New 28-bit DAC coming out.

[boXem]

It may be an issue of Amir's measurements, but again - if I've understood correctly, in the review I linked for the ARTHUR 2408/N2 it was measured at 115/116.

The 2408/N2 was discontinued 2 years ago. Amir also measured the Arthur 4222/E1 at 125 dB, it is indeed the "noisiest" amp we sell today. But we are off topic.

 

[antcollinet]

The 2408/N2 was discontinued 2 years ago. Amir also measured the Arthur 4222/E1 at 125 dB, it is indeed the "noisiest" amp we sell today. But we are off topic.

Fair enough (I didn't deliberately choose an old review by the way - when I search for Boxem or Arthur in the review index, the 2408 is the only one that comes up)

 

[rdytro]

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.

But what if I want to get a nice system for my cats

 

[signalpath]

Hi All. Thanks for the interesting discussion. The D-1 multi-path DAC is designed for the pro market. In recording, we stick microphones directly on snare drums and trumpets. These have around +155dB SPL peak level at the microphone. On the other end, the human ear can detect sound down to -8dB SPL. In recording, this is a real-world dynamic range of around 165dB, or around 27-bits.

Will everyone need 165dB dynamic range? No, of course not. We're creating a new professional standard, not a home standard. Most home listeners are fine with 100-110dB peaks, or even less with headphone applications (though significantly more if a large sub-woofer is used for explosions, hip hop, etc). The D-1 DAC is just the beginning. We seek to re-create every link in the audio signal path via multi-path architecture, including microphones and power amplifiers.

Ultimately, it's all about THD+N at low music levels -- where we are most sensitive to atmospherics and imaging. And this is a key advantage of multi-path. When we cross-fade to the top of the low-path (roughly -45dBFS), the low-path DAC IC begins converting at its most significant bit (bit-32) --- dramatically reducing low-level THD+N compared with legacy single-path architecture at the same level. This is immediately audible, and easily testable on an AP.

It's looking like production D-1 will achieve 28-bit performance, which is 40nVrms quiescent noise (broadband, unweighted) and +23dBu headroom, with zero ISOs. Output impedance is roughly 1.8 ohms. This performance is only achieved at the diff-bal XLR outputs. The RCA outputs are not specified. An AP cannot directly measure -146dBu broadband noise. Heroic measurement techniques are required to even get close to such a number. See Art Kay’s seminal papers on measuring ultra-low-noise.

The best multi-path analogy we've found is called HDR = High Dynamic Range Photography. HDR uses multiple exposures which are then intelligently combined to create a single image with much greater dynamic depth and detail than can be achieved with a single photo (multi-path v. single-path). Please read the AES paper (AES21106) for discussion on noise calcs, path continuity, linearity, etc..

 

[unpluggged]

These have around +155dB SPL peak level at the microphone

How then they don't clip the capsule?

We're creating a new professional standard

What a bold statement.

This is immediately audible

An even bolder statement.

 

[popej]

Ultimately, it's all about THD+N at low music levels

My guess is, that your product has higher amount of THD at high level. Care to provide measured or estimated SINAD, the way it is used on ASR?

 

[signalpath]

How then they don't clip the capsule?

Many pro microphones are designed to handle +155dB SPL, some even higher.

What a bold statement.


An even bolder statement.

Indeed. We're moving systemic dynamic range from 120dB to 160dB.

 

[signalpath]

My guess is, that your product has higher amount of THD at high level. Care to provide measured or estimated SINAD, the way it is used on ASR?

The high-path is on par with any very high quality DAC today, with THD in the -120dB range. It's in the low-path where we greatly improve today's single-path architecture.

 

[Salt]

So next generation of AP has to cover the beneath 160 dB path.
Or @IVX may go for it.
?

 

[Scytales]

In recording, we stick microphones directly on snare drums and trumpets. These have around +155dB SPL peak level at the microphone.

I know that is off-topic, but it should be strongly pointed out that the practice described in bold types is one way to record some sounds, but it is not the only one. Faithful recordings to live performances are routinely made with microphones at some distance from the instruments, where the sound pressure levels is much less than directly at the instruments.

 

[popej]

It's in the low-path where we greatly improve today's single-path architecture.

How could you improve THD above quality of DAC in low path? You can attenuate noise, but not the ratio between distortion and signal.

