Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Frequency measurements with different inputs
#21
New measurement, from a Macbook Air working on batteries, with Wifi Air. Totally galvanically isolated from the Devialet.

   

Basically, same problem with a different computer. Checked too sampling rate, etc.
Oslo: sources: Mac Mini/Itunes/Audirvana,  glassfiber toslink/Mapleshade clearview USB to D200;
Double helix to Gallo Strada 2 with mapleshade stands, TR3 subwoofer.

Spain: Oppo 95 vacuumstate modded, Kimber select 1021, Linn Classik Movie Di to Gallo Solos through Goertz Python cable


Reply
#22
Have you also done measurements with the same set-up using ethernet and USB? And optical?

If you get the same results as with the previous computer source, there may be a problem with your D200. Sorry to state the obvious!

Matt

Sonos Connect (W4S) > DSpeaker Antimode 2.0 > Sanders Magtech > Martin Logan Montis
Sonos Connect (W4S) > Devialet 200 > Vivid V1.5
Silver Phantoms (just the two)
London
Reply
#23
(29-Sep-2014, 23:36)Divasson Wrote: New measurement, from a Macbook Air working on batteries, with Wifi Air. Totally galvanically isolated from the Devialet.



Basically, same problem with a different computer. Checked too sampling rate, etc.

i think it's a comb-filter:
http://www.sengpielaudio.com/calculator-combfilter.htm
according to his diagrams, you are seeing a comb filter effect with a first notch at about 1100hz, first peak at 2.2khz. The other notches are consistent.
I'm not an expert, but he talks about disruptive room reflections with delays between 2-15ms. With your first notch at 1100hz, it looks like you have a disruptive delay of about 0.5ms (which is outside the 2-15 but Im getting a strong feeling of "disruptive reflections" here....).

So I would ask:
do you have well-damped walls ?
is your microphone placement sensible, or is it possibly at a room node, or halfway between 2 reflective walls, or (etc etc).

Why don't you try putting the mic somewhere entirely different, and put some temporary damping around the place just to see.....

I can't prove it, but I'm now 99% convinced it's a test placement/ setup issue.......

(or I'm completely wrong). Smile
 _________________________________________________________________________
Aurender X100, Audiophilleo, Devialet 200, Verity Audio Parsifal Ovation Monitors
Leiden, the Netherlands
Reply
#24
(30-Sep-2014, 20:08)krass Wrote:
(29-Sep-2014, 23:36)Divasson Wrote: New measurement, from a Macbook Air working on batteries, with Wifi Air. Totally galvanically isolated from the Devialet.



Basically, same problem with a different computer. Checked too sampling rate, etc.

i think it's a comb-filter:
http://www.sengpielaudio.com/calculator-combfilter.htm
according to his diagrams, you are seeing a comb filter effect with a first notch at about 1100hz, first peak at 2.2khz. The other notches are consistent.
I'm not an expert, but he talks about disruptive room reflections with delays between 2-15ms. With your first notch at 1100hz, it looks like you have a disruptive delay of about 0.5ms (which is outside the 2-15 but Im getting a strong feeling of "disruptive reflections" here....).

So I would ask:
do you have well-damped walls ?
is your microphone placement sensible, or is it possibly at a room node, or halfway between 2 reflective walls, or (etc etc).

Why don't you try putting the mic somewhere entirely different, and put some temporary damping around the place just to see.....

I can't prove it, but I'm now 99% convinced it's a test placement/ setup issue.......
..... and credit to f1eng who suggested the same thing earlier

(or I'm completely wrong). Smile

(29-Sep-2014, 23:36)Divasson Wrote: New measurement, from a Macbook Air working on batteries, with Wifi Air. Totally galvanically isolated from the Devialet.



Basically, same problem with a different computer. Checked too sampling rate, etc.

i think it's a comb-filter:
http://www.sengpielaudio.com/calculator-combfilter.htm
according to his diagrams, you are seeing a comb filter effect with a first notch at about 1100hz, first peak at 2.2khz. The other notches are consistent.
I'm not an expert, but he talks about disruptive room reflections with delays between 2-15ms. With your first notch at 1100hz, it looks like you have a disruptive delay of about 0.5ms (which is outside the 2-15 but Im getting a strong feeling of "disruptive reflections" here....).

So I would ask:
do you have well-damped walls ?
is your microphone placement sensible, or is it possibly at a room node, or halfway between 2 reflective walls, or (etc etc).

Why don't you try putting the mic somewhere entirely different, and put some temporary damping around the place just to see.....

I can't prove it, but I'm now 99% convinced it's a test placement/ setup issue.......

(or I'm completely wrong). Smile

This is what I thought from the beginning; but it should happen too with the Toslink input (remember that all measures have been made without any change to loudspeakers, room or connections. So Toslink should give the same comb filter effect!

The tests made on Monday of this week were done with the loudspeakers moved around 15cm back each, to avoid these problems. Still the comb filter appears - except on Toslink. Same frequencies.
Oslo: sources: Mac Mini/Itunes/Audirvana,  glassfiber toslink/Mapleshade clearview USB to D200;
Double helix to Gallo Strada 2 with mapleshade stands, TR3 subwoofer.

Spain: Oppo 95 vacuumstate modded, Kimber select 1021, Linn Classik Movie Di to Gallo Solos through Goertz Python cable


Reply
#25
I think moving the speakers is only one possibly less important part, since youre not changing the room node/ echo characteristics.

Maybe the position of he mic in relation to 1) the room nodes 2) the speakers 3) the combined room & speaker reflections is the problem?

