Simaudio's new 32 bit DAC

[Kellen]

Is 32 bits really needed or is this just a marketing ploy?

http://www.simaudio.com/moon750D.htm

[pfarrell]

Kellen wrote:
> Is 32 bits really needed or is this just a marketing ploy?

Needed for what?

Human's can not hear more than was a well recorded 16 bit signal contains.

Its good to use longer signals when mixing, doing effects, etc.
Its 99% marketing when they sell 24 bit recordings.

88.2 Hz sampling makes far more sense.

--
Pat Farrell
http://www.pfarrell.com/

[georgeh]

From their website:
...the MOON 750D is extremely flexible and exemplifies a luxurious lifestyle.

Seriously though, does anybody even publish 32 bit music?

[pfarrell]

georgeh wrote:
> Seriously though, does anybody even publish 32 bit music?

That can't be a serious question. If there is some audiofool who will
pay more for 32 bit or even 64 bit, then someone will sell it.

It may have only RedBook data on it, like many SACDs, but there is
nothing wrong with selling it.


--
Pat Farrell
http://www.pfarrell.com/

[michael123]

I think 32 bit is mostly used to lower noise floor.
Also, if transporter had 32bit input, I would send the output from Sox as 32-bit, to minimize rounding errors

RE: 16bit vs 24bit music: of course I hear the difference. I hear all the little nuances now, which were hidden by 16-bit precision. To my taste, I give preference having 44.1KHz/24bit than 88.2KHz/16bit

[Mnyb]

Marketing buzzwords , many dacs and processors use oversampling upsampling and all kind of mathematics to process audio, mostly to counteract filter artifacts and such problems move the noise spectra somewhere benign etc.
My meridian HT preamp uses 64bit floating point internally, but it still a "24 bit" machine. You must simply have processing space that is transparent to the incoming signal.

Btw there is no true 24bit dacs either how can it be an 32 bit, analog noise swamps the performance around 20bits for most practical systems, you need cryogenics to get to -144dB for 24bits. I read somewhere that the thermal noise in an 100 Ohm resistor is louder than that, the atoms is vibrating creating noise.

Btw nowdays with moores law and all, I'll expect even the most humble ht reciever to use dsp's with 64bit fpu. Maybe it was expensive 5-10 years ago.

But the kit looks very god anyway if you read the specs.
It seems to achieve *very* good data, so it looks like a *really* good DAC .
Just don't mind the marketing to much.
There must have been some serious engineering effort going on here. Not just some Audiophool sticking tubes into an existing design, I would not mind give this one a listen.

I copy pasted some data:

Configuration .................................................. ....... Fully balanced, dual-mono
Digital Power Supply Transformer ............................ 25VA
Analog Power Supply Transformer ............................ 10VA
Digital Power Supply Capacitance ............................ 17,600μF
Analog Power Supply Capacitance ........................... 8,800μF
Transport Mechanism ............................................. Proprietary Design
ESS ES9018S Sabre32 32-bit HyperstreamTM
Digital Filter / Digital-to-Analog Converters ...............
Frequency Response (audible) ................................. 20Hz - 20kHz +0/-0.1dB (with internal CD transport)
Frequency Response (full range) .............................. 2Hz - 100kHz +0/-3dB (with external digital source)
THD @1kHz, 0dBFS (A-weighted).............................. < 0.0003%
Intermodulation Distortion ...................................... < 0.0002%
Dynamic Range .................................................. .... > 120dB
Signal-to-noise Ratio .............................................. > 120dB @ full output
Slew Rate .................................................. ............ 50V/μs
Channel Separation ................................................ > 116dB
Low Level Linearity ................................................. < ±0.25dB at -90dBFS
Intrinsic Jitter .................................................. ....... 1 picosecond RMS
Analog Outputs – Balanced ..................................... 1 pair XLR
Max. Analog Output @ 0dBFS - XLR ....................... 2.0 Volts
Analog Output Impedance - XLR ........................... 100Ω
Analog Outputs – Single Ended ............................... 1 pair RCA
Analog Output Impedance - RCA ........................... 100Ω
Max. Analog Output @ 0dBFS - RCA ....................... 2.0 Volts
Digital Inputs (4)............................................... ...... AES/EBU (XLR), S/PDIF (RCA), TosLink, USB Type-B
Digital Outputs (2) ................................................. S/PDIF (RCA), AES/EBU (XLR)
Digital Input/ Output Impedance - S/PDIF ................ 75Ω (0.5 Volts p-p)
Digital Input/ Output Impedance - AES/EBU .............. 110Ω (3.7 Volts p-p)
Remote Control .................................................. .... All Aluminum Full-Function
Display Type .................................................. .......... 8 character dot matrix LED
Power Consumption @ idle ..................................... 25 Watts
AC Power Requirements ......................................... 120V / 60Hz or 240V / 50Hz
Shipping Weight .................................................. ... 35 lbs / 16 Kgs
Dimensions (W x H x D, inches) ............................... 18.75 x 4.0 x 16.81

[pfarrell]

michael123 wrote:
> I think 32 bit is mostly used to lower noise floor.

Huh? RedBook has a SNR of 96dB. That is already way below audible.
24 bit takes the SNR is 144 dB.

In absolute silence, 144 dB will cause ear damage. Not probably, will.

Or as OSHA says, maximum allowed exposure to 115 dB is about 30 second.
For every 3 dBs the permissible exposure time before possible damage can
occur is cut in half.

118 dB ~ 15 second
121 dB ~ 7 seconds
123 dB ~ 3.5 seconds
126 dB ~ 1.7 seconds

See http://en.wikipedia.org/wiki/Dynamic_range


--
Pat Farrell
http://www.pfarrell.com/

[Mnyb]

Is this the rms figures ? music peaks are very fast fortunately (milliseconds).
It's not a problem to see peaks ~115dB on a dance floor.

Here in Sweden i think we have an 85dB limit for 8 hour exposure ( a workday ) otherwise ear protection is mandatory. don't remember if it is a or c curve.

[Themis]

I think the word length used (32bit) has to do with the more recent chip used, NOT the music signal itself. It's used to lower artifacts, not SNR.

I doubt 16bit chips will still exist in a few years...

[michael123]

I think the word length used (32bit) has to do with the more recent chip used, NOT the music signal itself. It's used to lower artifacts, not SNR.

This is what I actually meant.
As with Transporter, where measurable resolution (e.g. output voltage) is merely 21bit, I do not think Simaudio will produce true 32bit on its outputs