Recommendation: Putting together a new desktop audio system

[o7_brother]

For example he didn't EQ Stax SR-404 below 50 Hz but according to this distortions measurements he can do it. Why? Maybe I missed something.

Because that kind of Stax headphones does not seal well on a measurement rig, but may seal better on an actual fleshy human head.

See the two graphs he made for the 404 Limited. First one is just the headphones sitting on his measurement rig, the second one is the same but with more pressure applied so they would seal better.

https://www.dropbox.com/s/gtqg9oykowxogrx/Stax SR-404LE.pdf?dl=0

https://www.dropbox.com/s/aaw4zzia8efznjo/Stax SR-404LE (with good seal).pdf?dl=0

Therefore it makes sense to EQ the bass by ear to some degree.

 

[bobbooo]

any thoughts on the Sennheiser 6xx and the HiFiMan Sundara? THose appear to be the top contenders (today).

The HD6XX is sonically the same as the HD650, just with a different paint job. In terms of distortion, it has more THD in the bass, and lacks sub-bass, so EQing that up might result in audible distortion, which would not be the case for the Sundara, which has both more sub-bass and less distortion there. However, the Sundara does have a bit of a distortion spike at ~6 kHz though, as well as a not entirely clean impulse response (the latter probably not audibly so though):

The HE4XX has no large distortion spikes however, and a cleaner impulse response tail:

 

[outerspace]

as well as a not entirely clean impulse response (the latter probably not audibly so though):

Impulse response view in case of headphones is just combination of frequency response and phase response. Non-linearities are too small to see on this graphs. You can emulate real IR by applying raw FR with minimum phase equalizer. So they are useless. CSD are useless too. And step responses.
Real measured (red):

Emulated:

Another example:

zoomed:

 

[bobbooo]

Impulse response view in case of headphones is just combination of frequency response and phase response. Non-linearities are too small to see on this graphs. You can emulate real IR by applying raw FR with minimum phase equalizer. So they are useless. CSD are useless too. And step responses.
Real measured (red):
View attachment 84530
Emulated:
View attachment 84533
Another example:
View attachment 84520View attachment 84519
zoomed:
View attachment 84521

Yep, I know. People on this thread wanted 'good impulse response', so that's what I showed them. I said it won't be audible.

 

[outerspace]

I said it won't be audible.

They are most probably audible because this "ringing" is different representation of frequency response peaks.

 

[bobbooo]

They are most probably audible because this "ringing" is different representation of frequency response peaks.

I was talking about the difference between the tail of the Sundara and HE4XX's impulse response.

 

[outerspace]

I was talking about the difference between the tail of the Sundara and HE4XX's impulse response.

This "tail" is time domain representation of 10 dB ~13 kHz peak.

 

[bobbooo]

This "tail" is time domain representation of ~13000 Hz peak.

Exactly, and the difference in the Sundara and HE4XX's impulse response tails and so level of the frequency response at 13 kHz is not that big, in a range where our hearing is less sensitive (even beyond the audible range of some older listeners), hence 'probably not audibly' different in this respect.