or Connect
Head-Fi.org › Forums › Equipment Forums › Sound Science › With headphones can two supersonic tones (say 30k and 33K) at high levels create IMD with products in the audible range ?
New Posts  All Forums:Forum Nav:

With headphones can two supersonic tones (say 30k and 33K) at high levels create IMD with products in the audible range ? - Page 2

post #16 of 31

LOL! I have the same problem: HP with Beats Audio! I tried using Ubuntu on it but it overheats after some time. One thing that could be done is to install Ubuntu to a flash drive and run from it briefly for testing these sorts of things. I think that should bypass the Beats Audio problem.

 

Also, you might want to update the drivers. I did, and think that improved things a bit.

post #17 of 31

Since the Beats stuff seems to be driver-specific can't you just completely uninstall all that crap and install vanilla drivers?

Onboard is usually Realtek: realtek.com.tw

post #18 of 31

Well. I actually nuked my Losedoze7 install, bought a copy and re-installed it. That was because I got fairly frustrated with Beats and other issues and went Ubuntu only to find out that the laptop was overheating.

 

In the end, after re-installing Effdoze7, I also reinstalled the HP drivers... which have Beats in it. Did Not. Want. Overheating Issues. Again. However, in the process I think I updated all the drivers and things don't sound nearly as awful directly out of the internal soundcard.

 

That said, I use the Focusrite 2i2 for acquisition (which bypasses Beats crapness), and still get some weirdness if using Audacity + WASAPI/MME and so forth (stv014 might remember my Audacity interesting captures). For measurements I use ASIO and have no problems. But Audacity for Shamedoze 7 does not support ASIO since it seems is proprietary or so... unless you compile it to it.


Edited by ultrabike - 12/9/13 at 5:35pm
post #19 of 31

I got around to using a better rig on this.  My DT880 fed from a TC Impact Twin.  Used Foobar 2k with WASAPI out to an Audiophilleo 2 USB/SPDIF converter.  Listening over headphones connected to the Impact Twin I heard exactly nothing.  Not one thing.   Did a loopback in the Impact twin.  Got the ultrasonic tones cleanly at about -6.7 db each which would have put the combination short of clipping barely.  The IMD at 3000 hz was running around -100db to -101 db.  No wonder I didn't hear it!! 

 

But using a built in sound card on a Gigabyte MB it was quite obvious.  It is a 192khz card, but apparently the analog portion of it spits out some crazy high levels of IMD above 20 khz.  It was plainly audible at a pretty high level using the same headphones.  Playing the same signal over desktop MAudio BX5a speakers it was also audible at a high level, but these were being fed from the same card. In the Impact Twin nothing. 

 

So in my case at least it was simply analog IMD. 

post #20 of 31

In all cases I used ASIO drivers through foobar2k.

 

Focusrite 2i2

 

Through my 2i2 > HD558s I could ONLY hear the "26kHz - 48kHz warbling tones" distortion (2i2 does not support 192kHz).

 

In this setup I could not hear "30kHz + 33kHz tone" distortion.

 

HP Laptop

 

Through my laptop > HD558, I could hear distortion in ALL files (except the ultrasonic Juanes - "A Dios Le Pido")...


Edited by ultrabike - 12/10/13 at 12:32am
post #21 of 31
I ran the test using Audio-GD FUN with USB32 interface and got IMD with all the tests, but I had to crank the volume really up. The IMD with the ultrasonic Juanes (my fellow compatriot) is almost quiet though.
post #22 of 31
Thread Starter 
Quote:
Originally Posted by aroldan View Post
I ran the test using Audio-GD FUN with USB32 interface and got IMD with all the tests, but I had to crank the volume really up. The IMD with the ultrasonic Juanes (my fellow compatriot) is almost quiet though.

 

Still a fail though :wink:

 

This leaves me with a quandary. Do I get a better DAC such as the $400 DACMagic with measurably superb IMD and a better Soundcard and bypass the onboard chip (Beats is bundled with the IDT Chip in my HP and if you update the IDT driver it (Beats) comes back again, plus on my system if you uninstall Beats parts of the Windows sound mixer just vanish - generally the recording parts which Audacity defers to) to make sure the optical out is clean (don't know if you can get IMD on a purely digital signal , does not seem likely ?)  or do I just run my DAC in 16/44.1 mode since I have no high res music files whatsoever.

 

Having something you know is flawed even if you cannot hear it normally picks at my British Anal-retentive nature

 

 

EDIT tried with a 3rd DAC - SMSL SD 1955 - exactly the same result 


Edited by nick_charles - 12/10/13 at 3:34pm
post #23 of 31

My best guess is that the IMD from a DAC is more a function of the conversion than the bit stream feeding it. The bitstream might get noisy and jittery though, which might can affect performance if the DAC does a poor job at dealing with these issues.

 

Anyhow, running 16/44.1 might be K, but I would go for the external half-decent but not uber-pricy soundcard route. It's a crappy laptop/desktop friendly solution.

post #24 of 31
Thread Starter 
Quote:
Originally Posted by ultrabike View Post
 

My best guess is that the IMD from a DAC is more a function of the conversion than the bit stream feeding it. The bitstream might get noisy and jittery though, which might can affect performance if the DAC does a poor job at dealing with these issues.

 

Anyhow, running 16/44.1 might be K, but I would go for the external half-decent but not uber-pricy soundcard route. It's a crappy laptop/desktop friendly solution.

 

 

Hmm, three different DACs all fed from the same optical feed giving the exact same problem to the exact same extent (afaict) , swapping in an external amp for the Zero and Fire Phoenix makes no difference, the M^3 gives the same result. I thought that the Fire Phoenix was the least distorted but that was just a function of it's less powerful internal amp, running it from its line out to the M^3 and the high volume whistle is the same as on the Zero.

