried out at 192kHz as with my M33. You would simply have to activate Dirac Live and perform a frequency response measurement up to 100kHz.

 

[Rja4000]

My measurements above are with Dirac off.

What's the input level on the ADC ?
It seems -10dBFS or so, which is quite low.
Could you increase it to reach -1dBFS or so ?

 

[Rja4000]

perform a frequency response measurement up to 100kHz.

With a 192kHz ADC ?

 

[Mnyb]

The white paper, which I regard as a bit of a joke to be honest - I commented somewhere on it - states that QRONO d2a is a filter set. The filters described in the paper are not guaranteed to be representative of what you get. Remember that MQA also had/has a filter set: it was just that 99.9% of MQA files used just the one. @mansr posted on that, IIRC. You may be right: the white paper looked like a poorly edited version of a description of old MQA in parts. And remember, their "new information" came from 2010.

I note that the update refers to QRONO as a codec. This may be the QRONO dsd bit, but also may be an indication that other processing is going on than just a filter. The white paper refers quite specifically to the MQA triangle where high frequency content was cut in MQA - remember a couple of GoldenSound's tests that didn't work because he hadn't realised that? - so, is QRONO also throwing away that information? Without proper reverse engineering of the codec, we don't actually know what is going on.

That's why I'm recommending caution here with buying a QRONO enabled device - it's possible that there may be a surprise when we do know what it's doing. Amd it's one thing to sell a device and advertise it with QRONO, and another to update older devices with a supposedly very different playback method, without it being defeatable on those older devices - or at least optional. It may be impossible to avoid that firmware update - I don't know as I don't own any of the devices concerned.

I know I was annoyed when our Samsung TV got a new half-working interface dumped on it, and that didn't change its actual performance as a display.

Except that the ringing does not happen with music, only with the test signals used to characterise filters ? So what are they suggesting that this solves ? Are they solving a non problem and replacing it with a real problem?

Edit : I think I replied to the wrong post

 

[pogo]

With a 192kHz ADC ?

You can also use a test signal up to 60kHz for analog in, optical in and USB in to answer my initial question of post-processing.

 

[Rja4000]

You can also use a test signal up to 60kHz for analog in, optical in and USB in to answer my initial question of post-processing.

Well, you can, for sure, see if the bandwidth is limited to 48kHz.