MSB's 28.5bit dac

For $90,000.

Wonder if it's measuring as claimed. Jason(Schiit) mention that no dac can exceed 22bits.

http://www.digitalaudioreview.net/2015/11/the-select-dac-ii-msb-technologys-90000-da-converter/

 

How many bits deep is the most dynamic recording? I think Jason'll eat his hat if we get past a certain number.

 

This chart on their page is the most intriguing. Did they measure that 28.5 bits in a supercomputer simulation or something?

 

What is the Richter Scale equivalent of 173 Db?

 

LOL. Show me the ADC used for measurement that has a -200dB noise floor. Hint: there isn't one. Because physics.

Seriously, check the measurement capability (noise floor, resolution, distortion) of the best systems out there (Audio Precision, Stanford, dScope,) and then decide for yourself what is credible.

 

For a sustained sound, the limit is about 194dB because maximum loudness is dependent on the density of the transport medium. Air has an average pressure of one atmosphere, so the largest deviation is +/-1 atmosphere. Overpressure beyond one atmosphere becomes a shock wave and that is where the Richter scale comes into play.

 

For reference, even if the measuring device was calibrated for 0dBFS = 24dBU (rather than the consumer 6dBV), a -200dB noise floor would equate to ~1nV of noise (RMS). I'd guess this is beyond the performance of any ultra-low-noise, temperature controlled, astronomical data acquisition electronics, and might very well be physically impossible with electronics of this century. Forgot how to run those numbers ATM. For reference, most audio analyzers are in the single-digital uV range in terms of noise performance, and that is considered TOTL.

I'd give MSB the benefit of the doubt and ask them to clarify their measuring devices and methods. That plot smells like it was put together by a sales guy.

 

-200dB noise floor is partly an FFT trick, but that plot still is not believable at all.

They describe their setup here.

 

Once you get past -130dB, shouldn't you be worrying about timbre and envelope? (Correct answer: you should worry about it before -130)

MolaMola's DAC is on some crazy overkill shit and yet they don't claim to be -200. They say:

Signal to Noise Ratio: 140 dB (standalone version), 130dB (preamp option board)
THD, IMD: not measurable (estimated -150 dB).​

http://mola-mola.nl/index.php/dak

Being -200dB would presume that you have a device with a noise floor below -200dB, as other have said.

 

The DAC output is run through a 22-bit ADC and somehow that turns into a 28-bit noise floor via this FFT SINAD thing? Someone wanna explain how that one works?

I guess they only claim "effective bits" (whatever that means) but it still reeks badly of pseudo-science.

 

When you amplify the output signal of the DUT (DAC Under Test) by 70 dB and that amplifier and following ADC have 10dB to spare you can 'measure' that deep.
That ADC can be a 'lowly' - 130dB one in that case.
All one has to do is move the scale down 70dB (so -130dB now says -200dB) and you have these numbers.
Of course the maximum input signal of the 'test system' will overload when large signals are introduced.
The -90dB will almost be a FSD signal going into the ADC.


Now to find an amplifier and a recording that has a noise floor that can keep up and find some speakers (and ears) that can handle that dynamic range.

Marketting BS ... who doesn't ?

 

200dB??? Isn't the max differential at standard atmospheric pressure something like 194dB because you can't go more negative than an actual vacuum?

 

Yes.

 

Not a matter of real SPL or acoustic waves. We're talking about the noise floor inherent to electronics components, fundamentally caused by non-zero temperature (and hence, random motion of electrons within components) as well as other noise causing mechanisms in semiconductors.

 

Maybe they are measuring that stuff in my classroom where the laws of physics are broken everyday!
Force of Gravity points up! Acceleration due to gravity has been measured from 1.08 m/s^2 to 1124 m/s^2!

 

That's easy peasy with a properly constructed free body diagram and assigned reference frame/axes. Although next week's class will combine friction and pulleys on an inclined plane. Students might cry at that one.

 

What do physics matter? For $90,000 I sure as hell expect specs that are -literally- out of this world.

 

Hello, the clock itslef Femto 33 from MSB is nearly 20 grands, snake oil?

http://www.msbtech.com/products/femto.php?Page=dacSelect

 

Yup. At room temp, assuming 30ohm output resistance of measured dac, signal at 1kHz has Johnson's noise of about 10 nV below which any signal is unmeasureable.
So yeah, 200dB noise floor at 1kHz and up better be referenced to some high voltage to make it valid.
Reference to 1V -200dB is 100 pV, hmm, do we have dacs that output 100+ volts ?