Out of phase due to oversampling

[Bow_Wazoo]

I'm wondering what effect the Neutron Player's internal oversampling has on the output signal. So I connected my smartphone to the SMSL DL100 as usual. The DL100 is connected to the line in my sound card via pre out.
I played a 44.1Khz white noise and recorded it twice.
1x without oversampling, and 1x with 8x OS. When analyzing both recordings, I noticed that the oversampled recording is more out of phase. Is that relevant in any way, or is it a disadvantage for the sound?

In the non-oversampled recording phase, it makes no difference which filter I set on the DL100. Filters 1 and 4, for example, look identical. I haven't tried 5.

 

[solderdude]

The ears don't use phase at such high frequencies.
The difference in roll-off is inconsequential for most adults as well.

 

[Sokel]

The ears don't use phase at such high frequencies.
The difference in roll-off is inconsequential for most adults as well.

There's no more high frequency content at the 8X sample than the native 44.1kHz as far as I can see.it's white noise after all.
Noisy maybe?

 

[KSTR]

White Noise is usually not band-limited and therefore it must and will contain ringing from the reconstruction filter used in the updampling.
This normal and totally unavoidable and of course inaudible... except when an very suboptimal reconstruction filter were used, one with minimal ringing and accordingly a very rolled-off high frequency behavior.

 

[KSTR]

There's no more high frequency content at the 8X sample than the native 44.1kHz as far as I can see.it's white noise after all.
Noisy maybe?

The upsampled signal cannot contain more HF than the orginal, assumming a proper filter being used.
But the result must and will contain dithering noise when the signal is converted back to integer domain (as any reasonable upsampler works with floating point) and that dithering may have significant noise shaping.

 

[Sokel]

The upsampled signal cannot contain more HF than the orginal, assumming a proper filter being used.
But the result must and will contain dithering noise when the signal is converted back to integer domain (as any reasonable upsampler works with floating point) and that dithering may have significant noise shaping.

Of course but won't this filtered out by the DAC (to a decent degree at least) ?
Chart shows even less content for the 8X result before the 22.050Hz mark.

Unless the app computes the out of band noise too of course.Couldn't this be defined or set by the app?

 

[KSTR]

Chart shows even less content for the 8X result before the 22.050Hz mark.

Exactly. This drop is coming from the upsampler's reconstruction filter as expected.

Unless the app computes the out of band noise too of course.Couldn't this be defined or set by the app?

I don't know as I'm not using this program but all in all it looks quite unprofessional.
EDIT: for example, why does it show 44.1k/16 for the second file? It should be 352k/24 (assuming the upsample spits out 24bit, which is what it should do).
Does it resample the second file?

 

[Bow_Wazoo]

So this is caused by the different filters? That surprises me, because filter 1 of the SMSL (red) and that of the Neutron Player on 8x OS (green) are similar as far as I understand?!

 

[Bow_Wazoo]

EDIT: for example, why does it show 44.1k/16 for the second file? It should be 352k/24 (assuming the upsample spits out 24bit, which is what it should do).
Does it resample the second file?

Neutron feeds the DAC with 352Khz, but the recording with Audacity was done in 16bit/44.1kHz.

 

[KSTR]

This is analog loopback? Why?
You should look at the output of the original and resampled source file directly

 

[Bow_Wazoo]

I was interested in comparing the output signal of my DAC/(Pre-)Amp, player oversampling, vs. no oversampling.

 

[KSTR]

I was interested in comparing the output signal of my DAC/(Pre-)Amp, player oversampling, vs. no oversampling.

Ah OK... but that is in conflict with your initial statement:

I'm wondering what effect the Neutron Player's internal oversampling has on the output signal.

Anyway, if you want to look at the analog output you should ideally use a higher sample rate than either of the playbacks, so 705.6kHz, and 24 bit.
Practically, the sample rate should at least be 2x higher than that of the original source file, so I'd use >= 176.4kHz, 24bit.