 

Even though my 3 DACs are very low end they are also quite different in terms of the chips and topologies, I'm just a bit suspicious that 3 very different DACs behave exactly the same to the same degree...

post #25 of 31
Dunno. Maybe the analog filters on those delta sigma DACS are not doing anything to ultrasonic stuff between 20 and 48 or more kHz, and the amp section acts up if those signals are sufficiently high...
Edited by ultrabike - 12/10/13 at 6:50pm
post #26 of 31
Thread Starter 
Quote:
Originally Posted by ultrabike View Post

Dunno. Maybe the analog filters on those delta sigma DACS are not doing anything to ultrasonic stuff between 20 and 48 or more kHz, and the amp section acts up if those signals are sufficiently high...

 

But for some tests I used the Line out on the DACs that have a built-in amp and sent the signal to a separate amp with the same result, the external amp is an M^3 , one of the DACs has no built-in amp so always used the M^3, of course the M^3 could be equally flawed

 

It is possible all three DACs are equally crappy but I hesitate to get another (more expensive) one and find the same problem

post #27 of 31

Your results are a bit strange.  A few posts back you showed a DAC to ADC trace with the 3 khz tone at -92 db.  Seems like that would be difficult to hear.  Have you done any more measuring of the output?

 

Another idea, I looked at the individual bit levels.  The highest individual bit levels are -3.07 db.  That should prevent any odd overshoot behavior from a digital signal feeding a DAC.  But maybe try putting the file in a sound editor, knock it down 3 db more and see if the whistles disappears.  Still could be analog or digital in that case, but if it disappears with all three DAC's I would suspect an odd digital issue.  And I would wonder if your digital output is being monkeyed with by your playback software.  If you have a way to record the SPDIF output digitally on another piece of gear, that might be instructive too.  See if the peak bit values are getting altered.


Edited by esldude - 12/10/13 at 8:44pm
post #28 of 31

an option that can help is to use the 2 channels of a DAC, one for each test tone, add them together with a pair of resistors: ~5 kOhm should be high enough to not load most DAC outputs and low enough not to add too much noise, standard 1/4 W metal film R should be adequate quality - this should give a vanishingly low IMD test signal even from poor DACs

 

another resolution increase could be had by few kHz RC low pass filter, after the DUT output, before going to the ADC, to knock down the test tones ~10x vs the IMD product you're looking for, reducing the chance the ADC circuitry IMD masks the DUT IMD - do use a quality C - polypropylene or polystyrene dielectric

 

all of this can be passive, no op amps or power required although soldering and a metal box for shielding would be nice


Edited by jcx - 12/10/13 at 9:03pm
post #29 of 31
Quote:
Originally Posted by nick_charles View Post
 

 

But for some tests I used the Line out on the DACs that have a built-in amp and sent the signal to a separate amp with the same result, the external amp is an M^3 , one of the DACs has no built-in amp so always used the M^3, of course the M^3 could be equally flawed

 

It is possible all three DACs are equally crappy but I hesitate to get another (more expensive) one and find the same problem

 

 

LOL! Maybe the amps are not that flawed. AFAIK audio amps should work well in the audio range (< 20 kHz). Asking an audio amp to be uber-ultra-mega-linear while playing stuff loud in the ultrasonic range for the house pets might be unnecessarily a bit too much.

 

Looking at the datasheet for the AD1852 (what I think the Zero DAC uses) it seems it uses a 3rd order Bessel with corner frequency at 75 kHz:

 

http://www.analog.com/static/imported-files/data_sheets/AD1852.pdf.

 

I dunno, but IMO 75 kHz seems a bit too much.

 

Furthermore, judging from the plots in the link below (not sure about their accuracy), it seems some delta sigma DACs may put out quite a bit of stuff in the ultrasonic range which may also indicate to high a corner frequency in the output low pass filter, or it's absence, perhaps requiring further filtering before signal hits the amp:

 

http://www.audiocircle.com/index.php?action=gallery;area=browse;image=17983

 

Lowering the BW of the analog low pass filter (or adding one) prior to the amp might improve things out. Or perhaps the Amps could add an input analog low pass filter.


Edited by ultrabike - 12/10/13 at 9:19pm
post #30 of 31
Thread Starter 
Quote:
Originally Posted by esldude View Post
 

Your results are a bit strange.  A few posts back you showed a DAC to ADC trace with the 3 khz tone at -92 db.  Seems like that would be difficult to hear.  Have you done any more measuring of the output?

 

Another idea, I looked at the individual bit levels.  The highest individual bit levels are -3.07 db.  That should prevent any odd overshoot behavior from a digital signal feeding a DAC.  But maybe try putting the file in a sound editor, knock it down 3 db more and see if the whistles disappears.  Still could be analog or digital in that case, but if it disappears with all three DAC's I would suspect an odd digital issue.  And I would wonder if your digital output is being monkeyed with by your playback software.  If you have a way to record the SPDIF output digitally on another piece of gear, that might be instructive too.  See if the peak bit values are getting altered.

 

 

The recording was at too low a level, I was having trouble with setting the levels, when I reinstalled the sound driver the distortion on a 2nd recording was nearer -70db and there was still some way to go to get the levels up high enough

 

I am using FooBar with WASAPI for playback but I could try a different player.

 

EDIT : Audacity showed the same problem, though I have devices where the digital out is "louder" than the input (WD HDTV) 

 

EDIT: I just tried the IDT card directly thru the headphone out, at first I thought it was perfectly clean then I put it through the M^3 amp and cranked up the 3K whistling was there


Edited by nick_charles - 12/11/13 at 9:03am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › With headphones can two supersonic tones (say 30k and 33K) at high levels create IMD with products in the audible range ?