 

[antcollinet]

It's looking like production D-1 will achieve 28-bit performance, which is 40nVrms quiescent noise

Correct me if I'm wrong - if the LSB is around 40nVrms, then Vfs is around 40nVx2^28 = aprox 11V (or is that the 23dBu = 15.5V?)

If that is genuinely hit, kudos on the massive technical acheivement, but...

These have around +155dB SPL peak level at the microphone

that is nothing like 155dB at the listening position - even if that 155dB was carried into the final distributed recording.

On the other end, the human ear can detect sound down to -8dB SPL

Possibly - in dead silence. But how many ears can detect -8dB in a typical room with around 30dB of noise?

Then, "The dynamic range of music as normally perceived in a concert hall does not exceed 80 dB"

Further, the dynamic range of recorded music can be from as low as 5dB for horribly compressed junk, up to maybe a max of around 40dB for some classical.

The 96dB of dynamic range of 16 bit redbook will allow for 96dB peaks, at around 80dB average SPL, and still leave the noise floor down at 0dB - 30dB below typical in room noise. With shaped dither, that could be improved to at least 112dB peaks, and 95dB plus averages. Personally I've never knowingly heard 16 bit quantisation noise at any listening level.

Clearly 24bit DACs with maybe 20 to 22 bits actually achieved and a 120dB+ dynamic range stomp all over the requirements of home audio reproduction.

I can see that very high dynamic range ADCs can be useful in recording applications to avoid the need for level setting. I guess I can assume that 28bit DACs might be useful in the recording studio - although I admit to not really understanding how. But I think it is clear that there is no application in home audio reproduction that this technology will ever be needed for.

Convince me I'm wrong

 

[IVX]

Super cool! It went from vaporware to likely shipping this year!

This is what it is doing.
View attachment 400435

Cirrus DACs 431** working this similar way 5 years+, and some of cheap ESS DACs as well, but fortunately, they have registers to On/Off the DRE(Dynamic Range Enhancer) feature. Cirrus didn't mention that trick, in fact, sort of a noise gate, but tests clearly show artifacts of such an approach. BTW, I don't remember if someone considered CS43198/43131 as a serious high-end DAC, DR 130db(A), and 5mA current draw - cute! Unfortunately, physics requires low impedance, hence, high currents to handle true-high DR. As an example, ES9039Pro is literally inapplicable without a heatsink, and shows just 137db(A) of the DR in stereo mode, and THD+N <-130db@1kHz 0dbfs.

 

[IVX]

that is nothing like 155dB at the listening position - even if that 155dB was carried into the final distributed recording.

why do we need to talk about a theoretical mic's DR if the topic is about the DAC? Let's talk about a theoretical DR for a speaker system, including ambient noise levels and we'll find that the DR/SNR itself isn't really an interesting part of the specs if lays under -120-130db(A). Distortions, maybe, but not DR/SNR anymore for sure. An ADC with the lowest noise is still interesting, and the mentioned mic's DR makes sense(slightly)

 

[AnalogSteph]

Cirrus DACs 431** working this similar way 5 years+, and some of cheap ESS DACs as well, but fortunately, they have registers to On/Off the DRE(Dynamic Range Enhancer) feature. Cirrus didn't mention that trick, in fact, sort of a noise gate, but tests clearly show artifacts of such an approach.

Don't forget AKM's headphone codecs (AK4377/A, AK4376/A, and AK4375A before that, going back almost 10 years now). They have always been more upfront about the trickery though, and ironically their parts pass some tests that trip up the Cirrus parts with flying colors.

I have my suspicions about ALC1220 as well.

BTW, I don't remember if someone considered CS43198/43131 as a serious high-end DAC, DR 130db(A), and 5mA current draw - cute! Unfortunately, physics requires low impedance, hence, high currents to handle true-high DR.

This is quite true. Fortunately, human hearing doesn't actually require instantaneous DR anywhere near that high (but rather maybe 110 dB and change tops, and ~15 dB less in headphones), we mainly just never want to hear noise even from the most sensitive of IEMs. And honestly, CS43198 when switched to Class AB (which does away with the artifacts potentially encountered in Class H) still makes a pretty impressive DAC.

 

[solderdude]

I know that is off-topic, but it should be strongly pointed out that the practice described in bold types is one way to record some sounds, but it is not the only one. Faithful recordings to live performances are routinely made with microphones at some distance from the instruments, where the sound pressure levels is much less than directly at the instruments.

