Sound quality of PC and laptop headphone/line outputs

[captain paranoia]

What have I done wrong???

Wrong test frequency...? I note the time axis isn't labelled in any of the plots, and I can't see any mention of sweep period or test freq in the annotation. [edit: freq looks okay in the spectrum plot...]

You mention low output voltage; what voltage are you expecting from a headphone output? And is 500mV very low (compared with 'low' 1.15V and 1V)?

 

[captain paranoia]

It looks awfully like a 16 bit sine wave,

Yes. Looks like a 2-bit signal, which a -90dBfs would be for a 16-bit DAC. Still not sure why the freq appears different, though...

 

[captain paranoia]

Can you measure the digital stream

Well, that obviously wouldn't test the DAC, but it might show what their DSP is doing...

 

[svart-hvitt]

Well, that obviously wouldn't test the DAC, but it might show what their DSP is doing...

And noise, jitter, PSU related artefacts could be illuminated?

 

[March Audio]

Wrong test frequency...? I note the time axis isn't labelled in any of the plots, and I can't see any mention of sweep period or test freq in the annotation. [edit: freq looks okay in the spectrum plot...]

You mention low output voltage; what voltage are you expecting from a headphone output? And is 500mV very low (compared with 'low' 1.15V and 1V)?

Im not sure what you mean by "wrong test frequency". Its 1kHz same signal as all other tests I have made. The FFT may have given that away for you. Harmonics are as high as the fundamental! This may be an android or Roon problem. I subsequently tested a Samsung S7 and samsung S2 tablet with the exactly same result. Needs to be investigated further.

Low voltage in the context of using it as a line level source where you would expect 2v rms. Considering headphones can be 300ohms impedance or more, 0.5 volts is absolute minimum output voltage IMO.

 

[March Audio]

Yes. Looks like a 2-bit signal, which a -90dBfs would be for a 16-bit DAC. Still not sure why the freq appears different, though...

2 bit ???? Oh I see what you mean by this, ok. The time waveform is just a manual zoom so not the same in every graph. The frequency is 1kHz.

It would be interesting if Roon or Android is limiting the bit depth, which seems like a reasonable explanation. Obviously not an issue on the ipad.

 

[captain paranoia]

Im not sure what you mean by "wrong test frequency".

I was going by the time domain traces. The other devices you test show a period of five gradations. The one you're puzzling about is only two. Why the discrepancy?

Note my edit: I had spotted the 1kHz tone in the freq domain plot

The time waveform is just a manual zoom so not the same in every graph. The frequency is 1kHz

Right. I see.

 

[captain paranoia]

And noise, jitter, PSU related artefacts could be illuminated?

Only in the digital domain. But, as we know, a decent external DAC ought to be resistant to noise and jitter on the digital input, provided it's not so bad that it compromises data recovery.

 

[Frank Dernie]

Well the outputs are presumably there for headphones. Taking the output to a preamp input and through a hifi would be a less stressful test, presumably, because the input impedance and power requirement would be trivial to deal with. I have tried a few headphones and ear buds over the years when I was on an aeroplane every week. If there was any limitation it was power output, not overall sound quality being noticeably disappointing, but I was listening on an aeroplane...

 

[captain paranoia]

Harmonics are as high as the fundamental

Representing a sinewave with four levels (two bits) will give a lot of harmonic distortion...

It does look like something is truncating the digital stream to 16 bits, as Cosmik suggested.

 

[Blumlein 88]

Another data point. I mentioned it earlier in the thread. A Lenovo T410 (circa 2010) headphone out. It has enough oomph for most headphones. Volume typically not a problem.

The following are all 32K FFTs as is the forum convention. So the noise floor for the audio band would be some 42 db higher than what you see on the graph.

Here is 1 khz. Noise floor is the weak point of this laptop. Noise floor though not shown separately runs around -88 db. In the presence of strong signals it goes up 3 or 4 db.

Next is the 18+19 khz IMD test at max level.

Finally the 12 khz J-test signal.

 

[March Audio]

Representing a sinewave with four levels (two bits) will give a lot of harmonic distortion...

It does look like something is truncating the digital stream to 16 bits, as Cosmik suggested.

I agree and will be testing other playback software.

 

[March Audio]

Just found this (for general interest) whilst investigating the issue

 

[March Audio]

OK I have loaded Foobar onto the Huawei P10+ . This is 48kHz 16bit file 1kHz at -90.31dBFS

48kHz 24bit file 1kHz at -90.31dBFS

48kHz 24 bit file 1kHz at -102.34dBfs

48kHz 24 bit file white noise

Chord Mojo

So it looks like with both Roon and Foobar playing 24 bit files they are being truncated to 16bit and by the look of the frequency response possibly being re-sample to 44kHz. Im pretty sure thats not Roon or Foobar, so does that leave the Android OS?

I will check on a Samsung S7 and Samsung S2 to see if its the same, but it looks like it will be.

...mmmmmmmmmm.........

 

[March Audio]

Same files as above with Samsung S7

So obviously being truncated to 16 bit, no output at -102dB, but Im not convinced its being resampled.

 

[amirm]

So it looks like with both Roon and Foobar playing 24 bit files they are being truncated to 16bit and by the look of the frequency response possibly being re-sample to 44kHz. Im pretty sure thats not Roon or Foobar, so does that leave the Android OS?

Android will resample from what I have researched unless you use Plex player.

 

[March Audio]

Android will resample from what I have researched unless you use Plex player.

Brilliant info thanks I will test it out

 

[Blumlein 88]

Neutron works at higher rates and bit depths. If the device supports it and the internal DAC is capable. Or over USB into another DAC, or over various wireless protocols to other devices. 768 khz/32 bit support over OTG USB to other DACs.
http://neutronmp.com/

Poweramp player is confirmed on later Android devices to accomplish it.

Audio on Android is a minefield. Some devices can't be made to work at more than 16 bit, and without just the right software that apparently is normal for Android. Others can, but without attention from the user will not. Some have 24 bit DACs, but they aren't enabled for more than 16 bit. To my knowledge there is no single player software that you can count on to provide 24 bit output nor unresampled output above 44 or 48 khz.

 

[amirm]

Neutron works at higher rates and bit depths. If the device supports it and the internal DAC is capable. Or over USB into another DAC, or over various wireless protocols to other devices. 768 khz/32 bit support over OTG USB to other DACs.
http://neutronmp.com/

That's really cool. Went to install it and it wants permission for device ID (i.e. phone number). Why oh why????

 

[Blumlein 88]

That's really cool. Went to install it and it wants permission for device ID (i.e. phone number). Why oh why????

Are you doing the demo or the paid app? Maybe that is for your free trial period. Then again who knows. It didn't do this a year or so ago.