Frequency measurements with different inputs

[f1eng]

Is there any chance the microphone was in a slightly different location or an adjacent piece of furniture was moved between measurements?

I can see no reason why differences between the various digital connections would cause what you are showing, but moving the measuring microphone, or a reflecting surface near it, a few cm would do exactly what your measurements show (except the optical result ???).

All your plots look like the errors are down to room reflections, as if the room is a very sparsely furnished modern style room (which are hopeless for good sound quality, unfortunately) or reflections from a hard surface near the measuring microphone.

Good questions!

All the measurements above were taken within a 5 minutes period, with absolutely no changes in microphone position (or anything else for that matter), and even I myself was standing in the same position, well outside of the "magic triangle" of both loudspeakers and the microphone.

I did not have to move since I was controlling the computer with a wireless mouse-like device.

No appliance was switched on or off, and only the refrigerator was on during the whole period. The only lights on were "old style" lightbulbs - the noisier led or fluorescent lighting was switched off for the test. No air conditioning (we're doing the test in Oslo!)

Immediately after performing the Toslink test I repeated the wifi and Ethernet again (due to my surprise) and the results were equivalent to the initial ones.

Wow! maybe that is why my Astell & Kern portable into the optical input sounds so good!
The comb filter treble is still typical of interference patterns from reflections.
It is conventional to average room measurements at a number of microphone locations around the listening position to cancel these out, which is a pita.
Does the sound change noticeably if you move about?

[Jwg1749]

I hate to say this, as it sounds critical of the original measurements, and it's not meant to be critical at all, but it would be all the more interesting if these results could be repeated, i.e. the same measurements done over a period of time.

Again: I'm not doubting the validity of the measurements, it's just that if the results are repeated on several occasions, then we have a sounder empirical base for what looks like a very interesting phenomenon.

Matt

[Divasson]

Don't fear doubting the measurements - that's the first thing I did too. Yesterday night I re-did the measurements, now putting the loudspeakers in a slightly different position (15 cm back). The microphone was as close to the original measurement position as I could get it (the listening position) The comb-like behavior continued for Ethernet and wifi (with different through frequencies in each input, as in the first case) and did not appear (again) on Toslink.

So there is something going on.

I won't be able to test more until Monday night when I return to Oslo. I will try with another Mac (a MacBook Air, most likely running off batteries), and I may try as well with a Windows laptop.

[windler]

Very interesting results.
Up to 600 Hz seems that the behavior is identical on every connection.
This Make the business even much more interesting .
Deviations were expected to be lower when the band was supposed to be different.
A difference of 20 dB in the mid very meaningful wonder ...
Did you share this result with DEVIALET ?



One more question ... there is GAIN deference between the connection .
Did you compare the GAIN on each measurement ?

I never thought to do this comparison because I'm running RC anyway.

[windler]

I am using Air Ethernet and this is my room measurement

[krass]

I just looked again, the sawtooth seems to be a classic harmonic beat frequency of about 2khz (compare the frequency of adjacent troughsand peaks to each other), and the beat is apparent across the full spectrum. Whatever it is I would say its not "real" but definitely something that is being introduced in the test setup. I am intrigued and can't wait to find out what's causing it.....

EDIT: just realised I'm talking rubbish so I've greyed out the rubbish ! This is a frequency domain plot, harmonic beat is time-domain feature. ie I have no clue what's going on. Sorry for the noise folks !

[Confused]

Has anyone else tried REW with reasonably good results? If it was me, I would retry the tests with a different software package for the frequency analysis. Any recommendations for decent room frequency analysis software anyone?

[Mohmm]

Has anyone else tried REW with reasonably good results? If it was me, I would retry the tests with a different software package for the frequency analysis. Any recommendations for decent room frequency analysis software anyone?

I use FuzzMeasure+ a measuring microphone with good results on my Mac.

[Divasson]

Very interesting results.
Up to 600 Hz seems that the behavior is identical on every connection.
This Make the business even much more interesting .
Deviations were expected to be lower when the band was supposed to be different.
A difference of 20 dB in the mid very meaningful wonder ...
Did you share this result with DEVIALET ?



One more question ... there is GAIN deference between the connection .
Did you compare the GAIN on each measurement ?

I never thought to do this comparison because I'm running RC anyway.

Yes, I did share it with Devialet. They have asked me for my configuration file (which I submitted. On Monday night I will try to use another measurement program.

[pete86]

I am using Air Ethernet and this is my room measurement

If you wouldn't mind, could you repeat those measurements with USB and AIR WiFi? Perhaps Toslink too if that's possible?