New 28-bit DAC coming out.

[pablolie]

... The fact they are announcing/showing it at AES instead of launching it at an audiophile show suggests that it's really backed by science....

Hmmmm...

 

[egellings]

Assuming a 2V FS, then the LSB is about 7.5nV (0.0075 microvolts)

Thermal noise on that 10K resistor mentioned up there ^ is around 1.8uV at room temperature. Some 240 times bigger.

It's like getting a drop forge when all you asked for was a tack hammer. I do have to admire the engineering making such performance possible, though.

 

[EJ3]

Hmmmm...

Since we now know who is behind it, I'll give it a strong (leaning to the "YES! side) MAYBE (but still a wee bit skeptical).

 

[pablolie]

So science indicates audio benefits from 28 bits? Aren't there already 32/768 DACs out there anyhow?

 

[Bleib]

So science indicates audio benefits from 28 bits? Aren't there already 32/768 DACs out there anyhow?

I see it as pure marketing. Yes, they exist but real world performance is closer to 22-23 bit for the best stuff. More than enough.

 

[AdamG]

Will we need new Bionic Ears to take advantage of this? Maybe a new Audiophile upgrade path to obsess over….

 

[MaxBuck]

Since eight bits = one dollar, I'm assuming this thing will be available for $3.50 US. Seems like a pretty good deal.

 

[pablolie]

I'd think these days you'd go for the 8 bit increment to gain notorierity. 28 seems chosen for differentiation rather than audio science, sorry. why not 64 bits for hell's sake while one is at it? the tech is there.

 

[Human Bass]

I wonder if people are misinterpreting the original post. Maybe they mean a 28 bit R2R dac?

Schiit has a 64bit dac, that ofc is technically inferior to delta sigma ones

 

[unpluggged]

Schiit has a 64bit dac

By this measure, the Topping D30pro is a 128-bit DAC then

 

[antcollinet]

I do have to admire the engineering making such performance possible, though.

So do I - if it is actually achieved at the DAC output.

 

[KSTR]

The first thing that signal is going to hit is the input impedance of the pre-amp or amp - typically 10K and up. The thermal noise alone coming from that input impedance (let alone everything else) is going to be around 40dB (100x) bigger than the output noise of the DAC.

No. The effective noise impedance is the parallel combination of input and output impedances, thus dominated by the latter.

 

[solderdude]

Assuming a 2V FS, then the LSB is about 7.5nV (0.0075 microvolts)

Thermal noise on that 10K resistor mentioned up there ^ is around 1.8uV at room temperature. Some 240 times bigger.

+28dBu output is 19.5V that's +19.8dB opposite 2V for the S/N ratio when assuming lowest possible noise.
source:

Yes. A 27-bit DAC should be demo'ed at the 2023 AES Conference, Oct 25. That's a noise floor of -145dBu (broadband, unweighted) and a maximum output of +22dBu. A 28-bit option will be available for professional applications (+28dBu max output). Shipping 2024. New company is called "imersiv"

 

[antcollinet]

No. The effective noise impedance is the parallel combination of input and output impedances, thus dominated by the latter.

Good point - so the output impedance of the DAC (including PCB trace, wire and connector) needs to be of the order of that 0.17 ohms mentioned by @AnalogSteph above. And all the resistors in the output stage need to be of a similar order or lower. This might be possible, If it is I'm happy to admit it is beyond me.

And the noise level of the preamp and amp (or whatever the next device in the chain is) need to be down at -160dB also.

 

[antcollinet]

Why is that clueless people

Nice. I'd have preferred a discussion where any errors of fact are corrected, and we can move on with increased understanding. Now you've just tripped my hair trigger ignore button.

 

[boXem]

I'd have preferred a discussion where any errors of fact are corrected

That's indeed what I did, didn't seem very efficient.
Reading this may prove enlightening, in the case of a non inverting configuration R4 is the input impedance.

 

[Tell]

I just don't see why anyone should ever care about this DAC? 168dB of dynamic range is as useful as a car going 0-100 in 0.1 sec, it'll hurt SO much and possibly even kill you.

 

[antcollinet]

That's indeed what I did, didn't seem very efficient.
Reading this may prove enlightening, in the case of a non inverting configuration R4 is the input impedance.

Fair enough, so you did. And I apologise. The specific statement of mine regarding input resistor noise was factually incorrect, and I steadfastly ignored your hint to that effect.

However the intent of my argument (magnitude of noise on the output of a DAC achieving 28 bit performance compared with the noise performance of subsequent devices in audio applications), is I think still valid.

Even your amps, which achieve stellar performance in the world of audio amps, have full power SNR some 60dB short of the dynamic range needing to be achieved by a genuine 28 bit DAC.

 

[boXem]

Why is that clueless people always question the performance of an expert product they have no clue about?
This just boosts the bad reputation that ASR already has even more.

To be fair, the first post of the thread links to marketing material that is, ihmo, not at the level of the invention. Better reading the audioxpress article, and eventually the patents, to have a better understanding of the system.
If you take the wrong reasoning "higher input impedance = higher noise", a designer from stellar measuring electronics has been repeating this as a mantra for years because his electronics use an el cheapo (single inverting opamp) input stage. I see no reason why other members would not repeat this.
Finally, you seem to have a good understanding of the needs of the pro world. Maybe explaining why a record DR is more than engineering excellence for the pros could stop the comments from home users?

Even your amps, which achieve stellar performance in the world of audio amps, have full power SNR some 60dB short of the dynamic range needing to be achieved by a genuine 28 bit DAC.

DR and SNR are different metrics.
Full power SNR from my amps is around 130-135 dB

 

[antcollinet]

DR and SNR are different metrics.

But (I think - and I'll bow to your greater knowledge if wrong ) closely related.

Full power SNR from my amps is around 130-135 dB

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.

I did look on your website product details for your published figures but in the 3 or 4 amps I checked the SNR was stated as TBD. For example:

But even at 135 dB, still short by 40dB of 28 bits