It should be easy enough to run a test
a) original placement
b) original placement with loads of wall damping, blankets hung from walls etc
that should tell you something

but to be honest, it looks like a hard reflection problem so I think you're stuck with first doing some room/furnishing treatment. (is it a boxy room/lots of reflective surfaces or is a room thats quite open-plan/ heavily damped ?)

the website I linked to suggests that this effect is linked to disruptive time delays etc. Since its a closed-loop system when performing REW measurements, I wonder if using different cable types is simply introducing different time delay at the speakers (due to different amp path), so REW sees a different time bias. We're only talking parts of a millisecond: This wouldn't affect sound-quality when listening normally, but might be important in a closed-loop test configuration where the software is both generating signals and simultaneously checking the response (the linked website shows there is a relationship between delay, distance, phase etc).
 _________________________________________________________________________
Aurender X100, Audiophilleo, Devialet 200, Verity Audio Parsifal Ovation Monitors
Leiden, the Netherlands
Reply
#26
(30-Sep-2014, 20:36)Divasson Wrote: This is what I thought from the beginning; but it should happen too with the Toslink input (remember that all measures have been made without any change to loudspeakers, room or connections. So Toslink should give the same comb filter effect!

The tests made on Monday of this week were done with the loudspeakers moved around 15cm back each, to avoid these problems. Still the comb filter appears - except on Toslink. Same frequencies.

This is the most important part of the discussion. I'd have thought all comments about room effects are idle until we can explain the difference between Toslink and the other inputs.
As I said earlier, it may be a faulty Devialet.
Matt

Sonos Connect (W4S) > DSpeaker Antimode 2.0 > Sanders Magtech > Martin Logan Montis
Sonos Connect (W4S) > Devialet 200 > Vivid V1.5
Silver Phantoms (just the two)
London
Reply
#27
It could indeed be some kind of hard reflection issue, but it does look a little too drastic. If it was me I would like to double check the validity of the results, for example, try alternative analyser software, maybe even try a different microphone. It is just possible that the mic or software is the culprit, maybe not, but it is at least possible. Also, if I could beg steal or borrow another dac / amp combo, I'd repeat the test without the Devialet. (Maybe you have a surround amp, that could provide a substitute in stereo mode?). Loads of hassle I know, but if you could try any of the above, then you would genuinely start to have some real evidence for what is occurring.
Reply
#28
(30-Sep-2014, 21:27)krass Wrote: I think moving the speakers is only one possibly less important part, since youre not changing the room node/ echo characteristics.

Maybe the position of he mic in relation to 1) the room nodes 2) the speakers 3) the combined room & speaker reflections is the problem?

It should be easy enough to run a test
a) original placement
b) original placement with loads of wall damping, blankets hung from walls etc
that should tell you something

but to be honest, it looks like a hard reflection problem so I think you're stuck with first doing some room/furnishing treatment. (is it a boxy room/lots of reflective surfaces or is a room thats quite open-plan/ heavily damped ?)

the website I linked to suggests that this effect is linked to disruptive time delays etc. Since its a closed-loop system when performing REW measurements, I wonder if using different cable types is simply introducing different time delay at the speakers (due to different amp path), so REW sees a different time bias. We're only talking parts of a millisecond: This wouldn't affect sound-quality when listening normally, but might be important in a closed-loop test configuration where the software is both generating signals and simultaneously checking the response (the linked website shows there is a relationship between delay, distance, phase etc).

I have done another test, using a very sophisticated spectrum analyzer: my ears. I have run several times diverse frequency sweeps, very slow, from 800 to 2000 hz, covering the biggest through. When they say that 1dB is the smallest frequency variation a trained ear can hear, and that 10dB is like doubling the apparent loudness at middle frequencies*, I guess that i should be able to detect a 30dB variation!

Well, I can't hear any significant variation. So, it is not real. Then, it must be the test setup. What happens that does not happen with Toslink?

After thinking for a while: it may be related to the latency of each method, that may interfere with the readings. Toslink has the lowest lag - it is just a direct digital connection. Air, even with ethernet, has at least the minimum latency fixed in the settings screen: 50 ms. USB should have a short latency too - but it may be enough to trump the REW measuring logic.

And USB sounds so much better than Toslink with Audirvana 2.0.2. So, let's relax and enjoy.


*while doubling power is 3dB, and doubling voltage is 6dB, the apparent loudness seems to double with each 10dB
Oslo: sources: Mac Mini/Itunes/Audirvana,  glassfiber toslink/Mapleshade clearview USB to D200;
Double helix to Gallo Strada 2 with mapleshade stands, TR3 subwoofer.

Spain: Oppo 95 vacuumstate modded, Kimber select 1021, Linn Classik Movie Di to Gallo Solos through Goertz Python cable


Reply
#29
Joining the discussion a bit late…but this looks really weird!

Do you have any chance to try other inputs to the Devialet: S/PDIF, AES/EBU or analog line in? Off the top of my head I'd expect S/PDIF and AES/EBU to behave the same as Toslink.

As Matt suggested, just maybe there's a problem with your D200. More data about how different inputs behave might help to track it down.

Did Devialet have any comments or suggestions other than "send the configuration file"?

Ian
Roon (Mac Mini), Wilson Benesch Full Circle, Expert 1000 Pro CI, Kaiser Chiara
Warwickshire, UK
Reply
#30
(30-Sep-2014, 22:00)Divasson Wrote: I have done another test, using a very sophisticated spectrum analyzer: my ears.

I use the same. (Although my ears, not @Divasson's) However inaccurate they may be, they are after all the only thing that really matters.

Having said that, I would refer back to my earlier post and ask, does anyone know of a decent analyser software / microphone combination that they have actually used successfully with the Devialet / AIR?

I am thinking of playing with speaker positioning / room treatment to deal with some minor issues my own ears have detected, is there anything out there that might be useful? Or should I just stick with my trusty lug holes?
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)