New 28-bit DAC coming out.

[popej]

This is what it is doing.

This looks so wrong ....

Is there overlap between high and low dac? -28dB in lower path suggest, that high path adds only less than 5 bits.
How DR 140dB at lower path can be reduced by adding high path signal, if it is already defined versus FSB? Or it isn't?
What about linearity? Do we get SINAD defined by 7-bit DAC without dither? Or high DAC true precision is 27bit and it perfectly matches low DAC? If it is the later, then why not make a single 27bit DAC?

 

[Chrispy]

Wouldn't it then be AD-DA? If it is just DAC, I am afraid it is just for audiophiles and SINAD chasers.
I could be wrong.

Assumed it had that capability, for consumers only, meh.

 

[johny_2000]

people listening to it with speakers that compress as early as 100dB

I can't find any loudspeakers or headphones that can deliver even 100dB of DNR. Everything I see is in the 60 to 90dB range before noise or distortion becomes noticeable, and most mainstream products fall in the 60 to 70 dB range.

 

[johny_2000]

How many recordings actually have a S/N ration >120dB anyway ?

Most of the content I see on streaming services these days is 16-bit (96 dB DNR). Very little of it is released at Hi-Rez 24-bit (144 dB DNR), and I'm not sure if it was recorded at that resolution in the past, or just remastered now to 24-bit.

 

[solderdude]

The vaccine is called ASR but it does not appear to work on everyone or maybe is not applied correctly

 

[boXem]

@MC_RME putting aside the "smart" level detection, what is the advantage of this method compared to what RME does with the auto ref level that seems to be a bit simpler ?

 

[Martini]

Stellar dynamic range (if real), is almost meaningless with the recording industry compressing the living daylights out all the music.

 

[KSTR]

This looks so wrong ....

Is there overlap between high and low dac? -28dB in lower path suggest, that high path adds only less than 5 bits.
How DR 140dB at lower path can be reduced by adding high path signal, if it is already defined versus FSB? Or it isn't?
What about linearity? Do we get SINAD defined by 7-bit DAC without dither? Or high DAC true precision is 27bit and it perfectly matches low DAC? If it is the later, then why not make a single 27bit DAC?

The inventor is a member of this forum and has already explained how it works in details, so please read up.

 

[KSTR]

@MC_RME putting aside the "smart" level detection, what is the advantage of this method compared to what RME does with the auto ref level that seems to be a bit simpler ?

It happens instantly on the fly in real-time according to momentary signal levels and, more importantly, the switchover is fully transparent (clickless, etc). However, similar to multipath ADCs, there is an unavoidable noise floor modulation when the high level path is muted/unmuted, see the encircled section of the above diagram

In other words, there is an increase in resolution, effective when the signal level is low, but there is no increase in signal to noise ratio when the signal level is high as the high level path dominates the noise and low level bits just get drowned out as usual.

And as noted, the main idea is to do away with any range switching of signal chains that is normally required (and at the burden of the user) to get the best performance at any given signal level. All gain/level setting can be done in the digital domain without compromises (well, the compromise is that the signal must not contain significant DC or ultra low frequencies that would keep the level going down below the transition threshold).

 

[antcollinet]

It's not impossible because a clever real-time range switching is done (actually, a cross-fade). Once the signal level is low enough, you switch to a heavily padded down output and re-adjust the digital input.

With ADC it's easier and it's been done for many years, if not decades. Any better guitar multi-effect pedal does this, these days.

The benefit of it is that you don't need to care about signal level ranges, Connect and forget. You can use one ADC and one DAC unit to cater for any signal from uV/mV range to several ten's of volts. For AC signal signals only, of course.

But surely the DAC output electronics still needs to deliver an overall SNR on the output of 160+dB.

You can't range switch the output buffer - it has to be able to output a signal at say 2V, AND at (2/2^28)V with the noise level being lower than the smaller of those two.

Is that even physically possible? What is the minimum thermal noise on a 100ohm resistor?

EDIT - found a thermal noise calculator - if the output impedance is 100ohm the thermal noise on that resistor is (on its own) -135dBV, so about -140dB compared with 2V. I find it it difficult to believe a full output stage could be constructed without significantly higher noise than that.

Sort of renders 28bit conversion redundant - unless I am seriously misunderstanding what is possible.

 

[SSS]

28-bit DAC is meaningless for home audio. Dynamic range of 24-bit 96k is more than enough to my experience and opinion. But there may be special cases where the higher dynamic will make sense. Perhaps in a big theatre where very high SPL is used and then it should be quiet without noise?

 

[boXem]

It happens instantly on the fly in real-time according to momentary signal levels and, more importantly, the switchover is fully transparent (clickless, etc).

That's indeed what I was calling 'smart'.
If one removes the real time part and the adc feedback for clickless switching (that's at least how I understand it), what is the advantage of the multipath over a single path with various analog gain/attenuations and associated digital gain?

 

[CapMan]

Sooo... if the maximum SPL peaks my ears can shortly tolerate = 125dB SPL and a 0dB sound pressure can only be detected when having acclimated in an acoustically dead room why would one need a DAC that can do 162dB dynamic range ?
What amp can match that ?
How many recordings actually have a S/N ratio >120dB anyway ?