Yep, but it is much easier to do close miking as you do not have to mind the acoustic place the recording is made in, especially when multiple instruments are recorded at the same time and one wants to balance instruments afterwards.

For ADC such a huge dynamic range is needed. In a studio that range is squashed anyway in order to produce a recording that is listenable at lower average SPL.

For DACs I can only see it beneficial in research and signal generation for testing such ADC's and not having to use attenuators to break up the measurement range.

Possibly - in dead silence. But how many ears can detect -8dB in a typical room with around 30dB of noise?

While basically true one has to include the noise spectrum. If the overall spectrum of the 30dB in noise is mainly in the LF area (most likely) and in the 'high pitched' hiss band of electronics then one can still hear sounds well below 30dB noise floor in the 1-6kHz band for sure where the average noise levels in that band are usually low.
You may need a set of young ears and some acclimating to silence time.
Late at night usually works when you can hear a clock tick where you don't hear it during the day.

Otherwise I agree with the sentiment.

Also detecting continuous or tone bursts at a certain frequency is something entirely different as hearing dynamic sounds in the entire audible spectrum (music)

I found that when playing music comfortable loud at the listening position some 70dB attenuation is enough to hear 'silence' even when arguably there still is sound but below our hearing threshold.

So yes, in engineering/research we may want to have 160dB dynamic range but for audio reproduction about 80dB is indeed more than sufficient. Make it 100dB for S/N ratio and audio reproduction is fine.
I also wonder how many mic pre-amps have 160dB S/N ratio. When doing close mic recordings the actual dynamic range is not 160dB but much smaller what concerns musical signals (the trumpet sounds that end up in the recording)

We don't need 160dB range for home audio reproduction for sure and that does not seem to be the target market either but audiophools will jump at it anyway.
Also I assume that range can only be achieved with +23dBu (11V) in balanced mode (home audio 'standard' = 4V)
When the input circuit of the following device has 9dB headroom that might still work if only the following circuits could maintain such a S/N ratio.

I applaud the technical achievements made here. I also assume some audiophiles and audiophools will use these DACs in home audio too so there certainly will be an additional market outside of recording/research and applications other than music recording.

 

[IVX]

AnalogSteph, right, Impressive like engineered well compromising. I bought Meizu HiFi DAC for $13, for that cost it is impressive even with a fake DR.

 

[IVX]

For DACs I can only see it beneficial in research and signal generation for testing such ADC's and not having to use attenuators to break up the measurement range.

I think it is a bad idea due to the obvious source of a high-order nonlinearity. Look, the top DAC is 7 bits(I hope it is just a typo, I see no reason to cut 25 bits of AK4490 or similar DAC), if it is so, try to imagine how will it work with -35dbfs sine. The top DAC should be muted 4 times per period, did you ever see such a commutation with no glitches? It should work smoother with two DACs both 32 bits, then the muting needs to be Off after the signal goes over the threshold for 10-50ms(common psychoacoustic time constant), and after 100-300ms(time hysteresis) mute should be On if the signal fades under the threshold. This way work Cirrus and ESS(only cheap portable models) DACs but they never go for 42db overlap, rather 6-10db to keep that trick surely inaudible. Glitches and short clippings are there but they are not periodical and nearly inaudible, especially for the mass-market portable dongles.
BTW, we can try such dual DAC combo with the single part $2 ADAU1701 + low-noise opamp, to get 150-170db(A) of the "DR"

 

[solderdude]

The top DAC should be muted 4 times per period, did you ever see such a commutation with no glitches?

Didn't the discussed DAC pull this off or intends to do so given:

It's looking like production D-1 will achieve 28-bit performance,

That would mean no glitches that would show up in measurements ?

BTW, we can try such dual DAC combo with the single part $2 ADAU1701 + low-noise opamp, to get 150-170db(A) of the "DR"

I won't stop you for sure.
Would be great to get that kind of performance but output voltage would have to be in the 10-20V range... because of physics.
20V AC (balanced) means at least +/- 15V supply rails is needed for the output circuits.

 

[IVX]

because of physics.
20V AC (balanced) means at least +/- 15V supply rails is needed for the output circuits.

that's why I mentioned the low-noise opamp