What does 'measures identical' mean in practice? How close do two DACs need to match to be considered 'identically measuring' ?
So THD+N percentage within 1% of each other?
In practice then a DAC measuring 0.0001% THD+N is not identical to one measuring 0.000102%. Did I understand correctly?
In practice then a DAC measuring 0.0001% THD+N is not identical to one measuring 0.000102%. Did I understand correctly?
Based on my own experience of measuring THD+N with an AP, even at the 0.001% level the third significant figure is not at all stable (if indeed its even displayed on the monitor). So I doubt your definition is usable in practice with current measuring kit. It might be workable though with heavy averaging over a significant time period. But then I would guess individual (say) Topping D30s would turn out to be non-identical to each other.
Last edited:
Show me two dacs that measure identical. It just does not exist.
Wov, Abraxalito You were quicker!
Wov, Abraxalito You were quicker!
I have a QuantAsylum QA401 which I believe can be used as an ADC. And it should be pretty decent.
I will look into this. It is indeed interesting.
How did you verify the bit-perfectness? Did you also include your USB-I2S converter?
Bit perfectness has been verified by countless people over the years for these platforms.
I am referring to WASAPI output from Foobar on a Windows machine (with no DSPs or volume control applied obviously) and ALSA output from Linux machines, with no resampling or level change applied.
My USB to I2S converter is my own XMOS-based design, using XMOS' reference libraries, which have also been verified to be bit perfect by the industry.
Plus, if any of these platforms were not bit perfect, DoP would not work. And it does.
I am referring to WASAPI output from Foobar on a Windows machine (with no DSPs or volume control applied obviously) and ALSA output from Linux machines, with no resampling or level change applied.
My USB to I2S converter is my own XMOS-based design, using XMOS' reference libraries, which have also been verified to be bit perfect by the industry.
Plus, if any of these platforms were not bit perfect, DoP would not work. And it does.
The problem here is that in async UAC2 if the result of the USB-I2S conversion is bit-perfect the only way the host could have impact is through conducted or radiated noise which would somehow impair the timings or voltage references. While that may be possible it should also be visible in measurements (though not necessarily in THD+N).
Agreed. Noise, after all, influences jitter.
The problem is, THD+N may be relatively easy to measure, but it does not correlate directly to how "well" a device sounds. "Well" extends well beyond non-distorted and noisy. Clarity and sound stage play a huge role too.
The problem is, THD+N may be relatively easy to measure, but it does not correlate directly to how "well" a device sounds. "Well" extends well beyond non-distorted and noisy. Clarity and sound stage play a huge role too.
Not necessarily the only way. Conducted noise could be passed through the DAC into pre- and/or poweramp and affect those through IMD. It would depend on cabling (bal or SE) and also (if balanced) if industry best practices had been used in pin1 wiring.
That may well be another possibility. But regardless of how the noise from host impacts the process the outcome should most probably be measurable. And avoidable.
The measurement would need to be taken at the output of the poweramp to encompass all the possibilities. Measuring just at the DAC would miss the potential of the conducted noise's effect on the downstream components.
@Dimdim Not at all wishing to rain on your parade but I doubt the QA401 with its ~99dB SINAD is up for the task. You could always invest in a Cosmos ADC though, its noisefloor is over 20dB lower.
@Dimdim Not at all wishing to rain on your parade but I doubt the QA401 with its ~99dB SINAD is up for the task. You could always invest in a Cosmos ADC though, its noisefloor is over 20dB lower.
Last edited:
That's about what I was expecting from the QA401..
I'll look into the Cosmos unit. It looks interesting.
I'll look into the Cosmos unit. It looks interesting.
There are hardly any DIY kits available yet for new AK4499EXEQ it seems!? Let alone for 4 or 8 channels (with two DAC chips). Also no commercial multi channel option with this chip yet...
I would like to own a very good 6 or 8 channel DAC solution, driven via USB or ethernet. OktoDAC with ES9028 is best I found so far in the still affordable range (around 1500-2000 euro). Or Merging Anubis, a bit more expensive.
Currently have one AK4454 per speaker in Fusion 253 amps, already sounds good. Going to triple AK4493 PCB solution may also be an option, but perhaps this step is too small. Any better alternative available?
I would like to own a very good 6 or 8 channel DAC solution, driven via USB or ethernet. OktoDAC with ES9028 is best I found so far in the still affordable range (around 1500-2000 euro). Or Merging Anubis, a bit more expensive.
Currently have one AK4454 per speaker in Fusion 253 amps, already sounds good. Going to triple AK4493 PCB solution may also be an option, but perhaps this step is too small. Any better alternative available?