XXHighEnd

Reading the new release of an upgraded NOS/OS Dac today which has this capability, I wondered about XX/Phasure running to this figure. Has it been tried?

Robert

Hi Robert,

It would be relatively easy to implement. However, that can only happen when I can test it. And that requires a physical (DAC) example.

Which DAC is that ?

Regards,
Peter

Hi Peter,
            https://www.denafrips.com/single-post/2019/08/02/Product-Launch-ARES-II

I wasn't thinking of buying one but as its setup to take 1536khz, I wondered if you had tried this at some point in the past.

Robert

FWIW, I find 352.8/384 the sweet spot with the NOS1.

Mani.

Robert,

*If* you buy one, buy it through me; In that case I can make it work first (this is by guarantee, unless they lie about the specs).

Notice that I see something fishy in there, because each upsampling step requires 1 additional bit. So, 16 bits is the base (for Redbook). Then we go 2x, 4x, 8, 16 (that would be 705.6) and 32. This is 5 additional steps, with thus 16 bits at the base. Where to we end at ? 21 bits.
But the DAC is 20 bits only …

Having said *that*, I now see it fail in advance. I mean, XXHighEnd will be able to do the upsampling, but it will output in 24 or 32 bits, 21 bits really utilized, while the DAC will accept 20 bits only and thus messes up with the upsampling. It will still do 1411.2 Khz (or 1536 Khz), however.

And additionally : thinking of that, a 24 bits file would also be subject to the same anomaly, for a 24 bit DAC. But, there's a difference for the level where this occurs. Thus, the 24 bit is 24 dB lower than the 20bit, which is a lot - and actually too much to be audible. For 20 bits I am not so sure, especially because a DAC could resolve to 21 bits (the NOS1 does just over 23 bits, which is an exception).
Thus, it is also related to the resolvement, and once in the noise, it should not matter any more (as in : when upsampling from 24 bits requires a 25 bit but resolvement is to 23 bits (or less) anyway, then it can't matter).

The THD+N of 0.004% is an other figure which makes my "upsampling" theory quite moot. This is a bit harder to reason out (with math) but since the 0.00059% I see pass by over here is at a noise level of ~-140dB and distortion at -120dB, that headroom of 20dB implies just over 3 bits. With the notice that this is different from "resolvement", implying error in that same area can't be the best (thus, NOS1 resolves to 23 bits, but THD+N implies less than 21 bits - I am sure you can't follow).
Now, 0.004% will come down to 3 bits worse again (compared to mentioned 0.00059), thus "less than 21" turns into "less than 18".
Conclusion (kind of) : The 20 bit DAC already shows distortion at less than 18 bits, and thus implying more distortion beyond bit 20, is harmless.

Go for it.
Not that you need by any means. I just had fun in this bit of reasoning; I hope I didn't make too many mistakes.

Peter

hi-fi plus, August 2019