[nick_charles]

components in the audible range ?

 

[ultrabike]

I think it's very possible.

 

[stv014]

The easiest way to find out is to actually try it with various headphones. In a quick test with random cheap headphones, I could only hear a 3 kHz IMD tone at higher than normal volume setting (and with a 0 dBFS ultrasonic signal that is unlikely to occur in music).

 

[nick_charles]

  The easiest way to find out is to actually try it with various headphones. In a quick test with random cheap headphones, I could only hear a 3 kHz IMD tone at higher than normal volume setting (and with a 0 dBFS ultrasonic signal that is unlikely to occur in music).

 
 
Which is pretty much what I got with Monty's test http://people.xiph.org/~xiphmont/demo/neil-young.html#toc_1ch but where did this spuriae come from?
 
Was it from the digital source (two different PCs with different sound chips IDT and Realtek),  DAC sections, the amp sections or the headphones.
 
All 3 of my "good" headphones (DT880, HD589, AD700) seem to produce the same effect to about the same degree and the effect is the same on two different DAC/headamp units with two different PCs as sources, maybe I'll swap in the M^3 amp and see if that makes a difference but I doubt it.
 
 
EDIT: Nope swapping in the M^3 made no difference. 

 

[esldude]

I got similar results to you.  Two headphones, one of which is also a DT880.  Using a 192 capable soundcard though nothing special.
 
Reducing the level of the two tones in Audacity by 3 db dropped the audible IM result lots more than 3 db.  I would estimate maybe 15 db or a bit more.  Another 3 db drop killed it even more than that.  At which point it was on the edge of even being heard with volume well up. 
 
An additional 3 db drop (now 9 db below the original signal which 3 db from max anyway) made it disappear. So that would indicate likely overload distortion somewhere in the chain.  Just guessing would think it in the analog portion.  Have a better rig I will try it on later, let you know what happens.

 

[xnor]

Anything with nonlinear distortion will produce IMD.
 
Any nonlinear signal processing, the DAC, the amp, the headphones .. all will produce harmonic and intermodulation distortion. Since audio stuff is optimized for a ~20 kHz range you will usually see quite an increase in distortion above that range.
 
The problem with IMD is that a completely inaudible 30 and 33 kHz tone will produce a (theoretically audible) 3 kHz intermodulation product.

 

[esldude]

Oh it isn't theoretical.  You definitely hear it. 

 

[xnor]

Yeah of course you can hear a 3 kHz tone, but under real circumstances it may be masked by music if the level of the tone is low enough.

 

[bigshot]

Not likely, that's the sweet spot for human hearing. (Fletcher Munson). It's more likely that the added distortion will mask the treble in the 6kHz range.

 

[stv014]

Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif
 
Was it from the digital source (two different PCs with different sound chips IDT and Realtek),  DAC sections, the amp sections or the headphones.

 
That could be tested with some measurements on the DAC and amplifier outputs.
 

  Yeah of course you can hear a 3 kHz tone, but under real circumstances it may be masked by music if the level of the tone is low enough.

 
Indeed, a signal with very high ultrasonic content and silence in the audio band is rather unnatural.

 

[nick_charles]

  Quote:
 
That could be tested with some measurements on the DAC and amplifier outputs.
 
 
Indeed, a signal with very high ultrasonic content and silence in the audio band is rather unnatural.

 
 
Right, I took the analog out from the DAC and recorded it using the line-in on my PC set to 24/96 and using Auadacity.  I tested the levels using music and then recorded the 30K/33K combination
 

 
 a very different result. The audible component was a very low level hiss/whistle,  the spectrum shows quite a lot of spreading, whether this is from the DAc or ADC I do not know - exporting the results to Excel I get
 

 
So the 3k component is there but is much less audible after going straight from the DAC and into an ADC and even cranking it up it is now barely audible. Not sure quite how to interpret this ?

 

[xnor]

Anything with nonlinear behavior is going to produce IMD. If your DAC and ADC are relatively "clean" then you won't see much IMD.

 

[ultrabike]

Nick, it might be the MME driver. ASIO might be better, but unfortunately Audacity needs to have ASIO compiled in. Under Linux possibly Audacity with ALSA may have less problems.

 

[stv014]

the spectrum shows quite a lot of spreading, whether this is from the DAc or ADC I do not know

 
It is probably from the relatively short Hann window used for creating the plots. It can be fixed by changing the window type to Blackman-Harris or Gaussian (a=4.5), which have much lower sidelobe levels. The second graph shows a lot of low level peaks, though, those could be some kind of DAC or ADC problem.

 

[nick_charles]

   
The second graph shows a lot of low level peaks, though, those could be some kind of DAC or ADC problem.

 
I am using the onboard soundcard (IDT) on my HP desktop with the annoying "never get rid of it" Beats Audio. I cannot vouch for the quality of the ADC. The DAC is a 2009 Zero so nothing special 
 
http://amarokcz.wz.cz/ZEROE.htm
 
I'm hesitant to invest in another soundcard just to unequivocably find out how awful my DAC is, but i suppose I could stretch to something like the Asus Xonar DG