It's just a numbers game and has no benefit to audio.

I'm gonna get my ears syringed just in case
So I'm ready

 

[popej]

The inventor is a member of this forum and has already explained how it works in details, so please read up.

Thanks. I'm not sure what explanation you are referring to, but I have read this article:

Imersiv D1 Multi-Path Digital-to-Analog Converter Completes 32-bit Audio Chain

Unveiled at the 2023 AES Conference in New York, the new Imersiv D1 Digital-to-Analog Converter is the first product from a new brand created by Millennia founder John La Grou, and the first to introduce to the market his approach to multi-path technology. The results of years of research and...

audioxpress.com

I get confirmation, that picture is really wrong and doesn't correspond to the article text. In summary: low path is attenuated by 34dB while high path is amplified by 8dB, which together make 7 bit for high DAC. Each DAC noise is estimated at -106dB. Total dynamic range should be 148dB, which is about 25bit.

This quote confirms my doubts: "Finally, one might consider the 7-bit high-path DAC to be a relatively simple affair, but it also must be accurate to the least significant bit value of the complete system." This is where we need SINAD measurement to validate implementation. Pity, no mention of THD in article.

 

[KSTR]

Thanks. I'm not sure what explanation you are referring to, but I have read this article:

Imersiv D1 Multi-Path Digital-to-Analog Converter Completes 32-bit Audio Chain

Unveiled at the 2023 AES Conference in New York, the new Imersiv D1 Digital-to-Analog Converter is the first product from a new brand created by Millennia founder John La Grou, and the first to introduce to the market his approach to multi-path technology. The results of years of research and...

audioxpress.com

I get confirmation, that picture is really wrong and doesn't correspond to the article text. In summary: low path is attenuated by 34dB while high path is amplified by 8dB, which together make 7 bit for high DAC. Each DAC noise is estimated at -106dB. Total dynamic range should be 148dB, which is about 25bit.

This quote confirms my doubts: "Finally, one might consider the 7-bit high-path DAC to be a relatively simple affair, but it also must be accurate to the least significant bit value of the complete system." This is where we need SINAD measurement to validate implementation. Pity, no mention of THD in article.

But surely the DAC output electronics still needs to deliver an overall SNR on the output of 160+dB.

You can't range switch the output buffer - it has to be able to output a signal at say 2V, AND at (2/2^28)V with the noise level being lower than the smaller of those two.

Is that even physically possible? What is the minimum thermal noise on a 100ohm resistor?

EDIT - found a thermal noise calculator - if the output impedance is 100ohm the thermal noise on that resistor is (on its own) -135dBV, so about -140dB compared with 2V. I find it it difficult to believe a full output stage could be constructed without significantly higher noise than that.

Sort of renders 28bit conversion redundant - unless I am seriously misunderstanding what is possible.

The inventor, going by screen name @signalpath, has posted some details in this thread: https://www.audiosciencereview.com/...theoretical-minimum-noise-levels.29004/page-3 (starting at post #51), and there might be more info in his other contributions.
He might add some comments here...

 

[popej]

The inventor, going by screen name @signalpath, has posted some details in this thread: https://www.audiosciencereview.com/...theoretical-minimum-noise-levels.29004/page-3 (starting at post #51)

There is still only general info. I wonder it there exist a product, that can be measured. Or if there is at least preliminary specification.

I think the split of samples to 7+19 bits leads to difficulties. While it all looks nice in digital domain, I doubt if the split signals obey the Nyquist condition. Getting perfect sum on analog side seems to be problematic. IMO there would be less problems if both DAC worked at full resolution. Then the final analog adder would be replaced with a switch or mixer, which basically is the same concept as adder, but with mute on both paths.

 

[antcollinet]

I've decided I don't care.

Even if the DAC can achieve 28 bits and SNR of 160dB. And even if the output stage can do justice to that:

No human ears on the planet can come close.
No Headphone on the planet can come close.
No speaker on the planet can come close.
No microphone on the planet can come close.
No room on the planet can come close without killing item 1 on this list.

I'm an engineer. Good enough is all that is needed. Any more is a waste of resources.

 

[popej]

No human ears on the planet can come close.

Yes, that's the part of the concept
From product description: "It is known from psycho-acoustic research that broadband noise can be inaudible when it is below a certain program level due to masking."

 

[sergeauckland]

I've decided I don't care.

Even if the DAC can achieve 28 bits and SNR of 160dB. And even if the output stage can do justice to that:

No human ears on the planet can come close.
No Headphone on the planet can come close.
No speaker on the planet can come close.
No microphone on the planet can come close.
No room on the planet can come close without killing item 1 on this list.

I'm an engineer. Good enough is all that is needed. Any more is a waste of resources.

Totally agree. Good enough for transparency plus a little in hand is all that's ever needed.

Just because one can, doesn't mean we should. That's what has got the world into the mess it's in.

Many years ago I was taught that Quality means Good Enough.

S

 

[AnalogSteph]

I imagine the military could use something like this. Though then there's the next problem, finding a bunch of power amps on the same level...